a Unweighted statistics of some key variables describing the study population in the youth pre-DM/DM data set overall and by pre-DM/DM status. More detailed statistics for all the variables in our data set can be found in the Data Exploration section of POND.
b Pre-DM/DM: pre–diabetes mellitus and diabetes mellitus.
c CHIP: child health insurance program.
d Hypertensive was defined by blood pressure ≥90th percentile or ≥120/80 mm Hg for children 13 years of age and older [ 2 ].
e eq: equivalent.
We estimated that the survey-weighted prevalence of pre-DM/DM in our study population rose substantially from 4.1% (95% CI 2.8-5.4) in 1999 to 22% (95% CI 18.5-25.6) in 2018 (Figure S3 and section S6 in Multimedia Appendix 1 ). This increasing trend of pre-DM/DM prevalence was consistent with that reported in other NHANES-based studies, which had pre-DM/DM prevalence ranging from 17.7% to 18% [ 18 , 19 ]. We also applied the study population and pre-DM definition criteria reported in a recent study [ 13 ] to NHANES data and derived a similarly sized study population (n=6656 vs n=6598 in the current vs previous analysis [ 13 ]) and youth pre-DM prevalence, which ranged from 11.1% (95% CI 8.9-13.3) to 37.3% (95% CI 31.0-43.6) in our analysis compared with from 11.6% (95% CI 9.5-14.1) to 28.2% (95% CI 23.3-33.6) in the study by Liu et al [ 13 ] (Table S6 in Multimedia Appendix 1 ).
We extracted 95 epidemiological variables from NHANES and organized them into 4 pre-DM/DM-related domains, namely, sociodemographic, health status, diet, and other lifestyle behaviors (Table S1 in Multimedia Appendix 1 ). Table 1 shows the unweighted statistics of some key study population characteristics. Among youth with pre-DM/DM (n=2010), the proportion of youth who were non-Hispanic Black, non-Hispanic White, Hispanic, and other race or ethnicity (including non-Hispanic persons who reported races other than Black or White and non-Hispanic Asian) were 33.6% (n=676), 21.4% (n=431), 35.4% (n=711), and 9.6% (n=192), respectively. Approximately, half (7719/15,149, 51%) of the population were male, and they represented 65.6% (1319/2010) of those with pre-DM/DM. Approximately 32.4% (4528/15,149) of the youth had a family income below poverty level, and 69.4% (7833/15,149) were from households receiving food stamps. The proportion of youth covered by private insurance was higher among those with than with no pre-DM/DM (5648/13,139, 43.8% vs 744/2010, 37.7%). Overall, 21.5% (3214/15,149) of the youth were obese as defined by having a BMI at or above the 95th percentile based on age and gender, and the proportion was 33.3% (663/2010) among youth with pre-DM/DM. Youth with pre-DM/DM tended to have less fruit and vegetable intake and ate lower amounts of protein and total grains than those with no pre-DM/DM. Youth with and with no pre-DM/DM showed similar amounts of physical activity with 209 and 210 minutes per week, respectively ( Table 1 ).
To facilitate other researchers’ use of our youth pre-DM/DM data set and make our methodology transparent and reproducible, we developed POND, which is available on [ 47 ]. Users can navigate POND through its built-in functionalities. For example, users are able to explore the details of the 95 individual variables ( Figure 3 A) and their distributions by pre-DM/DM status ( Figure 3 B), examine the risk factors of youth pre-DM/DM identified from the case studies described below ( Figure 3 C), as well as download the data for customized analysis and the analytical code to replicate our findings ( Figure 3 D). In addition, we make available all the code used to develop the data set, our case studies, and POND itself.
We examined the validity and use of our processed multidomain data set for translational studies on youth pre-DM/DM by the following 2 complementary types of data analyses.
In our bivariate analyses, we found 27 variables to be significantly ( P <.001, Bonferroni adjusted) associated with pre-DM/DM status ( Figure 4 [ 63 ] and Table S7 in Multimedia Appendix 1 ). These variables spanned all 4 domains and included gender, race or ethnicity, use of food stamps, health insurance status, BMI, total protein intake, and screen time. Similar results were found when repeating these bivariate association tests after accounting for NHANES survey design elements (Table S7 in Multimedia Appendix 1 ).
We used an ML framework, EI [ 53 , 54 ], to leverage the multidomain nature of our data set and predict youth pre-DM/DM status. We also compared EI’s performance with alternative prediction approaches, most prominently the widely used XGBoost algorithm [ 71 ].
The best-performing multidomain EI methodology, stacking [ 75 ] using logistic regression, predicted youth pre-DM/DM status (AUROC=0.67; BA=0.62) more accurately than all the alternative approaches ( Figure 5 ), namely, XGBoost (AUROC=0.64; BA=0.60; Wilcoxon rank sum FDR=1.7×10 4 and 1.8×10 4 , respectively), the ADA pediatric screening guidelines (AUROC=0.57, BA=0.57; Wilcoxon rank sum FDR=1.7×10 4 and 1.8×10 4 , respectively), and 4 single-domain EI (AUROC=0.63-0.54; BA=0.60-0.53; FDR <1.7×10 4 and 1.8×10 4 , respectively).
The multidomain EI also identified 27 variables (the same as the number of significant variables from bivariate analyses) that contributed the most to predicting youth pre-DM/DM status. Among these variables, 16 overlapped with those identified from the bivariate statistical analyses ( Figure 6 ; Fisher P of overlap=7.06×10 6 ). These variables identified by both approaches included some established pre-DM/DM risk factors such as BMI and high total cholesterol, as well as some less-recognized ones such as screen time and taking prescription drugs [ 2 ].
Leveraging the rich information in NHANES spanning nearly 20 years, we built the most comprehensive epidemiological data set for studying youth pre-DM/DM. We accomplished this by selecting and harmonizing variables relevant to youth pre-DM/DM from sociodemographic, health status, diet, and other lifestyle behaviors domains. This youth pre-DM/DM data set, as well as several functionalities to explore and analyze it, is publicly available in our user-friendly web portal, POND. We also conducted case studies using the data set with both traditional statistical methods and ML approaches to demonstrate the potential of using this data set to identify factors relevant to youth pre-DM/DM. The combination of the comprehensive public data set and POND provides avenues for more informed investigations of youth pre-DM/DM.
The future translational impact of pre-DM/DM research, facilitated by comprehensive data sets such as the one developed in this study, holds significant promise for advancing our understanding of the disease and its risk factors among youth. By enabling researchers to investigate multifactorial variables associated with pre-DM/DM, this data set contributes to several areas of research and has a broader impact on the scientific community. First, the data set’s comprehensive nature allows researchers to explore the collective impact of various risk factors across multiple health domains. By incorporating sociodemographic factors, health status indicators, diet, and lifestyle behaviors, researchers can gain a holistic understanding of the interplay between these factors and pre-DM/DM risk among youth. This knowledge can be used to generate hypotheses for further studies and inform the development of targeted interventions and prevention strategies that address the specific needs of at-risk populations. Furthermore, the data set provides an opportunity to delve into less-studied variables and their interactions in relation to pre-DM/DM risk. Variables such as screen time, acculturation, or frequency of eating out, which are often overlooked in traditional research, can be examined to uncover their potential influence on pre-DM/DM risk among youth. This expands the scope of translational research and enhances our understanding of the multifaceted nature of the disease.
One of the major contributions of our work was POND, our publicly available web portal, which provided access to all materials related to our data set and analyses, thus enabling transparency and reproducibility. Although several such portals are available in other biomedical areas, such as genomics [ 76 - 78 ], there is a general lack of such tools in epidemiology and public health. We hope that, in addition to facilitating studies into pre-DM/DM, POND illustrates the use of such portals for population and epidemiological studies as well.
The results of the case studies and validation exercises we conducted were also consistent with existing literature. The case studies identified known pre-DM/DM risk factors, such as gender [ 15 , 17 , 19 ], race and ethnicity [ 2 , 9 , 10 , 24 ], health measures (BMI, hypertension, and cholesterol) [ 2 , 55 ], income [ 9 , 11 ], insurance status [ 9 , 10 ], and health care availability [ 9 , 10 ], thus affirming the validity of the data set. In addition, our analyses revealed some less studied variables, such as screen time, home ownership status, self-reported health status, soy and nut consumption, and frequency of school meal intake, which may influence youth pre-DM/DM risk. Further study of these variables may reveal new knowledge about pre-DM/DM among youth. More generally, such novel findings further demonstrate the use of our data set and data-driven methods for further translational discoveries about this complex disorder.
Although our work has several strengths and high potential use for youth pre-DM/DM studies, it is not without limitations. First, as our data set was derived from NHANES, we adopt limitations to the survey in our data set. Since NHANES is a cross-sectional survey, the pre-DM/DM status and its related variables provide only consecutive snapshots of youth in the United States over time across the available survey cycles. Thus, the associations identified are better suited for hypothesis generation purposes and require in-depth investigation using prospective longitudinal and randomized trial designs. In addition, we modified the ADA guideline for determining pre-DM/DM status according to variable availability. Due to the high missingness of 45% in family history (DIQ170) and the complete missingness of maternal history (DIQ175S) from 1999 to 2010 in the raw NHANES data, we were unable to include family history of diabetes in the data set. Similarly, NHANES does not provide data regarding every condition associated with insulin resistance. Therefore, we used hypertension and high cholesterol as proxies for insulin resistance. On the other hand, as our main purpose is to use POND as a conduit between this comprehensive youth pre-DM/DM database and interested researchers, our method can be adopted to longitudinal data sets should they become available in the future. Second, for the prediction of pre-DM/DM status, EI’s performance was found to be significantly better than the alternative approaches, including a modified form of the suggested guideline [ 45 ]. However, this performance assessment was based only on cross-validation, which is no substitute for validation on external data sets that is necessary for rigorous assessment. Finally, while our preliminary case study analyses identified a wide range of variables associated with youth prediabetes and diabetes, other known risk factors, such as current asthma status [ 80 - 82 ], added sugar consumption [ 83 - 85 ], sugary fruit and juice intake [ 83 - 86 ], and physical activity per week [ 6 - 8 , 50 ], were not identified. This limitation can be addressed by using other data analysis methods beyond our bivariate testing and ML approaches, highlighting more potential use cases of our data set.
Overall, the future impact of translational pre-DM/DM research facilitated by comprehensive data sets and web servers like ours extends beyond individual studies. It creates opportunities for interdisciplinary collaboration and reproducibility, strengthens evidence-based decision-making, and supports the development of targeted interventions for the prevention and management of pre-DM/DM among youth. By providing rich resources, our work can enable researchers to build upon existing knowledge and push the boundaries of translational pre-DM/DM research, ultimately leading to improved health outcomes for at-risk populations.
This study was enabled in part by computational resources provided by Scientific Computing and Data at the Icahn School of Medicine at Mount Sinai. The Ensemble Integration used in this work was implemented by Jamie JR Bennett. This work was funded by National Institutes of Health grants R21DK131555 and R01HG011407.
The data set and code used in this study are available at Zenodo [ 87 ] and our web portal POND [ 47 ].
BL and GP contributed equally as cosenior and cosupervisory authors. NV, BL, and GP conceptualized the project. CM, YCL, NV, BL, and GP designed the methodology. CM and BL implemented the data curation and bivariate analyses. YCL implemented the ML case study and POND. CM and YCL conducted formal analysis and visualization. CM, YCL, NV, BL, and GP wrote the manuscript. NV, BL, and GP supervised the project.
None declared.
Supplemental materials.
American Diabetes Association |
area under the receiver operating characteristic curve |
balanced accuracy |
diabetes mellitus |
Ensemble Integration |
false discovery rate |
fasting plasma glucose |
glycated hemoglobin |
machine learning |
National Health and Nutrition Examination Survey |
Prediabetes/diabetes in youth Online Dashboard |
pre–diabetes |
social determinants of health |
extreme gradient boosting |
Edited by A Mavragani, T Sanchez; submitted 05.10.23; peer-reviewed by S El Khamlichi, C Zhao, Y Su; comments to author 09.01.24; revised version received 06.02.24; accepted 26.04.24; published 02.07.24.
©Catherine McDonough, Yan Chak Li, Nita Vangeepuram, Bian Liu, Gaurav Pandey. Originally published in JMIR Public Health and Surveillance (https://publichealth.jmir.org), 02.07.2024.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Public Health and Surveillance, is properly cited. The complete bibliographic information, a link to the original publication on https://publichealth.jmir.org, as well as this copyright and license information must be included.
Published on 3.7.2024 in Vol 26 (2024)
Authors of this article:
1 Division of Cardiology, Department of Internal Medicine, Seoul National University Bundang Hospital, Seoul National University College of Medicine, Seongnam, Gyeonggi-do, Republic of Korea
2 ARPI Inc, Seongnam, Gyeonggi-do, Republic of Korea
3 Department of Emergency Medicine, Seoul National University Bundang Hospital, Seongnam, Gyeonggi-do, Republic of Korea
4 Division of Cardiology, Department of Internal Medicine, Severance Hospital, Yonsei University College of Medicine, Seoul, Republic of Korea
*these authors contributed equally
Dong-Ju Choi, MD, PhD
Division of Cardiology, Department of Internal Medicine
Seoul National University Bundang Hospital
Seoul National University College of Medicine
82 Gumi-ro 173 Beon-gil
Seongnam, Gyeonggi-do, 13620
Republic of Korea
Phone: 82 317877007
Fax:82 317877041
Email: [email protected]
Background: Although several biomarkers exist for patients with heart failure (HF), their use in routine clinical practice is often constrained by high costs and limited availability.
Objective: We examined the utility of an artificial intelligence (AI) algorithm that analyzes printed electrocardiograms (ECGs) for outcome prediction in patients with acute HF.
Methods: We retrospectively analyzed prospectively collected data of patients with acute HF at two tertiary centers in Korea. Baseline ECGs were analyzed using a deep-learning system called Quantitative ECG (QCG), which was trained to detect several urgent clinical conditions, including shock, cardiac arrest, and reduced left ventricular ejection fraction (LVEF).
