A Machine Learning Model for Predicting the Occurrence of Early Heart Failure in Patients with Acute Myocardial Infarction

A Machine Learning Model for Predicting the Occurrence of Early Heart Failure in Patients with Acute Myocardial Infarction | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article A Machine Learning Model for Predicting the Occurrence of Early Heart Failure in Patients with Acute Myocardial Infarction Shuang Liu, Xuejin Chen, Yanan Hu, Jingjing Jin, Chunmei Qi, Ji Hao This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7859492/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 9 You are reading this latest preprint version Abstract Aims Heart failure (HF) remains a frequent and burdensome complication of acute myocardial infarction (AMI), posing a substantial challenge to global healthcare systems. This study aimed to develop and compare six machine learning (ML) algorithms to identify the optimal model for the early prediction of HF following AMI. Methods We retrospectively enrolled patients admitted for AMI at the Second Affiliated Hospital of Xuzhou Medical University between June 1, 2022, and December 31, 2024. Participants were categorized into HF and non-HF groups based on the occurrence of in-hospital heart failure. The cohort was randomly split into a training set (70%) and a validation set (30%) for model development and internal validation, respectively. Model performance was assessed using the receiver operating characteristic (ROC) curve, and clinical utility was evaluated via decision curve analysis (DCA). Results Among the six ML models evaluated, the extreme gradient boosting (XGBoost) algorithm demonstrated superior predictive performance. Feature importance analysis within the XGBoost model identified the top eight predictors, in descending order of contribution: high-sensitivity C reactive protein (hsCRP), age, aspartate aminotransferase (AST), left ventricular anterior–posterior diameter (LVAPD), blood urea nitrogen (BUN), albumin (ALB), glucose (GLU), and myocardial infarction type(MI). In the validation cohort, the model achieved an area under the ROC curve (AUC) of 0.818. DCA further confirmed its favourable net clinical benefit. Conclusion An XGBoost model incorporating eight readily available clinical features was developed and validated for the early prediction of HF after AMI, showing promising discriminative ability and clinical utility. This tool may assist clinicians in stratifying risk and guiding early intervention. Myocardial infarction Heart failure Machine learning Predict Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Heart failure (HF) is a complex clinical syndrome resulting from structural or functional cardiac abnormalities that impair ventricular filling or ejection. Key clinical manifestations include dyspnea, fatigue, and fluid retention (e.g., pulmonary and systemic congestion, peripheral edema). As a major global public health challenge, HF affects over 64 million individuals worldwide, with an estimated prevalence of 1–2% [ 1 , 2 ] . The condition imposes a substantial economic burden on healthcare systems, driven by high rates of morbidity, hospital readmissions, and mortality, as well as significant impairments in functional capacity and health-related quality of life [ 3 ] . In developed countries, HF accounts for an estimated 1–2% of total annual healthcare expenditures [ 4 ] . Indirect costs—such as productivity loss, early retirement, and informal caregiving—further amplify its socioeconomic impact [ 3 ] . Ischemic heart disease(IHD) remains the leading cause of HF. Following acute myocardial infarction (AMI), the irreversible loss of cardiomyocytes triggers a reparative fibrotic response. This process activates neurohumoral pathways, leading to ventricular remodeling, cardiac dilation, and impaired contractility. Subsequent hemodynamic overload often results in pulmonary congestion, manifesting as dyspnea and chest tightness, and poses a serious threat to patient survival. Against the backdrop of population aging and rising prevalence of comorbidities such as obesity, both the incidence and prevalence of HF are expected to increase. Missed or delayed diagnosis can worsen clinical outcomes and elevate treatment costs, underscoring the critical importance of early screening and risk stratification. Timely identification of high-risk patients not only saves lives but also enables personalized treatment strategies that can improve quality of life [ 5 ] . Recent advances in artificial intelligence (AI) have created new opportunities for enhancing clinical decision-making. Machine learning (ML), a core subset of AI, leverages computational models to identify patterns in large datasets and supports tasks such as