Machine learning-based prediction of return to work within 12 months in acute myocardial infarction patients

doi:10.21203/rs.3.rs-4619695/v1

Download PDF

Article

Machine learning-based prediction of return to work within 12 months in acute myocardial infarction patients

https://doi.org/10.21203/rs.3.rs-4619695/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Return to work is a critical indicator of recovery after acute myocardial infarction (AMI), and accurate identification of patients with low return-to-work rates is critical for timely intervention. The aim of this study was to develop a machine learning (ML) model for predicting the return to work in AMI patients. A retrospective study of data from 1473 patients was conducted using the Incidence Rate of Heart Failure After Acute Myocardial Infarction With Optimal Treatment database. Patients were randomly divided into a training cohort and a validation cohort (7:3). A total of five ML models were developed based on the training cohort to predict return to work. Model performance was assessed according to numerous learning metrics, including the area under the receiver operating characteristic curve (AUC), accuracy, F1-score, and Brier score. The ML models were constructed using 12 features, including age, occupation, income, anterior wall AMI, hypertension, fasting plasma glucose (FPG), beta-blockers, marriage, aspartate transaminase (AST), body mass index (BMI), TG (triglyceride) and phase II cardiac rehabilitation (CR). Among the five ML models, the LR model achieved the best performance, with an AUC of 0.793 (95% CI, 0.712-0.874), an accuracy of 0.719 (95% CI, 0.642-0.787), an F1 score of 0.800, and a Brier score of 0.135, and was subsequently transformed into a nomogram. A new return-to-work prediction model was developed based on a machine learning algorithm, which may help identify patients with low return-to-work rates and may become an effective management tool for AMI patients.

Clinical trial registration: Clinical Trials.gov ID: NCT03297164.

Health sciences/Cardiology

Health sciences/Health care/Prognosis

Acute myocardial infarction

Return to work

Cardiac rehabilitation

Machine learning

Prognosis

Acute myocardial infarction (AMI) is a significant health issue and has become a major disease burden worldwide^[¹^].Return to work is an essential part of the recovery process after AMI, signifying the transition from recovering from illness to re-entering society, and is associated with better physical functional abilities^[2], mental health^[3], and a greater quality of life^[^{4, 5}^]. Failure to return to work may increase the incidence of depression,^[6]increase economic burdens on patients and their families^[7], and cause labor shortages and socioeconomic losses^[8].AMI patients in China are younger than those in many other countries^[9], and the need to return to work becomes more urgent. Therefore, identifying the key factors affecting the return to work of AMI patients and establishing a predictive model are particularly important.

Machine learning (ML), which is a subset of AI wherein machines autonomously acquire information by extracting patterns from databases, has been increasingly used within the medical community, specifically within the domain of cardiovascular diseases. With the continuous development of machine learning technology^[10], it has surpassed existing clinical decision-making and traditional analytical techniques in the diagnosis and prognosis prediction of cardiovascular disease patients^[^{11, 1}^2]. The advantage of machine learning is its potential to offer innovative solutions for clinical problems, identify key factors that may be overlooked, and provide guidance for clinical decision-making. The accurate identification of patients with a low rate of return to work through predictive models and the provision of early and precise interventions may improve the rate of return to work.

Therefore, this study used multiple machine learning methods to establish an effective model for predicting the return-to-work rate within 12 months in AMI patients who received phase I cardiac rehabilitation during hospitalization, providing technical support for clinical judgment and intervention decisions.

General characteristics. From January 2018 to September 2019, a total of 1473 patients with AMI who received phase I cardiac rehabilitation during hospitalization were enrolled in this study. A total of 539 patients remained eligible for analysis. The process of patient recruitment is provided in Supplementary Fig. 1. After random allocation, 379 (70%) patients were assigned to the training cohort, while 160 (30%) patients composed the validation cohort. Among the 539 patients, the median age was 50.0 (45.0–54.0) years, 34 were females (6.3%), 98 had a higher education level (18.2%), and 516 were married (95.7%). Among the patients followed up for twelve months, 431 (80.0%) returned to work, and 344 (63.8%) received phase II cardiac rehabilitation. Table 1 presents the baseline characteristics used for model construction, with the complete baseline characteristics provided in Supplementary Table S1. Apart from income, variables selected for modeling showed no significant differences between the training and validation cohorts.