Results: Among the 1254 patients enrolled, in-hospital cardiac death occurred in 53 (4.2%) patients, and the QCG score for critical events (QCG-Critical) was significantly higher in these patients than in survivors (mean 0.57, SD 0.23 vs mean 0.29, SD 0.20; P <.001). The QCG-Critical score was an independent predictor of in-hospital cardiac death after adjustment for age, sex, comorbidities, HF etiology/type, atrial fibrillation, and QRS widening (adjusted odds ratio [OR] 1.68, 95% CI 1.47-1.92 per 0.1 increase; P <.001), and remained a significant predictor after additional adjustments for echocardiographic LVEF and N-terminal prohormone of brain natriuretic peptide level (adjusted OR 1.59, 95% CI 1.36-1.87 per 0.1 increase; P <.001). During long-term follow-up, patients with higher QCG-Critical scores (>0.5) had higher mortality rates than those with low QCG-Critical scores (<0.25) (adjusted hazard ratio 2.69, 95% CI 2.14-3.38; P <.001).
Conclusions: Predicting outcomes in patients with acute HF using the QCG-Critical score is feasible, indicating that this AI-based ECG score may be a novel biomarker for these patients.
Trial Registration: ClinicalTrials.gov NCT01389843; https://clinicaltrials.gov/study/NCT01389843
Heart failure (HF) is a major global health problem affecting millions of people worldwide, leading to significant morbidity, mortality, and health care expenditure [ 1 - 3 ]. Although several valuable biomarkers such as N-terminal prohormone of brain natriuretic peptide (NT-proBNP) [ 4 , 5 ] and cardiac troponins [ 6 ] have been introduced for patients with HF, their use in routine clinical practice is often constrained by their cost and limited availability.
Electrocardiogram (ECG) is an essential and cost-effective tool for evaluating cardiovascular diseases. ECG is widely available, noninvasive, and provides real-time information about cardiac electrical activity, which is crucial for detecting arrhythmias, ischemia, and other cardiac abnormalities. With advances in artificial intelligence (AI) and deep learning, there has been growing interest in employing AI algorithms to analyze ECG data and predict outcomes in patients with various cardiovascular conditions [ 7 , 8 ].
In this study, we investigated the utility of an AI algorithm that analyzes printed ECG images for outcome prediction in patients with acute HF. These findings will demonstrate the potential of AI-assisted ECG analysis for predicting outcomes in these patients, potentially overcoming the cost and availability constraints of current biomarkers.
This was a substudy of the prospective multicenter Korean Acute Heart Failure (KorAHF) registry, which enrolled 5625 consecutive patients upon initial hospital admission for acute HF at 10 tertiary university hospitals in Korea. Details on the KorAHF registry objectives, design, and population are available on the clinical trial registration site (ClinicalTrials.gov NCT01389843) and have been published previously [ 9 , 10 ]. Briefly, patients who had signs or symptoms of HF and met one of the following criteria were eligible for enrollment in the KorAHF registry: (1) lung congestion or (2) objective left ventricular systolic dysfunction or structural heart disease findings. There were no exclusion criteria.
In this study, we retrospectively analyzed the prospectively collected data from 1254 patients who were hospitalized for acute HF from March 2011 to February 2014 at 2 out of 10 participating tertiary centers (Seoul National University Bundang Hospital and Severance Hospital) using the KorAHF registry ( Figure 1 ). Additional ECG image data were collected for this study.
This study conformed with the principles outlined in the Declaration of Helsinki. The study protocol was approved by the institutional review board at Seoul National University Bundang Hospital (No. B-1104-125-014) and Severance Hospital (No. 2022-2166-001). The need for written informed consent was waived by the institutional review board. Our research strictly adheres to the Guidelines for Developing and Reporting Machine Learning Predictive Models in Biomedical Research [ 11 ].
Data collection methods have been previously described [ 9 ]. Briefly, data on patients’ clinical manifestations, biochemical parameters, medication, and outcome were collected using a web-based case report form for up to 60 months by research nurses. Outcome data on patients lost to follow-up were additionally collected from national death records.
The primary endpoint of this study was all-cause mortality. Secondary outcomes included in-hospital outcomes, especially in-hospital mortality. All deaths were considered to be cardiac-related unless a definite noncardiac cause could be established. All outcome data reported from the participating centers were reviewed by an independent clinical event adjudicating committee.
Quantitative ECG (QCG) is an AI analyzer composed of an encoder part and multiple task-specific networks. The encoder part is a modified convolutional neural network with residual connections, squeeze excitation modules, and a nonlocal block. The task-specific networks are multilayer percetron models. The encoder part accepts 2D ECG images as input to produce a common numerical feature vector for downstream tasks. The encoder part was pretrained on 49,731 open ECGs using self-supervised learning schemes and then fine-tuned on 47,194 annotated ECG images of over 32,968 patients who visited the Emergency Department of Seoul National University Bundang Hospital between 2017 and 2019 using multitask learning schemes. The tasks include the classification of 12 rhythms (with 35 subtypes) and production of 10 digital biomarkers correlated with the risk of (1) being critically ill (shock, respiratory failure, or cardiac arrest), (2) cardiac ischemia (acute coronary syndrome, ST-elevation myocardial infarction, or myocardial injury as defined by an elevated troponin level), (3) cardiac dysfunction (pulmonary edema, left and right heart dysfunction, pulmonary hypertension, and clinically significant pericardial effusion), and (4) hyperkalemia. Several validation studies of the system have been published previously [ 12 - 14 ]. The collection of these AI algorithms has been developed into a mobile app (ECG Buddy, ARPI), which has been approved by the Korean Ministry of Food and Drug Safety.
In this study, two QCG features were evaluated: QCG-Critical for critical conditions such as shock or mortality and QCG-HF for a reduced echocardiographic left ventricular ejection fraction (LVEF) of <40%. The QCG scores, representing probability, ranged from 0 to 1.0, with 0 indicating low and 1.0 indicating high probability. With a 9:1 ratio split of the training and test data sets, the internal validation results for these two QCG features showed an area under the curve (AUC) of 0.877 for QCG-Critical and 0.956 for QCG-HF. The composition of the training and validation data sets is presented as a flowchart in Figure 1 .
Categorical variables are reported as frequencies (percentages) and continuous variables are expressed as means (SD) or medians (IQR). The two key AI-driven scores (QCG-Critical and QCG-HF) were analyzed as continuous variables. The Student t test and χ 2 (or Fisher exact) test were used to compare the baseline clinical characteristics between the two groups. The discrimination performance of QCG scores for in-hospital outcomes was evaluated using receiver operating characteristic (ROC) curve analysis. The AUC values were compared using the DeLong test. The logistic regression model was used to estimate the odds ratios (ORs) and 95% CIs. Survival analysis was performed using the Kaplan-Meier method, and the Cox proportional hazard model was used to estimate the hazard ratios (HRs) and 95% CIs for the clinical outcomes. Multivariable analysis was performed with the inclusion of clinically relevant variables.
All tests were two-tailed and a P value <.05 was considered statistically significant. Statistical analyses were performed using R programming version 4.3.0 (The R Foundation for Statistical Computing).
Data of 1254 patients (716 from Seoul National University Bundang Hospital and 538 from Severance Hospital) were analyzed. Among the 1254 patients, 53 (4.2%) experienced in-hospital cardiac death. The baseline characteristics of the study population according to the in-hospital outcomes are shown in Table 1 . Compared with survivors, patients who died in the hospital were older, had a higher prevalence of ischemic heart disease, lower LVEF, and higher NT-proBNP levels. By contrast, atrial fibrillation (AF) was more frequent in survivors. The QCG-Critical and QCG-HF scores were significantly higher in patients who experienced in-hospital cardiac death than in survivors ( P <.001) ( Table 1 and Figure S1 in Multimedia Appendix 1 ).
Characteristics | Total (n=1254) | In-hospital cardiac death (n=53) | Survivors (n=1201) | value | |
Age (years), mean (SD) | 69.8 (14.7) | 74.0 (14.5) | 69.6 (14.1) | .03 | |
Male, n (%) | 673 (53.7) | 29 (54.7) | 644 (53.6) | .99 | |
Hypertension, n (%) | 843 (67.2) | 31 (58.5) | 812 (67.6) | .22 | |
Diabetes mellitus, n (%) | 499 (39.8) | 23 (43.4) | 476 (39.6) | .69 | |
Cerebrovascular disease, n (%) | 224 (17.9) | 7 (13.2) | 217 (18.1) | .47 | |
Chronic kidney disease, n (%) | 212 (28.3) | 11 (20.8) | 212 (28.3) | .69 | |
Ischemic heart disease, n (%) | 365 (29.1) | 25 (47.2) | 340 (28.3) | .005 | |
Valvular heart disease, n (%) | 217 (17.3) | 9 (17.0) | 208 (17.3) | >.99 | |
De novo HF , n (%) | 612 (48.8) | 29 (54.7) | 583 (48.5) | .46 | |
Atrial fibrillation, n (%) | 417 (34.7) | 10 (10.9) | 417 (34.7) | .03 | |
QRS duration≥120 ms, n (%) | 318 (25.4) | 17 (32.1) | 301 (25.1) | .32 | |
LVEF (%), mean (SD) | 35.3 (14.7) | 28.5 (11.9) | 35.6 (14.7) | .002 | |
NT-proBNP (pg/mL), mean (SD) | 10,373 (11,915) | 17,035 (1900) | 10,092 (11,879) | <.001 | |
scores, mean (SD) | |||||
QCG-Critical | 0.30 (0.21) | 0.57 (0.23) | 0.29 (0.20) | <.001 | |
QCG-HF | 0.65 (0.31) | 0.78 (0.18) | 0.64 (0.31) | <.001 |
a HF: heart failure.
b LVEF: left ventricular ejection fraction.
c NT-proBNP: N-terminal prohormone of brain natriuretic peptide.
d QCG: Quantitative electrocardiogram artificial intelligence system.
In the univariable logistic regression analysis, the QCG-Critical and QCG-HF scores were significant predictors of in-hospital cardiac death ( Table 2 ). Other than QCG scores, echocardiographic LVEF, NT-proBNP level, age, ischemic heart disease, and AF were significantly correlated with in-hospital cardiac death ( Table 2 ).
Variables | Univariate analyses | Model 1 | Model 2 | ||||||
OR (95% CI) | value | Adjusted OR (95% CI) | value | Adjusted OR (95% CI) | value | ||||
parameters (per 0.1 increase) | |||||||||
QCG-Critical | 1.66 (1.47-1.87) | <.001 | 1.68 (1.47-1.92) | <.001 | 1.59 (1.36-1.87) | <.001 | |||
QCG-HF | 1.21 (1.08-1.37) | .001 | 1.22 (1.08-1.39) | .002 | 1.02 (0.84-1.24) | .82 | |||
LVEF (per 5% decrease) | 1.21 (1.07-1.37) | .002 | 1.26 (1.09-1.45) | .001 | |||||
NT-proBNP (per 1000 pg/ml increase) | 1.03 (1.01-1.05) | <.001 | 1.04 (1.01-1.06) | <.001 | |||||
Age | 1.03 (1.00-1.05) | .03 | |||||||
Male | 1.05 (0.60-1.82) | .88 | |||||||
Hypertension | 0.68 (0.39-1.18) | .17 | |||||||
Diabetes mellitus | 1.17 (0.67-2.03) | .58 | |||||||
Chronic kidney disease | 1.22 (0.62-2.41) | .56 | |||||||
Cerebrovascular disease | 0.69 (0.31-1.55) | .37 | |||||||
Ischemic heart disease | 2.26 (1.30-3.93) | .004 | |||||||
Valvular heart disease | 0.98 (0.47-2.03) | .95 | |||||||
ADHF (vs de novo) | 0.78 (0.45-1.36) | .38 | |||||||
Atrial fibrillation | 0.44 (0.22-0.88) | .02 | |||||||
QRS duration>120 ms | 1.41 (0.78-2.55) | .25 |
a Adjusted for age, sex, hypertension, diabetes, chronic kidney disease, cerebrovascular disease, ischemic heart disease, valvular heart disease, heart failure type, atrial fibrillation, and QRS duration.
b When a variable was included as a covariate for adjustment, it was not adjusted for itself and QCG-Critical was added to the adjustment model (presented in italics).
c Adjusted for the same covariates as model 1 and further adjusted for left ventricular ejection fraction and N-terminal prohormone of brain natriuretic peptide.
d OR: odds ratio.
e QCG: Quantitative electrocardiogram.
f HF: heart failure.
g LVEF: left ventricular ejection fraction.
h NT-proBNP: N-terminal prohormone of brain natriuretic peptide.
i ADHF: acute decompensated heart failure.
After adjustment for age, sex, hypertension, diabetes, chronic kidney disease, cerebrovascular disease, ischemic heart disease, valvular heart disease, HF type, AF, and QRS duration, the two QCG scores remained significant predictors of in-hospital cardiac death. Moreover, the QCG-Critical score was an independent predictor of in-hospital cardiac death after further adjustment for echocardiographic LVEF and NT-proBNP level (OR 1.59, 95% CI 1.36-1.87; P <.001).
In a subgroup analysis, the QCG-Critical score was a significant predictor of in-hospital cardiac death regardless of the initial rhythm (AF or sinus rhythm), QRS width (wide or narrow), hypertension, diabetes, HF etiology (ischemic or nonischemic), HF type (de novo or acute decompensated HF), and LVEF (HF with reduced ejection fraction vs HF with preserved or mildly reduced ejection fraction), after adjustment for other clinical parameters ( Figure 2 ).
The QCG-Critical score was significantly higher in patients who experienced cardiac death within 1 day, 2 days, or during hospitalization than in survivors ( Figure 3 A). When the performance of the QCG-Critical score for predicting these events was analyzed using ROC curves, the AUC values for 1- and 2-day mortality and in-hospital cardiac death were 0.936, 0.917, and 0.821, respectively ( Figure 3 B).