diagnosis, risk prediction, and treatment optimization [ 6 , 7 ] . ML approaches can be broadly categorized into supervised, unsupervised, and reinforcement learning. In clinical settings, supervised learning algorithms—such as logistic regression, k-nearest neighbors, decision trees, support vector machines, and naive Bayes—are frequently used for classification tasks, including binary outcome prediction [ 8 ] . For instance, the extreme gradient boosting (XGBoost) algorithm has been employed to predict in-hospital mortality in acute HF [ 9 ] , and random forest models have been applied to HF classification [ 10 ] . These methods can uncover complex, non-linear relationships in multidimensional clinical data, offering valuable insights for disease prediction and diagnosis. In this study, we aimed to develop and compare six ML algorithms for early prediction of HF following AMI. Using routinely collected clinical data, we constructed prediction models and evaluated their performance to identify the most effective approach. We anticipate that a robust ML-based tool will assist clinicians in stratifying post-AMI risk, facilitating early intervention, and ultimately improving patient outcomes. Materials and Methods Study Design and Patient Selection This single-center, retrospective study enrolled consecutive patients diagnosed with acute myocardial infarction (AMI) at the Second Affiliated Hospital of Xuzhou Medical University between June 1, 2022, and December 31, 2024. Patients were stratified into two groups based on the occurrence of in-hospital heart failure (HF): the HF group and the non-HF group. The diagnosis of AMI (including ST-segment elevation MI [STEMI] and non-ST-segment elevation MI [NSTEMI]) was established upon admission according to contemporary guidelines for acute coronary syndrome [11] . Similarly, HF events during hospitalization were identified based on the latest clinical guidelines [12] . Inclusion and Exclusion Criteria Inclusion criteria were as follows: (1) age ≥18 years; (2) initial diagnosis of AMI at admission; and (3) no evidence of HF at the time of admission. Exclusion criteria comprised: (1) HF attributable to valvular heart disease or cardiomyopathy; (2) non-obstructive myocardial infarction; (3) concomitant rheumatic or autoimmune diseases; (4) history of malignant tumor; and (5) in-hospital mortality. The study protocol was approved by the Ethics Committee of the Second Affiliated Hospital of Xuzhou Medical University (Approval No. 2024071501), which waived the requirement for informed consent due to the retrospective nature of the analysis. Data Collection and Preprocessing Demographic, clinical, laboratory, and echocardiographic data were extracted from the hospital's electronic medical record system for all eligible participants and linked to a unique admission identifier. Data preprocessing involved cleaning and standardization: records with extensive missing data (>30%) were excluded, while remaining missing values in continuous variables were imputed using the mean or median after assessing normality with the Shapiro-Wilk test. The robustness of this imputation was evaluated by comparing model performance pre- and post-imputation. Finally, all laboratory values were converted to international standard units to ensure consistency. ML Algorithms and Model Development To develop the prediction model, a two-step feature selection process was employed. First, Least Absolute Shrinkage and Selection Operator (LASSO) regression with 5-fold cross-validation was applied to the initial 61 variables, identifying 24 non-zero coefficients as potentially significant predictors. These 24 variables were subsequently subjected to a multivariate logistic regression analysis, which further refined the feature set to 8 statistically significant indicators. These final 8 features were used as inputs for the following six machine learning algorithms: backpropagation neural network (BPNN), decision tree (DT), logistic regression (LR), random forest (RF), support vector machine (SVM), and XGBoost. The dataset was randomly partitioned into a training set (70%) for model construction and a validation set (30%) for performance evaluation. The models were trained on the training set, and their predictive performance was assessed on the validation set. Furthermore, 5-fold cross-validation was implemented during the training phase of each algorithm to optimize hyperparameters and ensure model robustness. Statistical Analysis Data management and statistical analyses were performed using Excel 2024 and SPSS software (version 23.0). Continuous variables were assessed for normality using the Shapiro–Wilk test. Normally distributed data are presented as mean ± standard deviation and were