Table 1

Comparison of Baseline Selected Characteristics and Rework Outcome Between the Training and Validation Cohorts
Characteristic	Overall, n = 539	Training, n = 379	Validation, n = 160	p-value
Rework, n (%)	431 (80.0%)	303 (79.9%)	128 (80.0%)	> 0.999
Age	50.00 (45.00, 54.00)	50.00 (45.00, 54.00)	50.00 (45.00, 54.00)	0.846
Married, n (%)	516 (95.7%)	365 (96.3%)	151 (94.4%)	0.352
BMI, kg/m²	26.00 (23.90, 28.40)	26.00 (23.90, 28.40)	26.00 (23.98, 28.40)	0.904
Occupation, n (%)				0.470
Blue-collar workers	270 (50.0%)	194 (51.2%)	76 (47.5%)
White-collar workers	118 (21.9%)	76 (20.1%)	42 (26.3%)
Self-employed	54 (10.0%)	39 (10.3%)	15 (9.4%)
Other	97 (18.0%)	70 (18.5%)	27 (16.9%)
Income, n (%)				0.001
< 2500	262 (48.6%)	199 (52.5%)	63 (39.4%)
2500–5000	184 (34.1%)	110 (29.0%)	74 (46.3%)
5000 − 1000	74 (13.7%)	54 (14.2%)	20 (12.5%)
> 10000	19 (3.5%)	16 (4.2%)	3 (1.9%)
Hypertension, n (%)	231 (42.9%)	161 (42.5%)	70 (43.8%)	0.860
Anterior wall AMI, n (%)	197 (36.5%)	139 (36.7%)	58 (36.3%)	> 0.999
Beta-blocker, n (%)	13 (2.4%)	10 (2.6%)	3 (1.9%)	0.825
FPG, mmol/L	7.61 (6.42, 10.22)	7.49 (6.41, 9.86)	7.73 (6.44, 11.02)	0.318
TG, mmol/L	1.52 (1.04, 2.24)	1.54 (1.02, 2.24)	1.49 (1.08, 2.17)	0.886
AST, U/L	35.00 (22.00, 81.00)	36.00 (22.00, 84.50)	34.50 (22.00, 72.50)	0.605
CR, n (%)	344 (63.8%)	239 (63.1%)	105 (65.6%)	0.640
FPG, Fasting plasma glucose; CR, Phase II cardiac rehabilitation; AST, Aspartate transaminase; TG, Triglyceride; BMI, Body mass index; Beta-blocker, Medication history of Beta-blocker.

Feature selection. Through the Lasso for feature selection, a Spearman correlation test was conducted on the factors identified by the best lambda value, as shown in Supplementary Figs. 2 and 3. Finally, 12 features, including age, occupation, income, anterior wall AMI, hypertension, fasting plasma glucose (FPG), beta-blocker, married, aspartate transaminase (AST), body mass index (BMI), TG (triglyceride) and phase II cardiac rehabilitation (CR), were identified to construct ML models (Supplementary Fig. 4).

Model performance. These factors were then combined with the LR, RF, XGBoost, SVM, and ANN algorithms, with the algorithm demonstrating the best performance being selected. Each model was optimized by 5 repeats of 10-fold cross-validation or tuned to the best parameters. Table 2 summarizes the performance of each model. In the validation cohorts, the accuracies of LR, XGBoost, RF, ANN, and SVM were 0.719, 0.831, 0.750, 0.669, and 0.763, respectively; the sensitivities were 0.703, 0.875, 0.766, 0.625, and 0.766, respectively; the F1-scores were 0.800, 0.893, 0.831, 0.751, and 0.838, respectively; and the ROC-AUC values were 0.793, 0.791, 0.793, 0.783, and 0.773, respectively. The AUCs for the five models ranged from 0.773 to 0.793, and the receiver operating characteristic (ROC) curves are shown in Fig. 1 (A, B). The LR model performed best, achieving an AUC of 0.793 (95% CI, 0.712–0.874), with 0.719 (95% CI, 0.642–0.787) accuracy, 0.703 sensitivity, 0.781 specificity, 0.928 PPV, and 0.800 F1 score on the validation cohort. Additionally, in the validation part, the XGBoost model achieved the highest accuracy, sensitivity, and F1 score. According to further model evaluation, the Brier score for the ability of the LR model to predict RTW was 0.135 in the validation cohort. Five repeats of Ten-fold cross-validation was also performed on the LR model, and its average accuracy were 0.813, indicating that our model was reliable.