Comparatively, the AUC values of echocardiographic LVEF and NT-proBNP level for predicting in-hospital cardiac death were 0.642 ( P <.001 vs QCG-Critical) and 0.720 ( P =.07 vs QCG-Critical) ( Figure 4 A). The AUC value of the QCG-Critical score (0.821) was significantly ( P =.02). higher than that of model 1 (0.705) established using traditional clinical variables, including age, sex, hypertension, diabetes, chronic kidney disease, cerebrovascular disease, ischemic heart disease, valvular heart disease, HF type, AF, and QRS duration. In addition, when the QCG-Critical score was added to model 1, it significantly enhanced the prediction for in-hospital cardiac death (AUC of model 1=0.705 vs AUC of model 1 with QCG-Critical=0.843; P <.001) ( Figure 4 B). When NT-proBNP and LVEF were further included in model 1 (model 2), the QCG-Critical score again demonstrated additional predictive value for in-hospital cardiac death compared to model 2 alone (AUC of model 2=0.787 vs AUC of model 2 with QCG-Critical=0.863; P =.01) ( Figure 4 C).
During a median follow-up of 2.7 years, 508 deaths occurred in the study population. To further analyze the performance of the QCG-Critical score for outcome prediction, we divided patients into three QCG-Critical score groups based on arbitrary cut-off values of 0.25 and 0.50 and then conducted survival analysis ( Figure 5 ).
After adjustment for age, sex, comorbidities, HF etiology and type, AF, and QRS widening, patients with higher QCG-Critical scores had significantly higher all-cause mortality rates during follow-up than those with lower QCG-Critical scores (<0.25). The adjusted HRs for patients with QCG-Critical scores between 0.25 and 0.50 and for patients with QCG-Critical scores higher than 0.50 were 1.57 (95% CI 1.28-1.93) and 2.69 (95% CI 2.14-3.38), respectively (all P <.001). With additional adjustment for LVEF and NT-proBNP to the previous model, the adjusted HRs were 1.61 and 2.27, respectively, consistent with the main analysis (Figure S2 in the Multimedia Appendix 1 ).
In a subgroup analysis, a higher QCG-Critical score (>0.50 vs ≤0.50) was significantly correlated with all-cause mortality during follow-up, regardless of the initial rhythm (AF or sinus rhythm), QRS width (wide or narrow), hypertension, diabetes, HF etiology, HF type (de novo or acute decompensated HF), and LVEF (HF with reduced ejection fraction vs HF with preserved or mildly reduced ejection fraction), after adjustment for other clinical parameters (Figure S3 in Multimedia Appendix 1 ).
Predicting outcomes in patients with HF is important for guiding management and improving prognosis [ 15 ] but is often hindered by the complexity of HF pathophysiology and the presence of other comorbidities. Recently, AI algorithms based on big data from medical records have been found to be helpful in predicting the outcomes of patients with HF [ 16 , 17 ]; however, these algorithms are difficult to apply in daily practice and their performance requires further improvements. In this study, the QCG-Critical score, a newly developed AI-based ECG score, was well correlated with early mortality and in-hospital cardiac death during the index after adjusting for traditional clinical risk factors. Moreover, the QCG-Critical score was an independent predictor of long-term all-cause mortality in this population, suggesting that this AI-based ECG score may serve as a novel biomarker for these patients.
ECG is a cost-effective, widely available, and easy-to-perform test, and is therefore often used as a first-line evaluation for patients with cardiovascular diseases. ST-elevation myocardial infarction is a quintessential disease where ECG evaluation is critical for a timely diagnosis. Although ECG is not deterministic for an HF diagnosis, several studies have demonstrated that some ECG features are correlated with the characteristics of HF [ 18 ]. In addition, the presence of AF or QRS widening may represent ECG features reflecting unfavorable underlying hemodynamics, thus correlating with a poor prognosis [ 19 , 20 ]. More subtle ECG changes have also been suggested as predictors of a poor prognosis in patients with HF; however, these require high levels of experience and skill for interpretation, which may limit their applicability [ 21 ].
Theoretically, the ECG signal may contain information regarding the electric and mechanical activities of the diseased heart beyond a physician’s perception. With the assistance of AI, ECG may provide valuable information beyond its current usage. For example, Attia et al [ 22 ] reported that LVEF reduction may be detected by ECG using AI. This new application of AI-ECG was reproduced by other researchers [ 23 , 24 ]. In this study, the QCG-HF score also showed good performance in predicting reduced echocardiographic LVEF of less than 40%, with an AUC value of 0.884 (Figure S4 in Multimedia Appendix 1 ). Notably, in the above-mentioned studies, the AI-ECG–predicted LVEF was correlated with the prognosis of patients with chronic HF, whereas the AI-based ECG score had a predictive value in patients with acute HF in this study. Thus, to the best of our knowledge, this study represents an initial effort in terms of predicting the outcomes of acute HF using AI-based ECG interpretation.
The QCG-Critical score was originally trained to detect critical medical conditions that may result in shock or mortality within 1 day [ 12 ]. In this study, the QCG-Critical score predicted early cardiac mortality in patients with acute HF with high accuracy. The AUC value of the QCG-Critical score was higher than that of echocardiographic LVEF for the prediction of in-hospital cardiac death and was also higher than the AUC value of the serum NT-proBNP level, but without statistical significance. Notably, the QCG-Critical score was available for all 1254 patients enrolled in the KorAHF study, whereas LVEF and NT-proBNP results were not available in 68 (5.4%) and 168 (13.4%) patients, respectively. Considering that the KorAHF study enrolled patients from tertiary centers in Korea, a high proportion of patients with acute HF might not have the opportunity to benefit from these echocardiographic or serum biomarker tests in real-world practice. Because ECG is a widely available evaluation tool and QCG scores are derived from ECG images, the QCG-Critical score may serve as an adequate alternative biomarker for risk stratification of patients with acute HF in real-world settings with limited resources. This score may also be useful even in well-equipped centers because it would be available immediately after the ECG exam, without requiring additional waiting for echocardiography or laboratory tests. This may be beneficial for timely risk stratification in the emergency department. The QCG-Critical score was not only correlated with in-hospital cardiac death but also showed a strong association with long-term mortality. In addition, the subgroup analysis demonstrated a consistent correlation between the QCG-Critical score and clinical outcomes. These results emphasize the potential of AI-based ECG interpretation as a novel biomarker in this field.
This study has several limitations. First, the study population predominantly consisted of Asian patients; hence, further studies are needed to validate our results across different ethnicities. Second, the AI algorithm tested in this study was derived from one of the participating centers (Seoul National University Bundang Hospital). However, there was a temporal difference between patient enrollment for algorithm training (2017 to 2019) and the test population (KorAHF enrollment, 2011 to 2014), and another external center (Severance Hospital) was involved in this study. Nevertheless, this may limit the generalizability of our findings. Third, the ECG format may affect the algorithm’s performance. Although the manufacturers of the ECG devices used in the two participating hospitals differed (Philips PageWriter TC 30 and TC 70 at Seoul National University Bundang Hospital and GE Healthcare MAC 5500 and MAC VU360 at Severance Hospital), there was no significant difference in the AI algorithm performance between the hospitals. However, because the system uses printed ECG images as input, there may be problematic scenarios where the qualities of the images influence the predictive power of the biomarkers. Although some recent AI algorithm–based studies suggest further interpretation analysis, the QCG system does not support gradient-weighted class-activation mapping or similar visualization for model explainability due to the custom network architecture used. Therefore, we could not evaluate which part of the ECG images the system uses for each prediction.
In conclusion, predicting outcomes in patients with acute HF using the newly developed AI-based ECG score appears to be feasible. Thus, this score may serve as a novel biomarker for patients with HF, potentially overcoming the cost and availability constraints of current biomarkers.
YC and MY are co-first authors, and SMK and DJC are co-corresponding authors. Author SMK can be reached at [email protected] for correspondence.
This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute, funded by the Ministry of Health & Welfare, Republic of Korea (grant RS-2023-00265933). This work was also supported by the Seoul National University Bundang Hospital Research Fund (grant 13-2024-0009).
JK and YC are employed by ARPI Inc and hold dual roles as CEO and Head of Research Collaboration Center, respectively. The other authors have no conflicts of interest to declare.
Distribution of QCG-Critical scores (Figure S1); Kaplan-Meier curves for long-term mortality according to the QCG-Critical scores and adjusted HRs with additional adjustment for LVEF and NT-proBNP (Figure S2); subgroup analysis results for predicting long-term mortality (Figure S3); performance of the QCG-HF score for diagnosing left ventricular dysfunction (Figure S4).
atrial fibrillation |
artificial intelligence |
area under the curve |
electrocardiogram |
heart failure |
hazard ratio |
Korean Acute Heart Failure |
left ventricular ejection fraction |
N-terminal prohormone of brain natriuretic peptide |
odds ratio |
Quantitative electrocardiogram |
receiver operating characteristic |
Edited by A Mavragani; submitted 27.08.23; peer-reviewed by S Khan, J Zeng; comments to author 08.02.24; revised version received 22.02.24; accepted 29.05.24; published 03.07.24.
©Youngjin Cho, Minjae Yoon, Joonghee Kim, Ji Hyun Lee, Il-Young Oh, Chan Joo Lee, Seok-Min Kang, Dong-Ju Choi. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 03.07.2024.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research (ISSN 1438-8871), is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Humanities and Social Sciences Communications volume 11 , Article number: 843 ( 2024 ) Cite this article
96 Accesses
Metrics details
Industrial The upgrading of industrial structure, as the main means of urban economic transformation, plays a crucial role in the process of achieving urban economic resilience construction. We conducted a study on the nonlinear impact mechanism of industrial structure upgrading on urban economic resilience based on panel data from 267 prefecture-level and above-level cities and above in China from 2008 to 2021, using globalization as a threshold variable. The obtained results demonstrated the following: (1) there existed a significant nonlinear relationship between industrial structure upgrading and rationalization and urban economic resilience, with a significant double threshold effect. (2) A robustness test was performed by removing extreme values from the sample, controlling for the time series and individual interaction terms while considering control variables, which did not change the basic conclusions based on the model. This demonstrated that the threshold regression model constructed in this study is robust and reliable. (3) From a regional heterogeneity perspective, the impact of industrial structure upgrading on urban economic resilience varied among different regions. Notably, industrial structure upgrading imposed a significant double threshold effect on urban economic resilience in the eastern and central regions, manifested as an inverted U-shaped trend. In the northeastern region, there was only a single threshold effect with globalization as the threshold variable, which still occurred on the left side of the inverted U-shaped curve, while no threshold effect was observed in the western region.
Introduction.
As the main focus area of economic activity and the core subject of mitigating various risks, enhancing the economic resilience of cities is a necessary means to effectively enhance the ability of their economic systems to resist risks and shocks (Fingleton and Palombi, 2013 ; Martin and Sunley, 2014 ; Cheng et al. 2022 ; Wang and Wang, 2021 ; Papaioannou, 2023 ; Fan et al. 2023 ). With the increasingly close global socioeconomic development, economic cooperation among countries is gradually moving toward a path of diversification, openness, and sharing (Zhou and Qi, 2023 ; Hynes et al. 2022 ; Jayasinghe et al. 2022 ; Gajewski, 2022 ). While countries worldwide share the fruits of economic development due to globalization, some countries have become “shock absorbers” for the cyclic regulation of the international economic system (Andrew et al. 2020 ; Ye and Qian, 2021 ; Ben and Ifergane, 2022 ). In particular, problems related to notable fluctuations in economic development, limited defense capabilities, and low competitiveness are particularly prominent in developing countries. These problems further exacerbate the vulnerability of cities in various countries in response to internal and external changes (Ženka et al. 2019 ; Wang et al. 2021 ; Mai et al. 2021 ). Therefore, how to build a strong and resilient economic system and employ resilience thinking to enhance the driving force and capacity of regional economic development has become an important research topic for countries globally, with the aim of promoting high-quality urban economic development.
The term resilience first evolved from the Latin word “resilio” (Yang et al. 2023 ). With the increase in the occurrence of uncertain events and external shocks, scholars have applied resilience in fields such as urban engineering resilience, ecological resilience, and economic resilience (Dario and Weterings, 2015 ; Paolo, 2017 ; Lemke et al. 2023 ; Du, 2023 ; Yu et al. 2023 ). The connotation of urban economic resilience has been widely investigated by the government, society, and academia, as it better conforms with the current stage of urban economic development in certain countries and the interpretation of the practical problems faced (Du et al. 2023 ; Gai and Yang, 2023 ; Hui and Tan, 2023 ). Urban economic resilience refers to the ability of a city to prevent and resist risks, as well as maintain efficient and sustainable economic development during a specific period (Drobniak, 2017 ; Tan et al. 2017 ; Pashapour et al. 2019 ; Erika and Mangirdas, 2020 ). Today’s world is vulnerable to severe impacts such as economic crises, epidemics, and natural disasters. Some regions may continue to maintain stable economic growth after impact, while others may suffer heavy losses and fail to recover (Deng et al. 2023 ; Wang et al. 2023 ; Lee and Wang, 2023 ). The reason for this difference is that countries with greater urban economic resilience often exhibit characteristics such as dynamic balance, redundant buffering, and self-healing, which can enable these countries to quickly eliminate risks and automatically adjust and recover, thus effectively resisting external shocks and mitigating internal disasters (Cheng et al. 2023 ; Zhou and Qi, 2023 ). Notably, 2008 and 2019 were critical time points in terms of global fluctuations and changes. Under the impacts of the global crisis and the COVID-19 pandemic, respectively, Compared with developed countries such as the United States and Japan, although China’s economic development has also encountered obstacles, its economic growth rate has decreased by 1.8% and 0.1% respectively compared to the previous year, But China’s long-term economic fundamentals have not changed, and the characteristics of sustained economic recovery, notable development potential, high resilience and broad space have not changed (Hao, 2023 ; Wang et al. 2024 ; Yang, 2023 ). For example, under the influence of the COVID-19 pandemic, the economies of the United States and Japan shrank by 3.5 and 5.3%, respectively, in 2021. On the one hand, the economy of China benefited from a series of policy documents issued by the Chinese government on “optimizing the environment, expanding the domestic demand, stabilizing growth, and promoting development”. On the other hand, it benefited from a strong labor force and high consumer market supply capacity.