compared between groups using the independent samples t-test. Non-normally distributed data are expressed as median (interquartile range, IQR) and were compared using the Mann–Whitney U test. Categorical variables are summarized as frequencies (percentages) and were compared using Pearson’s chi-square test or Fisher’s exact test, as appropriate. The predictive performance of the developed models was evaluated by the area under the receiver operating characteristic curve (AUC). The clinical utility of the optimal model was assessed using decision curve analysis (DCA). A two-sided p-value of less than 0.05 was considered statistically significant. Results Study Population and Baseline Characteristics A total of 664 patients with AMI were included in the final analysis. Based on the occurrence of in-hospital HF, 421 patients (63.4%) were classified into the HF group and 223 (33.6%) into the non-HF group. The baseline characteristics of the two groups are summarized in Table 1. Significant differences were observed in age, MI type (coded as 1 for STEMI and 2 for NSTEMI), history of diabetes, and sex between the two groups. Consequently, these variables were included as candidate features for subsequent model development. In contrast, no significant differences were found in the prevalence of hypertension or history of alcohol consumption between the groups. Table 1：Demographic characteristics of the participants. Variable Group Z/χ² p NHF HF Age 61.0（52.5, 70.0） 71.0（61.0, 79.0） -7.9692 ＜0.001 MI STEMI 116(18.0%) 264(41.0%) 6.887 0.09 NSTEMI 107(16.6%) 157(24.4%) Hypertension NO 107(16.6%) 180(28.0%) 1.612 0.212 YES 116(18.0%) 241(37.4%) Diabetes NO 152(23.6%) 249(38.7%) 5.044 0.026 YES 71(11.0%) 172(26.7%) Sex Male 183(28.4%) 293(45.5%) 11.75 0.001 Female 40(6.2%) 128(19.9%) Smoke NO 108(16.8%) 243(37.7%) 5.073 0.025 YES 115(17.9%) 178(27.6%) Alcohol NO 176(27.3%) 344(53.4%) 0.728 0.402 YES 47(7.3%) 77(12.0%) NHF: non-Heart failure group; HF: Heart failure group；MI：type of AMI. Model Construction and Feature Selection A total of 61 variables were initially extracted from patient baseline characteristics, initial laboratory tests, and echocardiographic examinations as potential predictors for model development. Missing quantitative data were imputed using the median value of each variable. Feature selection was performed in two stages: first, LASSO regression with 5-fold cross-validation was applied, which reduced the feature set to 24 non-redundant variables (Figures 1 and 2). These variables included type of myocardial infarction (MI), diabetes, sex, age, diastolic blood pressure (DBP), neutrophil count (NEUT), albumin (ALB), globulin (GLB), aspartate aminotransferase (AST), alkaline phosphatase (ALP), blood urea nitrogen (BUN), serum creatinine (Scr), glucose (GLU), high-density lipoprotein cholesterol (HDLC), lipoprotein(a) (Lpa), residual cholesterol (RC), high-sensitivity C-reactive protein (hsCRP), prothrombin time (PT), high-sensitivity cardiac troponin I (hscTnI), creatine kinase-MB (CKMB), left atrial anterior–posterior diameter (LAAPD), left ventricular anterior–posterior diameter (LVAPD), left ventricular posterior wall thickness (LVPW), and right ventricular anterior–posterior diameter (RVAPD). Subsequently, logistic regression was used to further refine the predictor set, identifying the eight most clinically meaningful features: MI type, age, ALB, AST, BUN, GLU, PT, hsCRP, and LVAPD, which were used to construct the final prediction model for post-infarction heart failure (Figure 3). Model Selection and Validation The cohort of 664 AMI patients was randomly divided into a training set (70%) and a validation set (30%). Six machine learning algorithms—BPNN, DT, LR, RF, SVM, and XGBoost—were trained and evaluated using these datasets. The sensitivity, specificity, and AUC values for both the training and validation sets are summarized in Table 2a, 2b and illustrated in Figures 4a and 4b. Table 2a: Diagnostic efficacy of Six classifiers（Train set） Model SPE SEN ACC PR Recall F1 AUC BPNN 0.503 0.913 0.765 0.765 0.913 0.832 0.830 DT 0.795 0.878 0.849 0.885 0.878 0.881 0.886 Logistic 0.609 0.854 0.766 0.791 0.854 0.821 0.846 RF 1.000 1.000 1.000 1.000 1.000 1.000 1.000 SVM 0.826 0.948 0.904 0.907 0.948 0.927 0.868 XGBoost 0.702 0.896 0.826 0.843 0.896 0.869 0.879 Table 2b: Diagnostic efficacy of Six classifiers（Test set） Model SPE SEN ACC PR Recall F1 AUC BPNN 0.450 0.910 0.767 0.786 0.910 0.843 0.794 DT 0.565 0.759 0.697 0.789 0.759 0.774 0.720 Logistic 0.710 0.767 0.749 0.850 0.767 0.806 0.795 RF 0.629 0.707 0.682 0.803 0.707 0.752 0.786 SVM 0.565 0.865 0.769 0.810 0.865 0.837 0.776 XGBoost 0.677 0.820 0.774 0.845 0.820 0.832 0.818 SPE = True Negative/( True Negative + False Positive); SEN = True