Table 2

Model performance in predicting return to work in the training and validation cohorts.
ML Models	AUC	Accuracy	Sensitivity	Specificity	PPV	NPV	F1	Brier
Training cohort
Logistic regression	0.784 (0.726–0.841)	0.826 (0.784–0.863)	0.908	0.500	0.879	0.576	0.893	0.124
XGBoost	0.814 (0.755–0.872)	0.797 (0.753–0.836)	0.819	0.711	0.919	0.495	0.866	0.116
Random Forest	0.807 (0.754–0.860)	0.678 (0.629–0.725)	0.640	0.823	0.937	0.366	0.761	0.160
ANN	0.836 (0.780–0.892)	0.807 (0.764–0.846)	0.822	0.750	0.929	0.514	0.872	0.111
SVM	0.815 (0.758–0.873)	0.847 (0.807–0.881)	0.911	0.592	0.899	0.625	0.905	0.159
Validation cohort
Logistic regression	0.793 (0.712–0.874)	0.719 (0.642–0.787)	0.703	0.781	0.928	0.397	0.800	0.135
XGBoost	0.791 (0.701–0.881)	0.831 (0.764–0.886)	0.875	0.656	0.911	0.568	0.893	0.130
Random Forest	0.793 (0.711–0.875)	0.750 (0.676–0.815)	0.766	0.688	0.907	0.423	0.831	0.158
ANN	0.783 (0.696–0.869)	0.669 (0.590–0.741)	0.625	0.844	0.941	0.360	0.751	0.135
SVM	0.773 (0.684–0.861)	0.763 (0.689–0.826)	0.766	0.750	0.925	0.444	0.838	0.160
ML, Machine learning; AUC, Area under the curve; XGBoost, Extreme gradient boosting; ANN,Artificial neural network; SVM, Support vector machine.

To determine the major predictors of return to work in our cohort, the importance of each permutation feature was measured from the LR, RF, XGBoost, ANN and SVM models. In the XGBoost model, Supplementary Fig. 5 shows that the 5 most important variables were occupation, age, income, TG, and FPG. In the LR model, Supplementary Table S2 shows that the 5 most important variables were occupation (coef = -1.785, OR = 0.168), marital status (coef = 1.484, OR = 4.409), beta-blocker history (coef = -1.134, OR = 0.322), phase II cardiac rehabilitation (coef = 0.667, OR = 1.948), and income (coef = 0.637, OR = 1.890). In the RF model, Supplementary Fig. 6 shows that the 5 most important variables were occupation, age, AST, FPG, and income. In the ANN model, Supplementary Fig. 7 shows that the 5 most important variables were occupation, beta-blockers, anterior wall AMI, income, and marital status.

Nomogram. Since the LR model performed best, with an AUC of 0.793, it was transformed into a dynamic nomogram for easier clinical application. From the 12 candidate factors identified after LASSO feature selection, a stepwise approach based on the Akaike information criterion (AIC) was used to select the optimal model (Supplementary Table S3), which eventually included 6 factors (age, occupation, anterior wall acute myocardial infarction, TG, AST, and phase II cardiac rehabilitation) that formed the dynamic nomogram (Fig. 2A). Points were assigned to each predictor by drawing a vertical line to the “points” axis, and then the total points were calculated as the sum of them. The dynamic nomogram with an intuitive web-based interface was also developed so as to facilitate the use for clinicians in clinical practices (Fig. 2B) (Dynamic Nomogram: https://returntowork.shinyapps.io/AMIRTWApp/)

To our knowledge, this is the first attempt to establish a machine learning-based model for predicting return to work in patients with AMI. In this study, we utilized data from 539 patients who underwent phase I cardiac rehabilitation during hospitalization for acute myocardial infarction to develop and internally validate five machine learning models for predicting the rate of return to work. These machine learning models demonstrated good predictive ability (AUC range: 0.773–0.793), potentially aiding in the management of AMI patients' return to work. Among them, the LR model showed the best predictive performance, with good discrimination and calibration effects.