Industrial structure upgrading is important for exploring the economic resilience of Chinese cities (Drobniak, 2017 ; Zhou et al. 2019 ; Betts and Buzzanell, 2022 ). Especially at this stage, China’s economy is transitioning from high-speed growth to high-quality development, and industrial structure upgrading plays an important role in the urban economy resilience process in different regions and stages (Zhang, 2022 ; Yin et al. 2023 ). On the one hand, industrial structure upgrading can cause an acceleration in regional industrial structure transformation from traditional high energy-consuming industries to high-tech industries. In this process, with the emergence of an advanced industrial structure and a specialized division of labor, new economic growth paths can be created, which can enhance the ability of cities to withstand market risks and can facilitate resilient growth of the urban economy (Tan et al. 2017 ; Cheng et al. 2023 ; Li et al. 2024 ). On the other hand, industrial structure upgrading can lead to enhancement in the service industry, but a high proportion of the service industry can easily cause the problem of hollowing out industries, which is not conducive to improving urban economic resilience (Gai and Yang, 2023 ; Hui and Tan, 2023 ). It should be noted that industrial structure upgrading entails a long and tortuous process, and its impact on urban economic resilience cannot be achieved overnight. Therefore, the causal relationship between industrial structure upgrading and urban economic resilience is not direct nor obvious. However, the impact of the former on the latter still objectively occurs through various transmission mechanisms. Moreover, due to the poor coherence and sustainability of relevant systems, this impact relationship may be repetitive and not simply linear.
From 1492 to 2023, globalization increased from the 1.0 era to the 4.0 era (Roberta et al. 2015 ; Aida et al. 2016 ; Gereffi et al. 2022 ). Globalization has firmly woven all countries into the network of the world system, and the commerce, economy, and system of each country have undergone tremendous transformation (Dunn, 2020 ; Carlos et al. 2022 ). In the globalization process, achieving high-quality and sustainable economic development has become a major challenge for the international community (Ngo, 2023 ). Since its accession to the World Trade Organization (WTO), China has received a large amount of foreign direct investment due to its own resource endowment. Foreign direct investment has, to a certain extent, accelerated the process of high-quality development of China’s urban economy, which is mainly reflected in two aspects: on the one hand, foreign direct investment can effectively promote industrial structure upgrading and optimization (Tao et al. 2023 ). Foreign direct investment usually leads to the introduction of technology, management, and market factors, accelerating the transformation of traditional Chinese industries to high-tech and high value-added industries and promoting sustainable development of the urban economy (Anis and Andreea, 2023 ). On the other hand, foreign direct investment not only provides financial support but also introduces advanced technology and management experience, providing Chinese industries with a larger market and more abundant resources (Knoke et al. 2022 ; Gereffi et al. 2022 ). Production scale increase, product quality improvement, and cost reduction further enhance the competitiveness of China’s industries (Carlos et al. 2022 ; Johnson and Mundell, 2023 ). Therefore, we must consider several questions: under the acceleration of globalization, what is the impact of industrial structure upgrading (with a focus on industrial structure upgrading and rationalization) on urban economic resilience? Are there certain stage characteristics? These questions should be explored in depth. To this end, this study links the above three aspects and focuses on determining whether there exists a threshold effect in terms of the impact of industrial structure upgrading on urban economic resilience with globalization as the threshold variable, examining its mechanism and effect, and investigating whether there is regional heterogeneity to provide an empirical basis for relevant departments to formulate targeted policies.
The remainder of this study is organized as follows: in the second section, a literature review is provided, in which the existing research on industrial structure upgrading, globalization, and urban economic resilience is summarized, providing a sound basis for this analysis. The third section constructs a theoretical framework and research hypotheses, elaborates on the theoretical basis of industrial structure upgrading, explores the impact of industrial structure upgrading and rationalization on urban economic resilience in the context of globalization, and proposes research hypotheses. In the fourth section, the research design is described in detail, including model settings, variable selection, data sources, and descriptive statistical analysis. In the fifth section, the empirical tests are introduced, including threshold effect assessment, threshold panel model-based estimation, and regional heterogeneity analysis. Finally, the sixth and seventh selections, respectively outline the conclusions, policy recommendations, and future prospects of this article.
The study of urban economic resilience has become an important topic for countries to explore high-quality and sustainable economic development (Guo et al. 2023 ). Boschma ( 2015 ) first proposed the concept of economic resilience, stating that economic resilience is the ability of an economic system to absorb shocks without catastrophic changes in its basic functional organization. Thereafter, the academic community investigated urban economic resilience from the perspectives of regional and geographic economics (Wang and Ge, 2023 ; Wang et al. 2024 ). Regional economics mainly explores how to cope with the decline in the regional economy, while geographic economics largely focuses on the study of spatial differences, correlations, and influencing factors of urban economic resilience (Zhang et al. 2023 ). In recent years, academic research on urban economic resilience has focused on three main aspects (Jesse, 2023 ; Cheng et al. 2022 ): measurement methods, influencing factors, and the impact of industries on urban economic resilience. First, from the perspective of measurement methods for urban economic resilience, scholars have measured the urban economic resilience index based on methods such as the core variable method and comprehensive indicator method from different disciplinary backgrounds (Qiang et al. 2020 ; Wang and Wang, 2021 ). However, due to the varying focuses, there is no consensus at present. Second, from the perspective of the influencing factors of urban economic resilience, scholars have considered that fiscal gaps, geographical location conditions, and resource endowments are key factors that constrain urban economic resilience (Gan and Chen, 2021 ; Cheng et al. 2022 ; Wang and Wang, 2021 ). The processes of urbanization, technological innovation, economic development, and industrial agglomeration are major factors driving urban economic resilience (Du et al. 2023 ; Gai and Yang, 2023 ). Third, from the perspective of the impact of industries on urban economic resilience, existing research has focused on two main aspects: (1) from a microscopic perspective, the impact of a single industry, such as the digital industry, financial industry, or manufacturing industry, on urban economic resilience has been examined; and (2) from a macroscopic perspective, the impacts of industrial structure upgrading, adjustment, transformation, and agglomeration on urban economic resilience have been explored. For example, Feng et al. ( 2023 ) empirically determined that industrial structure rationalization and upgrading are important ways for regional integration to affect urban economic resilience. However, the policy effects of regional integration on economic resilience vary over time, by region, and by urban structure. Zhang et al. ( 2023 ) noted that regional economic resilience is closely related to the state of the industrial structure, and there exists a spatiotemporal correlation in the evolution of the two systems. In the literature review process, we could conclude that the existing research on the relationship between industry and urban economic resilience has not yet reached a consensus, both at the micro- and macroscopic levels, thus providing a theoretical basis and new ideas for this study.
Regarding the relationship between globalization, industrial structure upgrading, and urban economic resilience, studies have mostly focused on the relationship between globalization and industrial structure upgrading, as well as the relationship between industrial structure upgrading and economic resilience (Carlos et al. 2022 ). In terms of the relationship between globalization and industrial structure upgrading, studies have suggested that globalization can promote industrial structure upgrading through the division of labor and cooperation in the industrial structure (Dunn, 2020 ) and that globalization can promote industrial structure upgrading through technological innovation and industrial transformation (Ngo, 2023 ). Globalization can improve factor allocation efficiency and drive industrial structure upgrading by influencing the direction and quantity of factor flow. In addition, scholars have noted that the development of globalization encompasses various stages, while its impact on the industrial structure is also cyclical, which can lead to instability in the impact of globalization on industrial structure rationalization and upgrading. In terms of the relationship between globalization and urban economic resilience, there are three specific viewpoints: globalization imposes a reducing effect on urban economic resilience (Tao et al. 2023 ), globalization exerts an expanding effect on urban economic resilience (Anis and Andreea, 2023 ), and the impact of globalization on urban economic resilience is dynamic (Martin et al. 2016 ). In terms of the relationship between industrial structure upgrading and urban economic resilience, industrial structure rationalization and upgrading can help to reduce the impact of international markets by improving the industrial configuration and quality level, providing greater development space for adaptive structural adjustment after impact occurrence and thus continuously enhancing urban economic resilience.
In the literature, scholars have empirically evaluated industrial structure upgrading and globalization as important factors affecting urban economic resilience based on econometric models, geographic models, and spatial econometric models (Maria, 2023 ; Cheng et al. 2023 ); however, few scholars have explored the relationships among these three factors. Within the context of globalization, the transformation of the new international division of labor model profoundly affects the process of industrial structure adjustment in various countries worldwide. However, existing research has overlooked the moderating effect of globalization on industrial structure upgrading and urban economic resilience. Moreover, the literature has mostly focused on analyzing the linear effect of industrial structure upgrading on urban economic resilience, and various conclusions have been obtained. This also reflects the complexity of the relationship between the two aspects, which suggests that they may not be characterized by a simple linear relationship, namely, there may be a nonlinear relationship. Compared with the literature, the marginal contribution of this study lies in coupling industrial structure upgrading, globalization, and urban economic resilience. It was preliminarily determined that the impact of industrial structure upgrading on urban economic resilience is nonlinear, and globalization was used as a threshold variable. Panel data from 267 prefecture-level cities and above in China from 2008 to 2021 were selected, and we empirically examined the nonlinear relationship between industrial structure upgrading and urban economic resilience and studied its stage characteristics. Moreover, we conducted robustness tests.
Industrial structure upgrading is important for exploring the economic resilience of Chinese cities. Especially at the current stage, China’s economy is shifting from high-speed growth to high-quality development, and industrial structure upgrading fulfills an important role in the resilience process of the urban economy in different regions and stages (Feng et al. 2023 ). The theory of industrial structure was first proposed by Fisher, who divided the overall industrial structure into the primary industry, secondary industry, and tertiary industry (Zheng et al. 2023 ; Li, 2024 ). William Petty established the theory of industrial structure upgrading, which refers to the process of industrial structure transformation from lower to higher stages, and constructed a theoretical analysis framework for industrial structure upgrading, including two aspects: the rationalization and advancement in the industrial structure (Feng et al. 2023 ). This theoretical framework laid the foundation for subsequent related research. With the deepening and development of theoretical research, the theory of industrial structure upgrading is the result of the joint action of the two forces of industrial structure upgrading and rationalization. There is widespread consensus among scholars at home and abroad, and this approach has been widely applied in fields such as social crises, regional environments, grassroots governance, and high-quality urban development. This has provided a new research approach for exploring the impact of industrial structure upgrading on urban economic resilience.
Since China joined the WTO, the Chinese economy has been further integrated into the world economy and become an important destination for FDI worldwide. Since then, FDI has become an important manifestation of globalization (Zhang et al. 2020 ). Globalization not only enhances the frequent exchange of resources such as technology, capital, and labor among countries but also promotes increasingly close economic development relationships among countries. Within the context of globalization, while countries achieve effective resource allocation in their industrial structures, they also move toward advanced and rational industrial structures, thereby promoting high-quality development of urban economic resilience through the release of structural dividends. In other words, industrial structure upgrading has become an important driving force for urban economic resilience within the context of globalization (Fig. 1 ).
In the figure, S denotes suppression, P denotes promotion, C denotes connection, and R denotes regular.
Industrial structure upgrading involves a process of rationalization. The singularity of the industrial structure is not conducive to urban entities overcoming internal and external market risks within an uncertain environmental context. At the primary stage of globalization, foreign direct investment can not only solve the problem of imbalanced and insufficient development of the regional industrial structure via the utilization of local resource endowments but also reduce production costs, provide stable and long-term global competitive advantages, and promote the intensive and large-scale development of local industries by achieving a reasonable division of labor and allocation of various industries in the regional industrial chain. The entry of labor- and capital-intensive industries into developing countries creates a large number of employment opportunities, promotes the flow of surplus labor from the agricultural sector to the nonagricultural sector, and enhances regional resistance to internal and external market risks. With the improvement in the globalization level and the increase in foreign investment, the formation of a diversified industrial system in cities is accelerating. Cities rely on industrial diversification, product richness, high-added value, and asynchronous industrial cycles to avoid drastic fluctuations in output and employment, thereby enhancing the ability of the urban economic system to overcome risks and adapt to shocks. Therefore, Hypothesis 1 was proposed.
Hypothesis 1: Industrial structure rationalization exerts a significant positive impact on urban economic resilience at different stages of globalization.
Another main feature of industrial structure upgrading is the upgrade process. At the early stages of globalization, foreign direct investment in factory construction, to some extent, dealt a heavy blow to traditional industries in developing countries, reducing the ability of urban economic entities to eliminate market risks. With the increase in globalization, on the one hand, FDI can promote the return of technology, labor, and capital and accelerate the replacement of high value-added industries and low value-added industries. On the other hand, foreign direct investment can not only promote the transition from traditional resource-intensive industries to resource-intensive industries but also attract more high-quality labor and technological resources, thereby increasing the resistance of urban economic systems to risks and the ability to adapt to impact factors. Therefore, Hypothesis 2 was proposed.
Hypothesis 2: There may be significant stage differences in the impact of industrial structure upgrading on urban economic resilience between different levels of globalization.
Model settings.
As mentioned earlier, industrial structure upgrading is accompanied by the reasonable flow and optimized allocation of production factors between industries and countries, which exerts a certain impact on urban economic resilience. Globalization also plays an important role in this process, but confirmation of the existence of threshold effects still requires further empirical testing. Therefore, in this study, globalization was adopted as a threshold variable, urban economic resilience was used as the dependent variable, and industrial structure upgrading and rationalization were employed as the core explanatory variables. Then, the following threshold panel model was constructed (Dou and Gao, 2023 ; Zheng et al. 2023 ):
where UER it denotes the urban economic resilience of city i during the t-th period; GL it denotes the globalization of city i during the t-th period as a threshold variable, with r 1 , r 2 …, r n representing n threshold values; RIS it and AIS it denote industrial structure rationalization and advancement, respectively; α 1 , α 2 , and α n, and β 1 , β 2 , and β n denote the regression coefficients for different threshold intervals; X it is a series of control variables; and θ and k are the regression coefficients of the control variables. Note that the only difference between Models (1) and (2) is the core explanatory variables (without considering differences in the parameter values), which are industrial structure upgrading and industrial structure rationalization, respectively. For the sake of brevity, Models (1) and (2) are distinguished by their core explanatory variables and are thus referred to as the RIS and AIS models, respectively.