Among the five machine learning models, the LR model demonstrated the best performance, achieving an AUC of 0.793, an accuracy of 0.719, and an F1 score of 0.800. The RF and XGBoost models also performed well, with AUCs of 0.793 and 0.791, respectively. When evaluating predictive models, the AUC is a commonly used metric for accuracy, but the sensitivity and PPV may have more clinical significance^[15]. In our study, the sensitivity range for the five ML models in the validation cohort was 0.625 to 0.875, and the PPV range was 0.907 to 0.941, indicating that they have a good ability to identify patients who return to work. In our study, the NPV of the models was lower (0.360–0.568), possibly because these models are based on objective indicators, but the return to work may be influenced by the patient's subjective willingness. However, the low NPV in our predictive models does not affect their clinical application, as we are more inclined to identify patients with a low return-to-work rate and enable them to achieve a positive outcome (return to work) through early intervention. Although there were some differences in the selected features, the results from the LR, RF and XGBoost models indicate that factors such as occupation, marital status, beta-blocker status, phase II cardiac rehabilitation status, income, age, AST, FPG, hypertension, and TG have significant clinical value in predicting the return to work in AMI patients.

Consistent with previous research, occupation and income are often combined as socioeconomic status,^[16] serving as a known predictive factor for the prognosis of cardiovascular diseases and related to returning to work^[17]. Participation in cardiac rehabilitation can enhance patients' return-to-work rates^{[18, 19]}, improve functional capacity^[20], mental health^[21], health-related quality of life^[22], adherence to treatment^[23, control of cardiovascular disease risk factors, and reduce readmission rates and cardiovascular mortality^[24]. In this study, all included patients underwent phase I cardiac rehabilitation treatment, which may be the reason for the higher return-to-work rate in this study than in previous studies. Additionally, the clinical value of phase II cardiac rehabilitation in predicting return to work was demonstrated across various machine learning analyses^[14]. These results highlight the importance of comprehensive cardiac rehabilitation. Although the benefits of CR have been well documented, participation rates among eligible patients remain low^[25], and the widespread use and implementation of cardiac rehabilitation are of significant clinical importance.

Hypertension, TG levels, and FPG levels, which are cardiovascular risk factors, are closely associated with cardiovascular health and prognosis^{[26, 27]}. The more cardiovascular risk factors a patient has, the less likely they are to return to work. Previous use of beta-blockers may indicate a poorer physical condition in patients and can reduce exercise tolerance^[28], and AST levels are positively correlated with cardiovascular disease risk and adverse outcomes^[29]. Therefore, previous use of beta-blockers and high AST levels are detrimental to the patient's return to work. Younger and married patients had a greater rate of return to work, which may be due to the lower incidence of adverse cardiovascular events in this group and the greater need to resume employment^[30].

Given the clinical application of the model, the clearer the focus on the indicators is, the more it can help physicians make decisions. In cases where the performance of the model does not significantly change, we believe that models with fewer factors are more advantageous for clinical application. Further logistic regression analysis revealed that occupation, CR, anterior wall AMI, TG, age, and AST were independent predictive factors, and we constructed a nomogram based on these factors to further extend the applicability of the model. It is more suitable for clinical practice than traditional prediction formulas, enabling the use of this tool to identify patients with a low return-to-work rate at an early stage.

This study has several limitations. First, this was a single-center retrospective study lacking an external dataset for model validation, so it is unclear how the model performs when applied to datasets from other sources. We hope to collect more external independent data in the future to test this model. Second, this study lacked variables related to patient subjective willingness and work, such as willingness to return to work and job satisfaction. We need to collect features from more dimensions to further improve the model's performance. The impact of these factors will be analyzed in future research.