Explained variable, calculation model for urban economic resilience.
The regional economic foundation often determines the lower limit of a region’s ability to withstand shocks, which, to a certain extent, affects its resilience and recovery level (Wang et al. 2021 ). Notably, urban economic resilience is not only related to the total economic output but also closely related to the economic structure. The higher the total economic output is, the higher the urban economic resilience when facing risks. The GDP is an appropriate indicator of the total economic output of a city, while a reasonable economic structure is also a key factor in ensuring healthy and dynamic growth of the urban economy. In contrast to the former, the economic structure more strongly reflects changes in economic growth rates. The main research methods for measuring urban economic resilience are the core variable method and the comprehensive indicator method, but a consensus has not yet been reached. Considering the representativeness and continuity of the core variable method, this study refers to existing research, and the output method was used to measure urban economic resilience. This indicator can directly reflect the degree of change in the urban economy in the face of pressure and shocks (Feng et al. 2023 ). Compared with previous studies, this approach avoids the subjectivity of using a comprehensive indicator system, but it neglects the relationships and dependencies between factors and does not fully reflect the actual situation. The specific calculation method is as follows:
where GDP denotes the standardized value of the GDP and ∆GDPV is the standardized value of the absolute change in the GDP growth rate in adjacent years. As the product of multiple standardized values is very small, to better visualize the differences in urban economic resilience, the regression coefficient is multiplied by 100.
According to existing research, industrial structure rationalization and advancement were used as alternative indicators to measure urban industrial structure upgrading (Gan and Chen, 2021 ; Yin et al. 2023 ). As expressed in the equation below, rationalization of the industrial structure (RIS) mainly reflects the coupling relationship between the input and output, where i denotes the primary industry, secondary industry, or tertiary industry, Y represents industrial economic output, L represents labor input, and I represents industrial sector (i = 1,2,3). Upgrading of the industrial structure (AIS) largely reflects the proportion of the service industry, measured as the ratio of the added value of the tertiary industry to that of the secondary industry.
The threshold variable selected in this study is globalization (GL). After China’s accession to the WTO, China gained a large amount of foreign direct investment due to its resource endowment. With the increasingly close relationship between China’s economic development and the global economy, FDI transformed globalization into a localization force through location selection, which, to a certain extent, promoted the development of the regional economy. However, the greater the degree of closeness to the global economy is, the greater the occurrence probability of unpredictable risks, thereby exacerbating the vulnerability of regional economic resilience. Therefore, the per capita foreign direct investment amount was selected as a characterization indicator to measure globalization.
(1) Industrial agglomeration level of industrial enterprises. We selected the total number of industrial enterprises/urban construction land area as an indicator to measure the level of industrial enterprise agglomeration (Wang et al. 2021 ). The agglomeration of industrial enterprises can accelerate the convergence of regional enterprises, goods, services, and highly skilled labor. The agglomeration of a large number of intermediate inputs, high-level services, and human capital imposes a significant positive effect on urban economic resilience. However, the excessive agglomeration of industrial enterprises could also lead to significant negative externalities with respect to factors such as infrastructure, residents’ health, and the ecological environment, reducing the responsiveness of urban economic systems to external shocks. (2) Local financial gap. We selected the indicator of (expenditure within the local fiscal budget—revenue within the local fiscal budget)/revenue within the local fiscal budget to measure fiscal gaps (Xiong et al. 2023 ). The level of local finance fulfills an important role in achieving the optimal allocation of regional resources. Research has shown that the smaller the local fiscal gap, the more capable local governments are of achieving a balance and structural optimization between the total social demand and supply when facing external shocks. In contrast, the greater the local fiscal gap is, the lower the ability of local governments to self-adjust and repair in the face of external shocks. (3) Technology investment level Footnote 1 . We selected scientific and technological investment/local general public budget expenditure as a measurement indicator (Zhang et al. 2020 ). Improving the technology investment level can, to a certain extent, accelerate the elimination of traditional industries and the growth of emerging industries in a given region, enhance the competitive advantage of the entire industry chain in key fields, and overcome the monopoly of core technologies in certain foreign fields, thus increasing urban economic resilience. (4) Population density. We selected the total number of permanent residents in cities within the province/the area of provincial jurisdiction as a measurement indicator (Tan et al. 2020 ). The population density is a key factor affecting urban economic resilience and varies across different regions. The regional agglomeration effect generated by the population density can cause various resources to gather in cities, while population aggregation can cause various high-quality resources to accumulate in cities, generating a positive agglomeration effect, which facilitates the construction of regional spatial governance systems, high-quality and intelligent public services, and infrastructure systems and has a certain significance in building urban economic resilience. When there is a turning point in the population density, the phenomenon of excessive development and utilization of regional resources may occur, requiring local governments to consume more financial and material resources to solve problems such as environmental damage, severe resource depletion, and high carrying pressure caused by population agglomeration. This exerts a significant inhibitory effect on urban economic resilience, so the total number of permanent residents in cities within the province/the area of provincial jurisdiction was chosen as a measurement indicator. (5) Infrastructure level. We selected the coverage of public transportation routes/total population as a measurement indicator (Zhang, 2022 ). Infrastructure, including transportation and communication facilities, provides the basic guarantee for urban economic activities. The convenience of infrastructure is directly related to the development of local economic resilience. Research has shown that improving infrastructure is not only conducive to high-quality development of the local economy but also enhances the ability of urban economic systems to overcome risks and adapt to shocks. (6) The economic development level was measured by the total GDP/total population ratio in this study (Deng et al. 2023 ). The economic strength and development level of a city significantly impact its resilience: the stronger the economy is, the more reasonable the structure, and the higher the innovation ability is, the greater the resilience level of a city. Research has indicated that the urban economic development level is related to the ability to withstand macroeconomic and financial risks.
We selected panel data for 267 prefecture-level cities and above in China (Excluding Hong Kong, Macao, Taiwan, Xinjiang, and Xizang) from 2008 to 2021 for empirical analysis. The panel data were obtained from the Statistical Yearbook of Urban Construction in China and the Statistical Yearbook of Chinese Cities from 2008 to 2021. Descriptive statistics of all variables are listed in Table 1 . The differences between the minimum and maximum values for measuring the urban economic resilience index were significant, indicating a significant difference in urban economic resilience among the 267 prefecture-level cities and above in China between 2008 and 2022. The maximum value of urban economic resilience was observed for Shanghai in 2019, with an urban economic resilience of 20.033, while the minimum value was obtained for Dingxi city in 2018, with an urban economic resilience of 0.000. The minimum and maximum values of the industrial structure rationalization indicator were 0.581 and 29.114, respectively, while the minimum and maximum values of the industrial structure upgrading indicator were 0.139 and 5.350, respectively. This reflects the considerable differences in industrial structure rationalization and upgrading among the 267 prefecture-level cities and above between 2008 and 2022. Similarly, there was a significant difference between the minimum value (−2.028) and the maximum value (9.970) of the globalization indicator, indicating a significant difference in globalization among the 267 prefecture-level cities and above between 2008 and 2022.
Hausman test.
Before conducting the Hausman test, we first applied the Durbin–Wu–Hausman (DWH) test to assess for endogeneity issues in the benchmark panel model (Zheng et al. 2023 ). According to the test results, the P value of the DWH test was 0.000, indicating that the null hypothesis of “all explanatory variables are exogenous” could not be rejected at a significance level of 1%. Therefore, the model did not exhibit endogeneity issues. Finally, we conducted a Hausman test of the relationship between industrial structure rationalization and upgrading and urban economic resilience. The results indicated that the P value of both models (1 and 2) in the Hausman test was 0.0000, indicating that the original hypothesis of random effects could be rejected at the 1% significance level and that the alternative hypothesis of fixed effects could be accepted. Therefore, we utilized the panel fixed effects model to estimate Models 1 and 2.
Before establishing a specific threshold effect model, two important tests were conducted: one involved testing for the existence of threshold effects, with the aim of exploring whether the parameter spaces within different threshold intervals divided by the threshold values significantly differ; the other was using the bootstrap method for consistency testing, with the aim of determining whether the estimated threshold value is consistent with the actual value. The former is generally evaluated by the F statistic, while the latter is assessed by the likelihood ratio (LR) statistic.
The test results for the industrial structure rationalization and upgrading models with globalization as the threshold variable are provided in Table 2 . First, a single threshold test was conducted, and the corresponding F values were 193.71 and 514.98, respectively, while the P values were 0.0033 and 0.0000, indicating that there exists a threshold effect in both the rationalized and advanced industrial structure models. Second, the double threshold effect existence test was performed, with corresponding F values of 55.85 and 90.97, respectively, and P values of 0.0700 and 0.0133, respectively, which were significant at the 10 and 5% levels. The original hypothesis of the existence of a single threshold could be rejected, and it could be considered that both models contain a double threshold effect. Finally, a triple threshold test was conducted, and the P values at this time indicated that the original hypothesis of the existence of double thresholds could not be rejected. Therefore, in this study, a dual threshold effect model was chosen for estimation. The model can be formulated as follows:
We used the estimation method of minimizing the sum of squares of residuals to determine the specific threshold values, and the results are provided in Table 3 . The first and second threshold values for the model variables of industrial structure rationalization and upgrading (GL) were 6.8724 and 6.5514, respectively, and 7.5034 and 6.8724, respectively. Table 3 also provides the confidence intervals for each threshold value at a 95% confidence level.
Next, we performed a consistency test between the estimated threshold values and the actual values. Based on the estimation results listed in Table 3 , a likelihood ratio function graph was created. The horizontal axis in Fig. 2 represents the threshold value for globalization, the vertical axis represents the likelihood ratio function value, and the dashed line represents the critical value at the 95% confidence level. Here, the upper half represents the confidence interval for the first threshold of globalization, while the lower half represents the confidence interval for the second threshold of globalization. According to Hansen’s likelihood ratio test model, for LR(γ) > C(θ), the original hypothesis can be rejected. For θ = 5%, the critical value of the LR statistic is 7.35. According to Fig. 1 , the threshold values of the LR statistic corresponding to industrial structure rationalization and globalization of the advanced models were significantly lower than the critical values, so the above threshold estimates could be considered true and effective.
Dual threshold values and likelihood ratio function of GL in the RIS and AIS model.
Regression estimates of the impact threshold of industrial structure upgrading on urban economic resilience are provided in Table 4 . Among them, the regression results for basic Models (I) and (V) indicated that the relationship between industrial structure upgrading and globalization could be divided into three intervals by the threshold variable of globalization, and there were significant differences between the different intervals. At a lower globalization level, the impact of industrial structure rationalization on globalization was significantly positive at the 1% level, while the impact of industrial structure upgrading on globalization was not significantly negative. On the one hand, this suggests that at a lower globalization level, industrial structure rationalization could significantly improve urban economic resilience. When faced with external shocks, the economic resilience of cities significantly differs. Cities with a reasonable industrial layout exhibit higher economic resilience, namely, cities with a more reasonable industrial structure can quickly adjust their industrial structure, thereby obtaining more persistent and robust economic resilience. On the other hand, this result indicates that the level of industrial upgrading in China is still low, and products are at a disadvantage in the global market competition process. At this time, globalization is still dominated by the negative siphon effect.
Notably, at the early stages of globalization, industrial structure rationalization is the foundation for the economy to overcome external shocks, while later, the same process is the source of sustained economic resilience. When globalization crosses the first threshold and reaches the intermediate stage, the regression coefficient between industrial structure rationalization and urban economic resilience remains significantly positive, The above research confirms the validity of Hypothesis 1. This may occur because, with the acceleration of globalization, industrial structure rationalization eliminates market barriers between regions and internationally through the rational division of labor and the allocation of various industries within the industrial chain, achieving effective resource allocation, releasing structural dividends, and providing stable and long-term global competitive advantages, thus enhancing the ability of urban economic systems to mitigate risks and adapt to shocks. The impact of an advanced industrial structure on urban economic resilience within this interval shifted from negative to positive, indicating that with the improvement in the level of the advanced industrial structure, the return of technology, talent, and capital is promoted, accelerating the replacement of high value-added industries and low value-added industries, improving the demand income elasticity of high value-added products, reducing the operational risks of urban entities, and increasing the ability of urban economic entities to withstand market risks. When globalization crosses the second threshold and reaches a high level, the impact coefficients of industrial structure rationalization and upgrading on globalization are significantly positive at the 1% significance level, with regression coefficients of 0.037 and 0.780, respectively. The main reason is that with the rapid development of globalization, the industrial structure layout becomes more reasonable, which is conducive to the formation of a diversified industrial system in cities. Cities rely on industrial diversification, product richness, high-added value, and asynchronous industrial cycles to prevent sharp fluctuations in output and employment, thereby improving the ability of the urban economic system to overcome risks and adapt to shocks. In contrast, the deepening of globalization promotes the transition from traditional resource-intensive industries to resource-intensive industries, thus vigorously enhancing the level of industrial intensification and, in return, the ability of the urban economic system to withstand risks and adapt to shocks. The above research confirms the validity of Hypothesis 2.
After benchmark regression, to ensure the reliability of the conclusions obtained, we conducted two robustness tests. First, considering that extreme values in the sample may impact the results, Beijing (which attained the highest urban economic resilience value in 2019) and Dingxi (which attained the lowest urban economic resilience value in 2018) were excluded, and the model was again regressed to obtain robustness test results (II and VI in Table 4 ). Second, robustness tests were performed by controlling for time and individual interaction terms, and the results of robustness tests III and VII are listed in Table 4 . Finally, by including control variables for robustness testing, additional robustness test results (IV) and (VIII) were obtained, as provided in Table 4 . The three additional control variables were the market size, financial development level, and cultural soft power (Guo et al. 2023 ; Zhang and Yao., 2023 ), where the market size can be measured by the proportion of the total retail sales of consumer goods to the GDP. The financial development level can be measured by the ratio of the bank loan scale to the GDP. Cultural soft power can be measured by the logarithmic value of book collection. Compared with the estimation results of the basic model, the core explanatory variables in the robustness test results—namely, industrial structure rationalization and industrial structure upgrading—basically exhibited regression coefficients of the same sign and significance within the various threshold ranges of globalization (represented by GL L , GL M , and GL H ). For example, robustness test result II was relatively close to that of basic Model I, while robustness test result VI was relatively close to that of basic Model V. The sign and significance of the regression coefficients for the core explanatory variable, i.e., industrial structure rationalization, within the various threshold intervals of globalization were highly similar. However, after removing the extreme values in the sample data, the regression coefficient and significance of industrial structure upgrading exhibited significant changes within the various threshold ranges of globalization, with significance decreasing from the previous 1% level to the 5% level. However, it should be noted that although the removal of extreme values generated a certain impact in this study, the existence of individual extreme values did not affect the basic conclusions of the model. For example, robustness test result III was related to basic Model I, while robustness test result VII was related to basic Model V. The regression coefficients for the core explanatory variables of industrial structure rationalization and industrial structure upgrading did not show significant changes within the various threshold ranges of globalization, as Model VII incorporated time and individual interaction terms, and the main difference between the two models was only the fact that within the third threshold range (GL H ) of globalization, the significance of the regression coefficient of Model VI slightly decreased. However, it was still significantly positive at the 5% level, and the conclusion remained significant. This indicates that, regardless of whether time and individual interaction terms were controlled, the conclusions of model analysis remained consistent. The results of robustness tests (IV) and (VIII) showed that the three additional control variables slightly but significantly affected the basic regression results. Therefore, the threshold regression model constructed based on panel data from 267 prefecture-level and above cities in China for assessing the impact of industrial structure upgrading on urban economic resilience was robust and reliable.
To eliminate the interference of regional heterogeneity factors and verify the moderating effect of globalization on industrial structure upgrading and urban economic resilience, we divided the 267 prefecture-level and above cities into four major regions—eastern, central, northeastern, and western regions—and conducted threshold effect testing and threshold model estimation. The specific results are listed in Table 5 . The industrial structure rationalization and upgrading models in both the eastern and central regions passed the double threshold test. The threshold values for globalization of the two models in the eastern region were 6.54 and 6.88, respectively, and 6.44 and 6.88, respectively, while the threshold values for globalization of the two models in the central region were 6.63 and 6.97, respectively, and 6.97 and 7.23, respectively. The industrial structure rationalization and upgrading models in the western region and the industrial structure upgrading model in the northeastern region did not pass the double threshold test, while the industrial structure rationalization model in the northeastern region exhibited only a single threshold effect.
The threshold panel regression results for the eastern, central, western, and northeastern regions are listed in Table 6 , which indicates certain differences between the regression results for each region and the national level.
In the eastern region, when globalization fell below the first threshold, the regression coefficients of industrial structure rationalization and upgrading on globalization were 0.056 and 0.275, respectively, and both passed the 1% significance test. When globalization crossed the first threshold but remained lower than the second threshold, the regression coefficients of both were significant. After globalization crossed the second threshold, the regression coefficients of the two variables were 0.109 and 1.546, respectively, and both passed the 1% significance test. These results indicated that with the continuous improvement in globalization, the relationship between industrial structure rationalization and upgrading and urban economic resilience in eastern China exhibits an inverted U-shaped trend. The reason for this may be that the eastern region has become the preferred region for accepting foreign investment due to its inherent resource endowment and location advantages. On the one hand, with increasing foreign investment, cities in the eastern region have launched a prelude to traditional industrial transformation. The optimization and improvement in industrial and product structures have, to a certain extent, increased the overall development level of regional industries and international competitiveness while also driving the improvement in urban economic resilience. On the other hand, with the limited urban carrying capacity in the eastern region, foreign direct investment is expected to be gradually transferred to the central and western regions, resulting in the phenomenon of urban economic resilience first increasing and then decreasing.
The threshold effect of industrial structure upgrading and rationalization on urban economic resilience in the central region was similar to that in the eastern region, also exhibiting an inverted U-shaped trend. On the one hand, in recent years, the central region has become an important position for China’s development of strategic emerging industries, committed to solving problems such as industrial structure convergence and chain fragmentation, and relies on diversified industrial structures (strategic emerging industries) to accelerate the rational allocation of industrial structures and regions to enhance the synergy between the secondary industry (manufacturing) and tertiary industry (productive services), improve the correlation and cohesion between industries within the region, promote the spillover of knowledge and innovation between departments, accelerate the formation of regional industrial economies of scale, and achieve long-term and robust urban economic resilience. On the other hand, the central region has adjusted its industrial structure through specific plans such as urban renewal, accelerating the circulation of key technological elements between different industries, improving the efficiency of regional resource allocation, increasing the size of the regional economy, and taking the lead in entering the post-industrialization stage, which, to a certain extent, serves to improve urban economic resilience.
Industrial structure rationalization exerted a single threshold effect on urban economic resilience in the northeastern region. When globalization occurred below the threshold value of 6.63, the regression coefficient of industrial structure rationalization was −0.003, which is not significant. When globalization exceeded the threshold value of 6.63, the impact of industrial structure rationalization on urban economic resilience remained inhibitory, with a regression coefficient of −0.021. The reason for this may be that, as an old industrial base in China, the northeastern region exhibits a heavy industrial structure and relatively slow development, which is less favored by foreign direct investment. During the sample period, with the promotion of the Northeast Revitalization strategy and the implementation of a series of supporting measures, the evolution of the industrial structure also showed a continuous trend of improvement. However, due to the failure to achieve the ideal goal of upgrading traditional industries, they often do not exhibit market competitiveness, so the promotion effect on urban economic resilience is relatively limited.
The western region differs from other regions and did not show a threshold effect. We believe that the reason for this may be that the Western region implemented only the Western Development strategy during the early 21st century. Transportation, communication, and other infrastructure in the western region are relatively outdated, and the opportunities for foreign direct investment are lower than those in the eastern, central, and northeastern regions. As a result, the industrial spillover effect and positive externality effect in the western region are not significant. Therefore, even at the advanced stage of globalization, industrial structure rationalization and upgrading in the Western region do not significantly impact urban economic resilience.
In this study, 267 prefecture-level cities and above in China were adopted as the research object, and globalization was considered a threshold variable to couple industrial structure upgrading globalization and urban economic resilience. We empirically assessed the nonlinear impact of industrial structure upgrading on urban economic resilience. The results indicated that, from a national perspective, the impact of industrial structure rationalization and upgrading on urban economic resilience exhibited a nonlinear relationship, and both showed a double threshold effect. At the initial stage of globalization, industrial structure rationalization promotes urban economic resilience, while industrial structure upgrading negatively impacts urban economic resilience. After globalization crosses the first threshold and enters the intermediate stage, industrial structure rationalization and upgrading exert a positive promoting effect on urban economic resilience, and industrial structure upgrading shifts from a negative to a positive promoting effect, which is significant. When urban economic resilience crosses the second threshold and enters the advanced stage, industrial structure rationalization and upgrading exert a significant positive effect on urban economic resilience. By removing extreme values from the sample, controlling for time and individual interaction terms, including control variables, and then modeling again, it was found that the above conclusion still holds and passes the robustness test. To empirically analyze regional heterogeneity, we divided the sample into eastern, central, northeastern, and western regions. We found that industrial structure upgrading still imposes a dual threshold effect on urban economic resilience in the eastern and central regions with globalization as the threshold variable, but the performance varied among the different regions. The threshold effect of industrial structure upgrading (including rationalization and upgrading) in the eastern and northeastern regions on urban economic resilience was characterized by an inverted U-shaped trend. In particular, the northeastern region still occurred on the left side of the inverted U-shaped curve, with only a single threshold effect. No threshold effect was observed for the western region.
Based on the research findings presented here, the following policy recommendations are proposed: firstly, we should continue to accelerate industrial transformation and upgrading, thus promoting high-quality development of urban economic resilience. We should actively promote regional industrial structure upgrading and, through the development of high value-added new economy sectors, we should embark on accelerated internal economic structure optimization, thereby enhancing urban economic resilience. Consequently, to improve the rationalization of the regional industrial structure, reasonable leading industries and strategic industries should be selected for driving economic development to support the development of leading industries, extend the industrial chain of products and enhance their added value, achieve differentiated competition and gradient transfer of industries, construct a circular economic development model, optimize the layout and management of the entire industrial chain, and establish a reasonable distribution of upstream, midstream, and downstream resources in the industry. Secondly, we should increase openness and innovation efforts to promote high-quality development through high-level openness. We should make good use of the national independent innovation demonstration zone platform, highlight open innovation, enhance innovation capabilities, promote independent innovation through open innovation, and continuously improve our competitiveness in opening up to the outside world. This would allow for increased efforts to “bring in and go out,” i.e., cultivating more market entities that are open to the outside world. We should continuously optimize the export structure, encourage and support local enterprises to “go global”, attract high-quality foreign investment, and promote the “optimal entry and exit” of opening up to the outside world. Thirdly, regional governments should implement differentiated measures based on the actual situation in their respective regions to avoid one-size-fits-all measures. The level of globalization in the eastern and central regions is already high, and it is necessary to focus on accelerating the development of high-tech industries, especially the updating and development of industries such as those in information, biology, new materials, aerospace, and ocean fields. The industrial structure and globalization levels in the northeastern and western regions are relatively low. As such, it is necessary to accelerate the replacement of new and old industries, vigorously promote the integration of informatization and industrialization, and thereby improve urban economic resilience. In the western region, it is necessary to continue industrial transfer from the eastern and central regions, optimize the regional industrial structure layout, create high-quality service platforms, focus on cultivating leading enterprises, fully leverage the siphon effect to drive industrial agglomeration, and create a bridgehead for opening up to the west, leveraging the advantages of large ports, channels, logistics, hubs, and trade. In Northeast China, efforts should be made to consolidate inventories, achieve incremental expansion, extend the industrial chain, and increase the added value. The digital, networked, and intelligent transformation of traditional manufacturing industries should be accelerated, promoting the extension of the industrial chain both upstream and downstream and finally creating a relatively complete industrial chain and cluster.
Industrial structure upgrading is an effective way to improve the urban economy, while globalization is an important driving force for regulating the correlation between urban industrial structure upgrading and urban economic resilience (Cheng et al. 2022 ). Although scholars have explored the coupling relationship between industrial structure upgrading and economic resilience, there is still insufficient research on the relationship between industrial structure upgrading, globalization, and urban economic resilience using globalization as a threshold. Compared with existing research, the similarity with previous studies lies in the fact that both have validated the nonlinear relationship between industrial structure upgrading and urban economic resilience. The difference from previous research lies in that, from a research perspective, we have taken the lead in placing industrial structure upgrading, globalization, and urban economic resilience within the same framework, and systematically elaborated on the relationship between the three, Fully considering the impact of industrial structure upgrading on urban economic resilience under the regulation of globalization is a beneficial supplement to existing theoretical research. From the perspective of research content, further subdividing the upgrading of industrial structure, using globalization as a threshold variable, exploring the threshold effect of industrial structure upgrading and industrial structure rationalization on urban economic elasticity, revealing the nonlinear relationship and stage characteristics between industrial structure upgrading and urban economic elasticity, is a strong challenge to the traditional research conclusion that there is a coupling relationship between the two. From the empirical results, although it cannot be said that the framework constructed in this article solves the “black box” problem of urban economic elasticity, it can indeed indicate that there is a linear relationship between the advancement and rationalization of industrial structure and urban economic elasticity, and it has a significant dual threshold effect and regional heterogeneity.
For developing countries, globalization is a double-edged sword. On the one hand, globalization can accelerate the development of international trade and close the global wealth gap (Xuan et al. 2023 ). On the other hand, globalization can create a more stimulating environment of international market competition (Shang, 2022 ). Research has also shown that some developing countries have deviated from their actual problems and needs, are blindly integrated into the global development system, and are constrained by factors such as industrial technology, product quality, and talent, becoming the weakest link in the global industrial chain (Carlos et al. 2022 ). The invasion of Western cross-border capital and industries has undoubtedly enormously impacted the economic and social development of developing countries, leading to issues such as the fragility of urban economic resilience and limited defense capabilities. Since China’s integration into the global economic system, it has adhered to a mutually beneficial and win-win strategy of opening up to the outside world. While fully leveraging its own endowments, it is adept at innovative integration of foreign and local technologies. Based on the advanced and rational industrial structure, it has increased investment in intelligent manufacturing and digital transformation, achieving the transformation from traditional manufacturing to intelligent manufacturing in China. At the same time, it enriches the content of the urban industrial system, improves industrial production efficiency and quality, expands international market competitiveness advantages, and promotes the formation of urban economic resilience. In addition, in recent years, China has also actively promoted the the Belt and Road Initiative, aiming to optimize the industrial structure system of countries along the Belt and Road through infrastructure construction, trade and investment, financial cooperation, and other means, while increasing the economic closeness of countries along the Belt and Road, so as to promote high-quality economic development of cities along the Belt and Road. This is also the main reason why China’s economy is still able to recover and maintain rapid growth under the impact of the financial crisis in 2008 and the COVID-19 epidemic. Although China still has a long way to go in enhancing urban economic resilience, it can undoubtedly provide important experiences for other developing countries to achieve the same goals.
It should be noted that this study has certain limitations. Firstly, regarding the measurement of urban economic resilience indicators, in existing research, the core variable method and the comprehensive indicator method have mostly been used to measure urban economic resilience, but no consensus has been reached thus far. The measurement indicator method selected in this study focused more notably on the representativeness and continuity of the core variable method, while the output method was used to measure urban economic resilience. Should periodicity be considered? This question may yield a direction for future research. Secondly, there may be certain shortcomings in the measurement of indicators for industrial structure advancement and rationalization, which cannot restore the comprehensiveness of industrial structure upgrading. Can industrial structure upgrading be further classified? This issue must be improved upon in the future. Finally, there were certain missing values for the research area considered in this article (especially in the western region of China). If further consideration is given to all prefecture-level cities and above, it cannot be ruled out that this may impact the conclusions of this article, which may also provide a direction for future research.
The original data for this article is included in the supplementary material of the article, and further inquiries can be directed to the corresponding author.