In conclusion, we constructed machine learning-based predictive models of RTW. Machine learning algorithms can potentially help predict the 12-month return-to-work rate of AMI patients who receive phase I cardiac rehabilitation during hospitalization. Additionally, the dynamic nomogram may serve as a beneficial management tool for cardiac rehabilitation practitioners at present.

Study design and patient enrollment. The data for this retrospective study were obtained from the Incidence Rate of Heart Failure After Acute Myocardial Infarction With Optimal Treatment database. The clinical study was registered on ClinicalTrials.gov (NCT03297164). We consecutively collected data from 1473 patients with AMI who had received phase I cardiac rehabilitation and were discharged alive from the Second Affiliated Hospital of Harbin Medical University in China from January 15, 2018, to September 17, 2019. To screen for suitable patients, the following exclusion criteria were used: (1) patients who reached retirement age at the 12-month interview (man ≤ 60 years, woman ≤ 55 years); (2) unemployed before AMI hospitalization; and (3) lost to follow-up. The study was approved by the Ethics Committee of the Second Affiliated Hospital of Harbin Medical University, and informed consent was obtained from all participants involved in the study (Ethics Number: KY2017-254). We followed the Transparent Reporting of Multivariate Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) statement for reporting multivariable prediction model development and validation^[13]. All methods were performed in accordance with the relevant guidelines and regulations.

Feature selection. Clinical characteristics were collected based on previous studies related to return to work^[14], and the optimal variables were identified through the least absolute shrinkage and selection operator (Lasso) method to establish the ML model. Additionally, the Spearman correlation coefficient was utilized to analyze the variable correlation of the factors in the model, and the results are presented in a heatmap.

Outcome. Our outcome was the development of a return to work prediction model. Return to work was defined as being employed either full-time or part-time at the 12-month mark following hospital discharge for AMI patients who received phase I cardiac rehabilitation during hospitalization. Patients were followed up by telephone or outpatient interviews to assess their return to work status at the 12-month mark after discharge from the hospital. Cardiac rehabilitation on the baseline table was defined as whether the patient underwent phase II cardiac rehabilitation within one month after discharge.

ML model development and validation. Based on the selected features, an ML model was constructed. The entire cohort was randomly divided into two datasets: a training cohort (70%) and an internal validation cohort (30%). A total of five ML algorithms were selected to develop the model, including logistic regression (LR), random forest (RF), extreme gradient boosting (XGBoost), support vector machine (SVM), and artificial neural network (ANN) models. Each model was optimized by 5 repeats of 10-fold cross-validation or tuned to the best parameters. We then assessed the models' performance in the training and validation cohorts using the area under the receiver operating characteristic curve (AUC), positive predictive value (PPV), negative predictive value (NPV) accuracy, sensitivity, specificity, and F1-score. Model calibration was measured by the Brier score.

Statistical analysis. All the statistical analyses were performed using R 4.3.2. Continuous variables with normal distributions are expressed as the mean ± standard deviation (mean ± SD) and were compared using the t test. Continuous variables with a nonnormal distribution are presented as medians and quartiles [M (Q1,Q3)] and were analyzed via the Mann–Whitney U test. The chi-square test or Fisher’s exact test was used to compare categorical variables, which are expressed as the number of cases and the constituent ratio [N (%)]. A dynamic nomogram based on the LR model output was established as a tool for clinical application. P values < 0.05 were considered to indicate statistical significance.

Conflicts of interest

The authors declare no competing interests.

Funding

This study was supported by the National Key R&D Program of China (Grant No. 2016YFC1301105), National Natural Science Foundation of China (Grant nos. 82172537 and 82372565), and Key R&D Program of Heilongjiang Province of China (Grant No. 2022ZX01A28).

Author Contribution

X.W.#, and S.W.# conceived the study. H.C. #wrote the main manuscript text. X.H., Z.W., X.Z.,Q.L., and T.C. collected and analyzed the data. Y.B., Y.Z.* and J.W. *critically revised the manuscript. All authors reviewed and approved the final manuscript.#X.W, S.W. , and H.C. contributed equally to this work and should be considered co-first authors.*Corresponding author.