It should be noted that our measurement of the technology investment level is based on the government’s perspective, confirming whether the technology investment level of the government impacts urban economic resilience. Therefore, private investment and foreign direct investment are not included in the measurement of the technology investment level.
Aida CS, Morten R, Oliver R (2016) Economic resilience: what role for policies? J Int Commer Econ Policy 7(2):1650009–1650009
Article Google Scholar
Andrew PB, Steven GT, Joana, F (2020) Disadvantage and economic viability: characterising vulnerabilities and resilience in upland farming systems. Land Use Policy, 96(C). https://doi.org/10.1016/j.landusepol.2020.104698
Anis B, Andreea CB (2023) Globalization beyond business sustainability, energy and the economy of the future. Proc Int Conf Bus Excell 17(1):1569–1583
Ben ZN, Ifergane T (2022) Firing restrictions and economic resilience: protect and survive? Rev Economic Dyn 43:93–124
Betts T, Buzzanell PM (2022) Enacting economic resilience: a synthesis of economic and communication frameworks. J Risk Financ Manag. 15(4):178–178
Boschma R (2015) The evolution of path dependence. Edited by Lars Magnusson and Jan Ottosson. Economic Geogr. 87(3):363–364
Carlos M, Antonios K, Carsten, H (2022) International trade resilience and the Covid-19 pandemic. J Bus Res. https://doi.org/10.1177/00081256211069420
Cheng T, Zhao, YH, Zhao CJ (2022) Exploring the spatio-temporal evolution of economic resilience in Chinese cities during the COVID-19 crisis. Sustain Cities Soc. https://doi.org/10.1016/J.SCS.2022.103997
Cheng, T, Zhao, YH, Song, YZ, Ma, L, Zhang, ZH, Luo, P, Gao, P, Zhang, MN, Zhao, CJ (2023) Towards resilience effectiveness: assessing its patterns and determinants to identify optimal geographic zones. J Clean Prod. https://doi.org/10.1016/J.JCLEPRO.2023.139596
Dario D, Weterings ABR (2015) The resilience of regional labour markets to economic shocks: exploring the role of interactions among firms and workers. J. Economic Geogr. 15(4):723–742
Deng Y, Jiang WY, Wang ZY (2023) Economic resilience assessment and policy interaction of coal resource oriented cities for the low carbon economy based on AI. Resour Policy. https://doi.org/10.1016/J.RESOURPOL.2023.103522
Dou QQ, Gao, XW (2023) How does the digital transformation of corporates affect green technology innovation? An empirical study from the perspective of asymmetric effects and structural breakpoints. J Clean Prod. https://doi.org/10.1016/J.JCLEPRO.2023.139245
Drobniak A (2017) Economic resilience and hybridization of development – A case of the Central European Regions. Region Stat. 7(1):43–62. https://doi.org/10.15196/rs07103
Du CY (2023) Research on evaluation of Thailand’s economic resilience and coupling coordination. Financ Eng Risk Manag. https://doi.org/10.23977/FERM.2023.060802
Du YN, Wang QX, Zhou JP (2023) How does digital inclusive finance affect economic resilience: Evidence from 285 cities in China. Int Rev Financ Anal. https://doi.org/10.1016/J.IRFA.2023.102709
Dunn HS (2020) Creative resilience and globalization from within: evolving constructs for analysing culture, innovation, and enterprise in the global south. Ann Int Commun Assoc. https://doi.org/10.1177/00081256211069420
Erika Q, Mangirdas M (2020) The economic resilience of the Austrian agriculture since the EU accession. J Risk Financ Manag 13(10):236–236
Fan WR, Wang ZX, Lv WQ (2023) Effects of the resilience of energy systems on economic sustainability. Sustain Prod Consum 41:379–390
Feng Y, Li CC, Peng DY (2023) Does regional integration improve economic resilience? Evidence from urban agglomerations in China. Sustain Cities Soc. https://doi.org/10.1016/J.SCS.2022.104273
Fingleton B, Palombi S (2013) Spatial panel data estimation, counterfactual predictions, and local economic resilience among British towns in the Victorian era. Region Sci Urban Econ 43(4):649–660
Gai M, Yang QF (2023) Synergistic study of the green efficiency and economic resilience of agriculture from a sustainable development perspective: evidence from Northeast China. Environ Sci Pollut Res Int 30(31):77568–77592
Article PubMed Google Scholar
Gajewski P (2022) Regional resilience to the Covid-19 shock in Polish regions: how is it different from resilience to the 2008 Global Financial Crisis? Region Stud Region Sci 9(1):672–684
Gan QH, Chen SM (2021) The impact of market integration on regional economic growth from the perspective of upgrading industrial structure. Ind Econ Res. https://doi.org/10.13269/j.cnki.ier.2021.05.004
Gereffi G, Pananond P, Pedersen T (2022) Resilience decoded: the role of firms, global value chains, and the state in COVID-19 medical supplies. Calif Manag Rev 64(2):46–70
Guo AJ, Zhang XY, Li Q (2023) Network infrastructure construction and urban economic resilience: a quasi natural experiment based on the demonstration city construction of “Broadband China”. Financ Trad Res. https://doi.org/10.19337/j.cnki.34-1093/f.2023.05.003
Hao YY (2023) The dynamic relationship between trade openness, foreign direct investment, capital formation, and industrial economic growth in China: new evidence from ARDL bounds testing approach. Humanit Soc Sci Commun. https://doi.org/10.1057/S41599-023-01660-8
Hui J, Tan QM (2023) Dynamic evaluation of regional economic resilience under major public emergencies: based on an improved dynamic evaluation model of grey incidence projection-fuzzy matter element. Wirel Netw 29(7):3223–3238
Hynes W, Trump BD, Kirman A, Haldane A, Linkov I (2022) Systemic resilience in economics. Nat Phys 18(4):381–384
Article CAS Google Scholar
Jayasinghe N, Fernando S, Haigh R, Amaratunga D, Fernando N, Vithanage C, Ratnayake, RC (2022) Economic resilience in an era of ‘systemic risk’: Insights from four key economic sectors in Sri Lanka. Prog. Disaster Sci. https://doi.org/10.1016/J.PDISAS.2022.100231
Jesse S (2023) Arcidiacono Alessia,Torrisi Gianpiero,Arku Robert Nutifafa. Regional economic resilience: a scoping review. Prog Hum Geogr 47(4):500–532
Johnson LC, Mundell M (2023) Regional resilience and an interventionist state: the case of Geelong, Victoria, 1990–2020. Urban Policy Res 41(3):247–261
Knoke T, Paul C, Gosling E, Jarisch I, Mohr J, Seidl R (2022) Assessing the economic resilience of different management systems to severe forest disturbance. Environ Resour Econ 84(2):343–381
Lee SM, Wang S (2023) Impacts of political fragmentation on inclusive economic resilience: examining American metropolitan areas after the Great Recession. Urban Stud 60(1):26–45
Lemke LK, whang G, Sakdapolrak P, Trippl M (2023) Unresolved issues in regional economic resilience: Conceptual ways forward. Prog Hum Geogr 47(5):699–717
Li XM, Zhou SW, Zhao YF, Wan GX (2024) A CRITIC-TOPSIS and optimized nonlinear grey prediction model: a comparative convergence analysis of marine economic resilience. Expert Syst Appl. https://doi.org/10.1016/J.ESWA.2023.121356
Li J (2024) Impact of green finance on industrial structure upgrading: implications for environmental sustainability in Chinese regions. Environ Sci Pollut Res Int 31(9):13063–13074
Maria T (2023) Social capital and characteristics of economic dualism as determinants of regional resilience. Eur Plan Stud 31(12):2568–2589
Mai X, Zhan CQ, Chan Roger CK (2021) The nexus between (re)production of space and economic resilience: an analysis of Chinese cities. Habitat Int. https://doi.org/10.1016/J.HABITATINT.2021.102326
Martin R, Sunley P (2014) On the notion of regional economic resilience: conceptualization and explanation. J Econ Geogr 15(1):1–42
Martin R, Sunley P, Gardiner B (2016) How regions react to recessions:Resilience and the role of economic structure. Region Stud 50(4):561–585
Ngo HT (2023) Causal relationship between globalization, economic growth and CO2 emissions in Vietnam using wavelet analysis. Energy Environ 34(7):2386–2412
Paolo DC (2017) Testing and explaining economic resilience with an application to. Pap Region Sci 96(1):93–113
Papaioannou SK (2023) ICT and economic resilience: evidence from the COVID-19 pandemic. Econ Model. https://doi.org/10.1016/J.ECONMOD.2023.106500
Pashapour SA, Azadeh A, Ghaderi SF, Keramati A (2019) Performance optimization of organizations considering economic resilience factors under uncertainty: a case study of a petrochemical plant. J Clean Prod. https://doi.org/10.1016/j.jclepro.2019.05.171
Qiang Y, Huang QX, Xu JW (2020) Observing community resilience from space: Using nighttime lights to model economic disturbance and recovery pattern in natural disaster. Sustain Cities Soc. https://doi.org/10.1016/j.scs.2020.102115
Roberta C, Andrea C, Ugo F (2015) Spatial heterogeneity in the costs of the economic crisis in Europe: are cities sources of regional resilience? J Econ Geogr 15(5):951–972
Shang ZC (2022) China’s ecological civilization and economic development model under globalization. Financ Eng Risk Manag. https://doi.org/10.23977/FERM.2022.050606
Tan JT, Zhang PY, Kevin L, Li J, Liu SW (2017) Conceptualizing and measuring economic resilience of resource-based cities: case study of Northeast China. Chin Geogr Sci 27(3):471–481
Tan JT, Zhao HB, Liu WX, Zhang PY, Qiu FD (2020) Regional economic resilience and influential mechanism during economic crises in China. Sci Geogr Sin 40(2):173–181
Google Scholar
Tao MM, Stephen P,Wen L, Sheng MY, Wang JD, Wang GH, Zheng YH (2023) Appraising the role of the digital economy in global decarbonization: a spatial non-linear perspective on globalization. J Environ Manag 347:119170–119170
Wang JJ, Ma CC, Jiang T, Phillips A, Wu X, Wang Y, Wang ZM, Cao Y (2024) A spatial econometric investigation into road traffic accessibility and economic growth: insights from the Chengdu-Chongqing twin-city economic circle. Humanit Soc Sci Commun. https://doi.org/10.1057/S41599-024-02695-1
Wang H, Li Y, Liu YZ, Qing FL, Zhou YD, Chen YM, Fang DD (2021) Study on the Influencing factors of urban economic resilience in post epidemic Era—A case study of Kunming City. J Urban Manag. https://doi.org/10.1016/J.JUM.2021.06.006
Wang HP, Ge Q (2023) Spatial association network of economic resilience and its influencing factors: evidence from 31 Chinese provinces. Humanit Soc Sci Commun 10(1):290–290
Article MathSciNet PubMed PubMed Central Google Scholar
Wang KW, Ma HT, Fang CL (2023) The relationship evolution between urbanization and urban ecological resilience in the Northern Slope Economic Belt of Tianshan Mountains, China. Sustain Cities Soc. https://doi.org/10.1016/J.SCS.2023.104783
Wang SJ, Wang ZH (2021) The spatio-temporal differentiation of urban land expansion and influencing factors in the Yangtze River Delta in the economic transition context. J Nat Resour 36(4):993–1007
Wang YW, Xu JG, Liu D, Zhou YY (2023) Analysis and comparison of the industrial economic resilience in the Taihu Lake Basin under the 2008 financial crisis and the 2018 Sino-US Trade War. Land 12(2):481–481
Xiong M, Cheng SM, Guo H, Zhao JZ (2023) The impact of local government fiscal gaps on public-private partnerships: government demand and private sector risk aversion. Int Public Manag J 26(4):589–608
Xuan D, Jian XY, Fang Y (2023) Can globalization and the green economy hedge natural resources? Functions of population growth and financial development in BRICS countries. Resour Policy. https://doi.org/10.1016/J.RESOURPOL.2023.103414
Yang YY, Chen WK, Gu RD (2023) How does digital infrastructure affect industrial eco-efficiency? Considering the threshold effect of regional collaborative innovation. J Clean Prod. https://doi.org/10.1016/J.JCLEPRO.2023.139248
Yang ZS (2023) Human capital space: a spatial perspective of the dynamics of people and economic relationships. Humanit Soc Sci Commun 10(1):145–145
Article PubMed PubMed Central Google Scholar
Ye SS, Qian Z (2021) The economic network resilience of the Guanzhong plain city cluster, China: a network analysis from the evolutionary perspective. Growth Change 52(4):2391–2411
Yin Y, Su XY, Xie XQ (2023) The spatial effect of industrial structure upgrading on urban economic resilience. Econ Geogr. 43(8):86–92
Yu HJ, Wang YZ, Li DH (2023) Using the AHP-entropy weight method to study regional marine economy resilience in China. J Coast Res 39(5):960–969
Ženka J, Slach O, Pavlík A (2019) Economic resilience of metropolitan, old industrial, and rural regions in two subsequent recessionary shocks. Eur Plan Stud 27(11):2288–2311
Zhang J (2022) Spatial effects of tourism development on economic resilience: an empirical study of Wenchuan earthquake based on dynamic spatial Durbin model. Nat Hazards 115(1):309–329
Zhang LX, Bi X, Huang ZJ (2020) Urban industrial land use efficiency under the background of economic transformation in the Yangtze River Economic Belt. Resour Sci 42(9):1728–1738
Zhang L, Yao L (2023) A study on the impact of digital technology innovation on urban economic resilience: empirical evidence from 278 prefectural and above cities in China. J Xinxiang Educ Coll 36(05):38–59
Zhang XB, Fu WS, Ma HT (2023) Coupling correlation between economic resilience and industrial structure evolution in the Beijing-Tianjin-Hebei region. Acta Geogr Sin. https://doi.org/10.11821/dlxb202310002
Zheng ZY, Xu YB, Chen JY (2023) Digital economy, industrial structure upgrading and green innovation efficiency of family enterprises. Int Entrep Manag J 20(1):479–503
Zheng HZ, Qian J, Liu GH,Wu YG, Delang, CO, He HM (2023) Housing prices and household consumption: a threshold effect model analysis in central and western China. Humanit Soc Sci Commun. https://doi.org/10.1057/S41599-023-02258-W
Zhou D, Chen A, Wang J (2019) Impact of disaster risks on regional economic resilience in China: a case study of Wenchuan earthquake. Soc. Sci. 8(5):245–254
Zhou Q, Qi ZP (2023) Urban economic resilience and human capital: an exploration of heterogeneity and mechanism in the context of spatial population mobility. Sustain Cities Soc. https://doi.org/10.1016/J.SCS.2023.104983
Download references
This work was supported by the National Natural Science Foundation of China [42271284;41801205;71974071;42171286].Fundamental Scientific Research Business Expense for Higher School of Central Government[CCNU23ZZ009].
Authors and affiliations.
Faculty of Political Science, School of Public Administration, Central China Normal University, Wuhan, China
Lu Zhang & Guodong Lin
School of Humanities and Law, Northeastern University, Shenyang, China
School of Management, Guangzhou College of Technology and Business, Guangzhou, China
You can also search for this author in PubMed Google Scholar
Conceptualization: LZ; methodology: WS; software: XL; validation, formal analysis, investigation, data curation, writing-original draft preparation, writing—review and editing, and visualization: GL; supervision: LZ; All authors have read and agreed to the published version of the manuscript.
Correspondence to Guodong Lin .
Competing interests.
The authors declare no competing interests.
This research was not required to receive any ethical approval because it did not involve human research participants and no primary data were collected. It uses data collected by statistical organizations such as the Statistical Yearbook of Urban Construction in China and the Statistical Year book of Chinese Cities from 2008 to 2021.
This article does not contain any studies with human participants performed by any of the authors.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .
Reprints and permissions
Cite this article.
Zhang, L., Lin, G., Lyu, X. et al. Suppression or promotion: research on the impact of industrial structure upgrading on urban economic resilience. Humanit Soc Sci Commun 11 , 843 (2024). https://doi.org/10.1057/s41599-024-03329-2
Download citation
Received : 12 December 2023
Accepted : 11 June 2024
Published : 27 June 2024
DOI : https://doi.org/10.1057/s41599-024-03329-2
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
The debate has raged for decades: Was it humans or climate change that led to the extinction of many species of large mammals, birds, and reptiles that have disappeared from Earth over the past 50,000 years?
By "large," we mean animals that weighed at least 45 kilograms -- known as megafauna. At least 161 species of mammals were driven to extinction during this period. This number is based on the remains found so far.
The largest of them were hit the hardest -- land-dwelling herbivores weighing over a ton, the megaherbivores. Fifty thousand years ago, there were 57 species of megaherbivores. Today, only 11 remain. These remaining 11 species have also seen drastic declines in their populations, but not to the point of complete extinction.
A research group from the Danish National Research Foundation's Center for Ecological Dynamics in a Novel Biosphere (ECONOVO) at Aarhus University now concludes that many of these vanished species were hunted to extinction by humans.
Many different fields of research
They present this conclusion in a review article invited by and published in the scientific journal Cambridge Prisms: Extinction . A review article synthesizes and analyses existing research within a particular field.
In this case, the researchers from Aarhus University incorporated several research fields, including studies directly related to the extinction of large animals, such as:
- The timing of species extinctions
- The animals' dietary preferences
- Climate and habitat requirements
- Genetic estimates of past population sizes
- Evidence of human hunting
Additionally, they included a wide range of studies from other fields necessary to understand the phenomenon, such as:
- Climate history over the past 1-3 million years
- Vegetation history over the past 1-3 million years
- Evolution and dynamics of fauna over the past 66 million years
- Archaeological data on human expansion and lifestyle, including dietary preferences
Climate change played a lesser role
The dramatic climate changes during the last interglacial and glacial periods (known as the late Pleistocene, from 130,000 to 11,000 years ago) certainly affected populations and distributions of both large and small animals and plants worldwide. However, significant extinctions were observed only among the large animals, particularly the largest ones.
An important observation is that the previous, equally dramatic ice ages and interglacials over the past couple of million years did not cause a selective loss of megafauna. Especially at the beginning of the glacial periods, the new cold and dry conditions caused large-scale extinctions in some regions, such as trees in Europe. However, there were no selective extinctions of large animals.
"The large and very selective loss of megafauna over the last 50,000 years is unique over the past 66 million years. Previous periods of climate change did not lead to large, selective extinctions, which argues against a major role for climate in the megafauna extinctions," says Professor Jens-Christian Svenning. He leads ECONOVO and is the lead author of the article. He adds, "Another significant pattern that argues against a role for climate is that the recent megafauna extinctions hit just as hard in climatically stable areas as in unstable areas."
Effective hunters and vulnerable giants
Archaeologists have found traps designed for very large animals, and isotope analyses of ancient human bones and protein residues from spear points show that they hunted and ate the largest mammals.
Jens-Christian Svenning adds, "Early modern humans were effective hunters of even the largest animal species and clearly had the ability to reduce the populations of large animals. These large animals were and are particularly vulnerable to overexploitation because they have long gestation periods, produce very few offspring at a time, and take many years to reach sexual maturity."
The analysis shows that human hunting of large animals such as mammoths, mastodons, and giant sloths was widespread and consistent across the world.
It also shows that the species went extinct at very different times and at different rates around the world. In some local areas, it happened quite quickly, while in other places it took over 10,000 years. But everywhere, it occurred after modern humans arrived, or in Africa's case, after cultural advancements among humans.
…in all types of environments
Species went extinct on all continents except Antarctica and in all types of ecosystems, from tropical forests and savannas to Mediterranean and temperate forests and steppes to arctic ecosystems.
"Many of the extinct species could thrive in various types of environments. Therefore, their extinction cannot be explained by climate changes causing the disappearance of a specific ecosystem type, such as the mammoth steppe -- which also housed only a few megafauna species," explains Jens-Christian Svenning. "Most of the species existed under temperate to tropical conditions and should actually have benefited from the warming at the end of the last ice age."
Consequences and recommendations
The researchers point out that the loss of megafauna has had profound ecological consequences. Large animals play a central role in ecosystems by influencing vegetation structure (e.g., the balance between dense forests and open areas), seed dispersal, and nutrient cycling. Their disappearance has resulted in significant changes in ecosystem structures and functions.
"Our results highlight the need for active conservation and restoration efforts. By reintroducing large mammals, we can help restore ecological balances and support biodiversity, which evolved in ecosystems rich in megafauna," says Jens-Christian Svenning.
FACTS: The numbers of extinct and surviving species come from the freely accessible database PHYLACINE 1.2.1, which lists all known mammals that have lived in the past 129,000 years, including those that have gone extinct recently or are only found in captivity.
Story Source:
Materials provided by Aarhus University . Original written by Peter Gammelby. Note: Content may be edited for style and length.
Related Multimedia :
Journal Reference :
Cite This Page :
Strange & offbeat.
IMAGES
VIDEO
COMMENTS
Table of contents. Step 1: Restate the problem. Step 2: Sum up the paper. Step 3: Discuss the implications. Research paper conclusion examples. Frequently asked questions about research paper conclusions.
A conclusion in a research paper is the final section where you summarize and wrap up your research, presenting the key findings and insights derived from your study. The research paper conclusion is not the place to introduce new information or data that was not discussed in the main body of the paper. When working on how to conclude a ...
Research Paper Conclusion. Definition: A research paper conclusion is the final section of a research paper that summarizes the key findings, significance, and implications of the research. It is the writer's opportunity to synthesize the information presented in the paper, draw conclusions, and make recommendations for future research or ...
The conclusion is intended to help the reader understand why your research should matter to them after they have finished reading the paper. A conclusion is not merely a summary of the main topics covered or a re-statement of your research problem, but a synthesis of key points derived from the findings of your study and, if applicable based on your analysis, explain new areas for future research.
The conclusion is where you describe the consequences of your arguments by justifying to your readers why your arguments matter (Hamilton College, 2014). Derntl (2014) also describes conclusion as the counterpart of the introduction. Using the Hourglass Model (Swales, 1993) as a visual reference, Derntl describes conclusion as the part of the ...
Step 1: Restate the problem. Always begin by restating the research problem in the conclusion of a research paper. This serves to remind the reader of your hypothesis and refresh them on the main point of the paper. When restating the problem, take care to avoid using exactly the same words you employed earlier in the paper.
Begin your conclusion by restating your thesis statement in a way that is slightly different from the wording used in the introduction. Avoid presenting new information or evidence in your conclusion. Just summarize the main points and arguments of your essay and keep this part as concise as possible. Remember that you've already covered the ...
Step 1: Answer your research question. Step 2: Summarize and reflect on your research. Step 3: Make future recommendations. Step 4: Emphasize your contributions to your field. Step 5: Wrap up your thesis or dissertation. Full conclusion example. Conclusion checklist. Other interesting articles.
The conclusion pushes beyond the boundaries of the prompt and allows you to consider broader issues, make new connections, and elaborate on the significance of your findings. Your conclusion should make your readers glad they read your paper. Your conclusion gives your reader something to take away that will help them see things differently or ...
Begin with a clear statement of the principal findings. This will reinforce the main take-away for the reader and set up the rest of the discussion. Explain why the outcomes of your study are important to the reader. Discuss the implications of your findings realistically based on previous literature, highlighting both the strengths and ...
Conclusions. One of the most common questions we receive at the Writing Center is "what am I supposed to do in my conclusion?". This is a difficult question to answer because there's no one right answer to what belongs in a conclusion. How you conclude your paper will depend on where you started—and where you traveled.
1. New Data: In a research paper conclusion, avoid presenting new data or evidence that wasn't discussed earlier in the paper. It's the time to summarize, analyze, or explain the significance of data already provided, not to introduce new material. 2. Irrelevant Details: The conclusion is not the spot for extraneous details not directly ...
Phrases for Conclusions of Research Papers. All this requires us to (propose the next action or an alternative idea). Altogether, these findings indicate (point out the logical result). Finally, it is important to note (make your strongest point and follow with a recommendation). In conclusion (restate your thesis with greater emphasis).
Writing a Conclusion. A conclusion is an important part of the paper; it provides closure for the reader while reminding the reader of the contents and importance of the paper. It accomplishes this by stepping back from the specifics in order to view the bigger picture of the document. In other words, it is reminding the reader of the main ...
When writing a research paper, it can be challenging to make your point after providing an extensive amount of information. For this reason, a well-organized conclusion is essential. A research paper's conclusion should be a brief summary of the paper's substance and objectives; what you present in your research paper can gain impact by having a strong conclusion section.
A conclusion is the final paragraph of a research paper and serves to help the reader understand why your research should matter to them. The conclusion of a conclusion should: Restate your topic and why it is important. Restate your thesis/claim. Address opposing viewpoints and explain why readers should align with your position.
The point of a conclusion to a research paper is to summarize your argument for the reader and, perhaps, to call the reader to action if needed. 5. Make a call to action when appropriate. If and when needed, you can state to your readers that there is a need for further research on your paper's topic.
Key Takeaways. Because research generates further research, the conclusions you draw from your research are important. To test the validity of your conclusions, you will have to review both the content of your paper and the way in which you arrived at the content. Mailing Address: 3501 University Blvd. East, Adelphi, MD 20783.
The conclusion of a research paper has several key objectives. It should: Restate your research problem addressed in the introduction section. Summarize your main arguments, important findings, and broader implications. Synthesize key takeaways from your study. The specific content in the conclusion depends on whether your paper presents the ...
Open With The Research Topic. To begin a conclusion paragraph, use the first sentence to reiterate the comprehensive subject matter that your paper covered. Since this is just a sentence-long retelling of your research topic and why it's important, it doesn't have to be specific, but it does need clarity. Example.
The conclusion of a research paper should contain a summary of the main findings or results of the study, a restatement of the thesis statement or research question, and a discussion of the broader implications or significance of the findings. It should also reflect on the limitations of the study and suggest directions for future research.
A research paper that presents a sustained argument will usually encapsulate this argument in a thesis statement. A research paper designed to present the results of empirical research tends to present a research question that it seeks to answer. It may also include a hypothesis—a prediction that will be confirmed or disproved by your research.
Writing a strong conclusion for your research paper is crucial, and WPS Office is designed to be exceptionally student-friendly. It offers accessible options and advanced features for free, making it an excellent tool for students. One of the standout features is WPS AI, which integrates AI into its writing and proofreading abilities. ...
Background Studies of implementation strategies range in rigor, design, and evaluated outcomes, presenting interpretation challenges for practitioners and researchers. This systematic review aimed to describe the body of research evidence testing implementation strategies across diverse settings and domains, using the Expert Recommendations for Implementing Change (ERIC) taxonomy to classify ...
With the benefit of hindsight, one can now draw many conclusions about the US policy responses to the two biggest economic crises of the past decade and a half: the 2008 financial crisis and its aftermath, and the COVID-19 pandemic. ... DC, the Hoover Institution is the nation's preeminent research center dedicated to generating policy ideas ...
Through this, we hope to address this significant gap and facilitate youth pre-DM/DM research. Methods: Building on data from the National Health and Nutrition Examination Survey (NHANES) from 1999 to 2018, we cleaned and harmonized hundreds of variables relevant to pre-DM/DM (fasting plasma glucose level ≥100 mg/dL or glycated hemoglobin ≥ ...
Objective: Group therapy is the most common form of treatment for substance use disorders (SUDs). However, a research-practice gap has limited the implementation of group evidence-based treatments. This article aims to synthesize the best available evidence of the effectiveness of group psychotherapy for SUDs. Method: Building upon previous reviews and meta-analyses, we conducted a brief ...
Background: Although several biomarkers exist for patients with heart failure (HF), their use in routine clinical practice is often constrained by high costs and limited availability. Objective: We examined the utility of an artificial intelligence (AI) algorithm that analyzes printed electrocardiograms (ECGs) for outcome prediction in patients with acute HF.
Industrial The upgrading of industrial structure, as the main means of urban economic transformation, plays a crucial role in the process of achieving urban economic resilience construction. We ...
Many different fields of research. They present this conclusion in a review article invited by and published in the scientific journal Cambridge Prisms: Extinction. A review article synthesizes ...