Data Availability

The data supporting the findings of this study are available from the correspondence author upon reasonable request. The data are not publicly available due to privacy or ethical restrictions.

Reed, G. W., Rossi, J. E. & Cannon, C. P. Acute myocardial infarction. Lancet 389, 197-210 (2017). https://doi.org:10.1016/S0140-6736(16)30677-8
Prabhu, N. V., Maiya, A. G. & Prabhu, N. S. Impact of Cardiac Rehabilitation on Functional Capacity and Physical Activity after Coronary Revascularization: A Scientific Review. Cardiol Res Pract 2020, 1236968 (2020). https://doi.org:10.1155/2020/1236968
Allan, B. A., Kim, T. & Shein, B. Underemployment and mental health: A longitudinal study. J Couns Psychol 69, 578-588 (2022). https://doi.org:10.1037/cou0000610
Du, R. et al. Health-related quality of life and associated factors in patients with myocardial infarction after returning to work: a cross-sectional study. Health Qual Life Outcomes 18, 190 (2020). https://doi.org:10.1186/s12955-020-01447-4
Cauter, J. V. et al. Return to work and associations with psychosocial well-being and health-related quality of life in coronary heart disease patients: Results from EUROASPIRE IV. Eur J Prev Cardiol 26, 1386-1395 (2019). https://doi.org:10.1177/2047487319843079
Warraich, H. J., Kaltenbach, L. A., Fonarow, G. C., Peterson, E. D. & Wang, T. Y. Adverse Change in Employment Status After Acute Myocardial Infarction: Analysis From the TRANSLATE-ACS Study. Circ Cardiovasc Qual Outcomes 11, e004528 (2018). https://doi.org:10.1161/CIRCOUTCOMES.117.004528
Bokenberger, K. et al. Work disability patterns before and after incident acute myocardial infarction and subsequent risk of common mental disorders: A Swedish cohort study. Sci Rep 9, 16086 (2019). https://doi.org:10.1038/s41598-019-52487-w
Varianou-Mikellidou, C. et al. Occupational health and safety management in the context of an ageing workforce. Safety Science 116, 231-244 (2019). https://doi.org:10.1016/j.ssci.2019.03.009
Jing Li, et al. ST-segment elevation myocardial infarction in China from 2001 to 2011 (the China PEACE-Retrospective Acute Myocardial Infarction Study): a retrospective analysis of hospital data. The Lancet, 385 (9966), 441-451. https://doi.org/10.1016/s0140-6736(14)60921-1
Haug, C. J. & Drazen, J. M. Artificial Intelligence and Machine Learning in Clinical Medicine, 2023. N Engl J Med 388, 1201-1208 (2023). https://doi.org:10.1056/NEJMra2302038
Kresoja, K. P., Unterhuber, M., Wachter, R., Thiele, H. & Lurz, P. A cardiologist's guide to machine learning in cardiovascular disease prognosis prediction. Basic Res Cardiol 118, 10 (2023). https://doi.org:10.1007/s00395-023-00982-7
Al'Aref, S. J. et al. Clinical applications of machine learning in cardiovascular disease and its relevance to cardiac imaging. Eur Heart J 40, 1975-1986 (2019). https://doi.org:10.1093/eurheartj/ehy404
Collins, G. S., Reitsma, J. B., Altman, D. G., Moons, K. G. & Group, T. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. The TRIPOD Group. Circulation 131, 211-219 (2015). https://doi.org:10.1161/CIRCULATIONAHA.114.014508
Jiang, Z. et al. Factors Associated With Return to Work After Acute Myocardial Infarction in China. JAMA Netw Open 1, e184831 (2018). https://doi.org:10.1001/jamanetworkopen.2018.4831
Monaghan, T. F. et al. Foundational Statistical Principles in Medical Research: Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value. Medicina (Kaunas) 57 (2021). https://doi.org:10.3390/medicina57050503
Zhu, Y. et al. Socioeconomic disparity in mortality and the burden of cardiovascular disease: analysis of the Prospective Urban Rural Epidemiology (PURE)-China cohort study. The Lancet Public Health 8, e968-e977 (2023). https://doi.org:10.1016/s2468-2667(23)00244-x
Stendardo, M. et al. Predicting return to work after acute myocardial infarction: Socio-occupational factors overcome clinical conditions. PLoS One 13, e0208842 (2018). https://doi.org:10.1371/journal.pone.0208842
Pedersen, S. M. et al. Return to work: does cardiac rehabilitation make a difference? Danish nationwide register-based study. Scand J Public Health 51, 179-187 (2023). https://doi.org:10.1177/14034948211062656
Batsis, J. A. et al. Development and Usability Assessment of a Connected Resistance Exercise Band Application for Strength-Monitoring. World Acad Sci Eng Technol 13, 340-348 (2019). https://doi.org:10.5281/zenodo
Hiruma, T. et al. Comprehensive Cardiac Rehabilitation Following Acute Myocardial Infarction Improves Clinical Outcomes Regardless of Exercise Capacity. Circ J 88, 982-992 (2024). https://doi.org:10.1253/circj.CJ-23-0668
Blumenthal, J. A. et al. Enhancing Cardiac Rehabilitation With Stress Management Training: A Randomized, Clinical Efficacy Trial. Circulation 133, 1341-1350 (2016). https://doi.org:10.1161/CIRCULATIONAHA.115.018926
Hurdus, B. et al. Association of cardiac rehabilitation and health-related quality of life following acute myocardial infarction. Heart 106, 1726-1731 (2020). https://doi.org:10.1136/heartjnl-2020-316920
Shah, N. D. et al. Long-term medication adherence after myocardial infarction: experience of a community. Am J Med 122, 961 e967-913 (2009). https://doi.org:10.1016/j.amjmed.2008.12.021
Dibben, G. O. et al. Exercise-based cardiac rehabilitation for coronary heart disease: a meta-analysis. Eur Heart J 44, 452-469 (2023). https://doi.org:10.1093/eurheartj/ehac747
Randal, J. T. Cardiac Rehabilitation — Challenges, Advances, and the Road Ahead. The New England Journal of Medicine (2024). https://doi.org:10.1056/nejmra2302291
Atul, J. & Andrew, M. D. Primary Prevention of Cardiovascular Disease. Jama (2019). https://doi.org:10.1001/jama.2019.15915
Donald, M. L.-J. et al. Life’s Essential 8: Updating and Enhancing the American Heart Association’s Construct of Cardiovascular Health: A Presidential Advisory From the American Heart Association. Circulation (2022). https://doi.org:10.1161/cir.0000000000001078
Murata, M., Adachi, H., Oshima, S. & Kurabayashi, M. Influence of stroke volume and exercise tolerance on peak oxygen pulse in patients with and without beta-adrenergic receptor blockers in patients with heart disease. J Cardiol 69, 176-181 (2017). https://doi.org:10.1016/j.jjcc.2016.02.017
Ndrepepa, G. Aspartate aminotransferase and cardiovascular disease—a narrative review. Journal of Laboratory and Precision Medicine 6, 6-6 (2021). https://doi.org:10.21037/jlpm-20-93
Dhindsa, D. S., Khambhati, J., Schultz, W. M., Tahhan, A. S. & Quyyumi, A. A. Marital status and outcomes in patients with cardiovascular disease. Trends Cardiovasc Med 30, 215-220 (2020). https://doi.org:10.1016/j.tcm.2019.05.012

No competing interests reported.

Supplementary.doc

Download PDF

Reviewers invited by journal
25 Jul, 2024
Editor assigned by journal
05 Jul, 2024
Editor invited by journal
25 Jun, 2024
Submission checks completed at journal
24 Jun, 2024
First submitted to journal
21 Jun, 2024

You are reading this latest preprint version

Machine learning-based prediction of return to work within 12 months in acute myocardial infarction patients

Status:

Version 1

Abstract

Figures

Introduction

Results

Discussion

Methods

Declarations

Conflicts of interest

Funding

Author Contribution

Data Availability

References

Additional Declarations

Supplementary Files

Status:

Version 1