Positive /( True Positive + False Negative); ACC = (True Positive + True Negative)/( Positive + Negative); PR = True Positive/( True Positive + False Positive); Recall = True Positive /( True Positive + False Negative); F1= 2*Precision*Recal / (Precision + Recal) Based on a comprehensive assessment of sensitivity, specificity, and AUC—while also accounting for potential overfitting or underfitting—the XGBoost model was identified as the best-performing predictor. Feature importance analysis within the XGBoost model revealed the following predictors in descending order of contribution: hs-CRP, age, AST, LVAPD, BUN, albumin, glucose, and MI type (Figure 5). DCA further demonstrated that the XGBoost model provided substantial net clinical benefit across both the training and validation sets (Figure 6). Discussion The growing global burden of AMI is closely linked to contemporary shifts in lifestyle and increasing exposure to psychosocial stressors, contributing to its rising incidence. As a leading cause of HF, AMI imposes a substantial economic and healthcare burden worldwide. This challenge is further compounded by population aging, which has been widely recognized as an independent risk factor for HF. Epidemiological studies indicate that adults over 60 years of age face a markedly elevated risk of developing HF compared to younger individuals [ 13 ] , and aging is strongly associated with the pathogenesis and progression of the disease [ 14 ] . In our study, the mean age of both the HF and non-HF groups exceeded 60 years, with patients in the HF group being significantly older. Those affected by HF frequently experience recurrent hospitalizations, long-term pharmacotherapy, and severely impaired functional capacity, leading to considerable loss of productivity and diminished quality of life. These factors underscore the critical importance of early prediction of HF following AMI [ 15 , 16 ] . The widespread adoption of electronic medical record (EMR) systems in healthcare institutions has generated vast repositories of clinical data, which—while requiring rigorous information security measures—provide a valuable foundation for applying ML in clinical research. Conventional statistical methods often face limitations in handling high-dimensional and complex multimodal health data, which can restrict the predictive performance of resulting models. In contrast, ML techniques offer a powerful alternative by leveraging computational algorithms to automatically learn patterns from data without relying solely on pre-specified hypotheses. ML models are trained by establishing mappings between input features and output labels, enabling the prediction of outcomes through classification or regression mechanisms [ 17 ] . These methods can capture both linear and non-linear relationships, but their performance heavily depends on appropriate feature selection and model design. Consequently, model development is an iterative, empirically driven process that involves continuous adjustment of algorithms and hyperparameters based on performance feedback [ 18 ] . Current diagnosis of heart failure relies on comprehensive assessment of patient symptoms, physical signs, laboratory tests, imaging findings, and medical history. In this study, we developed and validated an XGBoost-based machine learning model using routinely collected clinical data—including baseline characteristics, laboratory results, and echocardiographic parameters—to predict early-onset heart failure following AMI. Feature importance analysis revealed that hsCRP was the most influential predictor in our model, with significantly higher levels observed in the HF group compared to the non-HF group. As a sensitive inflammatory marker, hsCRP reflects underlying chronic inflammation, which contributes to atherosclerotic plaque vulnerability and progression of coronary artery disease. Elevated hsCRP levels have been documented in patients with non-ST-segment elevation acute coronary syndrome (NSTEMI-ACS) and have shown predictive value for new-onset HF [ 19 ] . Similarly, elevated hsCRP in stable ACS patients has been associated with a two-fold increased risk of new or worsening HF within two years [ 20 ] . AST and ALB were also identified as significant predictors in our model. Elevated AST following AMI may result from oxidative stress, cardiomyocyte necrosis, or ischemia-reperfusion injury, although direct evidence linking AST to HF progression remains limited [ 21 , 22 ] . Thus, AST may serve as an auxiliary predictive marker in this context. Hypoalbuminemia, frequently observed in HF patients (affecting up to 40% in some cohorts [ 23 ] ), is associated with higher NYHA functional class, reduced renal function, and increased comorbidity burden [ 24 ] . Higher serum albumin levels have been correlated with improved outcomes in HF with preserved ejection fraction (HFpEF) [ 25 ] , supporting ALB’s relevance as a prognostic indicator. LVAPD also contributed to model performance. Following AMI, compensatory ventricular dilation may initially maintain stroke volume but can progress to adverse remodeling and functional deterioration. Previous studies have linked LVAPD to adverse outcomes in both HFpEF and HF with reduced ejection fraction (HFrEF) [ 26 , 27 ] , consistent with our findings. BUN emerged as another relevant predictor. In HF, reduced cardiac output activates neurohormonal systems such as the sympathetic nervous system and renin–angiotensin–aldosterone system, promoting renal sodium retention and elevated BUN. These pathways are associated with worse cardiac function and prognosis [ 28 , 29 ] , and BUN has been consistently linked to HF outcomes [ 30 , 31 ] . Diabetes and elevated admission blood glucose are well-established risk factors for HF, with diabetic patients facing a two-fold increase in HF risk in men and up to five-fold in women after age adjustment [ 32 – 35 ] . Hyperglycemia at admission may reflect impaired myocardial energy metabolism and oxidative stress, further aggravating cardiac injury [ 36 – 38 ] and contributing to HF onset [ 39 ] . Finally, MI type (STEMI vs. NSTEMI) was incorporated into the model. STEMI, typically resulting from transmural ischemia, often leads to more extensive myocardial damage and higher HF incidence compared to NSTEMI [ 40 ] . Although MI type had the lowest feature importance in our model, it still provided complementary predictive value, consistent with prior studies [ 41 ] . Strengths and Limitations This study possesses several notable strengths. First, the use of initial patient data obtained upon admission enabled the early identification of key predictors, creating a valuable time window for preventive strategies against post-infarction heart failure. Through a rigorous two-step feature selection process—incorporating LASSO regression followed by logistic regression—we distilled 8 clinically meaningful predictors from an initial set of 61 variables, achieving an optimal balance between model performance and simplicity. Furthermore, the model relies exclusively on routinely collected electronic medical record data, requiring no additional costs or specialized examinations, which enhances its potential for real-world implementation. Finally, by systematically comparing six machine learning algorithms and selecting XGBoost based on its AUC performance, robustness against overfitting and underfitting, and net clinical benefit, this study offers a reliable prediction tool. Several limitations should also be acknowledged. As a single-center retrospective analysis, the findings may reflect local patient characteristics and clinical practices, and missing data, though handled, may introduce bias. Future multicenter prospective studies are needed to validate and generalize the results. The relatively limited sample size may also constrain the model’s predictive power; expanding the cohort in subsequent research could improve accuracy and stability. In addition, this study incorporated basic demographic, laboratory, and echocardiographic variables but did not include electrocardiographic, coronary CTA, or angiographic data. Integrating these parameters in future models could further enhance predictive comprehensiveness. Conclusion In this study, we developed and compared six machine learning models for predicting in-hospital heart failure following acute myocardial infarction. The XGBoost algorithm demonstrated superior performance, forming the basis of a final prediction model that incorporates eight key clinical features: hs-CRP, age, AST, LVAPD, BUN, albumin, glucose, and MI type. This model exhibits strong predictive capability and clinical utility, offering a reliable, data-driven tool for early risk stratification that may support clinical decision-making and improve patient management. Abbreviations AI = artificial intelligence ALB = albumin ALP = alkaline phosphatase AMI = acute myocardial infarction AST = aspartate aminotransferase AUC = an area under the ROC curve BPNN = backpropagation neural network BUN = blood urea nitrogen CKMB = creatine kinase-MB DBP = diastolic blood pressure DCA = ecision curve analysis DT = decision tree EMR = electronic medical record GLB = globulin GLU = glucose HDLC = high-density lipoprotein cholesterol HF = heart failure HFpEF = heart failure with preserved ejection fraction HFrEF = heart failure with reduced ejection fraction hsCRP = high-sensitivity C reactive protein hscTnI = high-sensitivity cardiac troponin I IHD = Ischemic heart disease LAAPD = left atrial anterior–posterior diameter LASSO Regression = Least Absolute Shrinkage and Selection Operator regression Lpa = lipoprotein LR = logistic regression LVAPD = left ventricular anterior–posterior diameter LVPW = left ventricular posterior wall thickness ML = type of AMI NEUT = neutrophil count NSTEMI = non-ST-segment elevation MI NSTEMI-ACS = non-ST-segment elevation acute coronary syndrome PT = prothrombin time RC = residual cholesterol RF = random forest ROC = receiver operating characteristic RVAPD = right ventricular anterior–posterior diameter Scr = serum creatinine STEMI = ST-segment elevation MI SVM = support vector machine XGBoost = extreme gradient boosting Declarations Acknowledgements None. Authors’ contributions Shuang Liu conceptualized and designed the study. Xuejin Chen and Yanan Hu collected the data. Shuang Liu, Xuejin Chen and Jingjing Jin performed the analyses and produced the results. Shuang Liu, Xuejin Chen and Yanan Hu analysed the results and wrote the manuscript. Chunmei Qi and Ji Hao provided funding and reviewed the manuscript. Funding None. Data Availability The datasets used and analysed during the current study available from the corresponding author on reasonable request. Ethics approval and consent to participate A statement to confirm that all methods were carried out in accordance with relevant guidelines and regulations. The study protocol was approved by the Ethics Committee of the Second Affiliated Hospital of Xuzhou Medical University (Approval No. 2024071501), which waived the requirement for informed consent due to the retrospective nature of the analysis. Consent for publication Not applicable. Competing interests The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. References Savarese G, Becher PM, Lund LH, et al. Global burden of heart failure: a comprehensive and updated review of epidemiology. Cardiovasc Res. 2023 Jan 18;118(17):3272-3287. doi: 10.1093/cvr/cvac013. Khan MS, Shahid I, Bennis A, et al. Global epidemiology of heart failure. Nat Rev Cardiol. 2024 Oct;21(10):717-734. doi: 10.1038/s41569-024-01046-6. Hessel FP. Overview of the socio-economic consequences of heart failure. Cardiovasc Diagn Ther. 2021 Feb;11(1):254-262. doi: 10.21037/cdt-20-291. Ong SC, Low JZ, Yew WY, et al. Cost analysis of chronic heart failure management in Malaysia: A multi-centred retrospective study. Front Cardiovasc Med. 2022 Nov 2;9:971592. doi: 10.3389/fcvm.2022.971592. Mohammad MA, Olesen KKW, Koul S, et al. Development and validation of an artif-icial neural network algorithm to predict mortality and admission to hospital for heart failure after myocardial infarction: a nationwide population-based study. Lancet Digit Health. 2022 Jan;4(1):e37-e45. doi: 10.1016/S2589-7500(21)00228-4. Li F, Wang Y, Xu T, et al. Deep learning-based automated detection for diabetic reti-nopathy and diabetic macular oedema in retinal fundus photographs. Eye (Lond). 2022 Jul;36(7):1433-1441. doi: 10.1038/s41433-021-01552-8. Ski CF, Thompson DR, Brunner-La Rocca HP. Putting AI at the centre of heart failu-re care. ESC Heart Fail. 2020 Oct;7(5):3257-3258. doi: 10.1002/ehf2.12813. Yasmin F, Shah SMI, Naeem A, et al. Artificial intelligence in the diagnosis and det-ection of heart failure: the past, present, and future. Rev Cardiovasc Med. 2021 Dec 22;22(4):1095-1113. doi: 10.31083/j.rcm2204121. Tanaka M, Kohjitani H, Yamamoto E, et al. Development of interpretable machine le-arning models to predict in-hospital prognosis of acute heart failure patients. ESC He-art Fail. 2024 Oct;11(5):2798-2812. doi: 10.1002/ehf2.14834. Chulde-Fernández B, Enríquez-Ortega D, Guevara C, et al. Classification of Heart Fai-lure Using Machine Learning: A Comparative Study. Life (Basel). 2025 Mar 19;15(3):496. doi: 10.3390/life15030496. Rao SV, O'Donoghue ML, Ruel M, et al. 2025 ACC/AHA/ACEP/NAEMSP/SCAI Gui-deline for the Management of Patients With Acute Coronary Syndromes: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2025 Apr;151(13):e771-e862. doi: 10.1161/CIR.0000000000001309. Chinese Society of Cardiology, Chinese Medical Association; Chinese College of Card-iovascular Physician; Chinese Heart Failure Association of Chinese Medical Doctor Association; Editorial Board of Chinese Journal of Cardiology. [Chinese guidelines for the diagnosis and treatment of heart failure 2024]. Zhonghua Xin Xue Guan Bing Za Zhi. 2024 Mar 24;52(3):235-275. Chinese. doi: 10.3760/cma.j.cn112148-20231101-00405. Goyal P, Maurer MS, Roh J. Aging in Heart Failure: Embracing Biology Over Chron-ology: JACC Family Series. JACC Heart Fail. 2024 May;12(5):795-809. doi: 10.1016/j.jchf.2024.02.021. Epub 2024 Apr 8. Fang Z, Raza U, Song J, et al. Systemic aging fuels heart failure: Molecular mechani-sms and therapeutic avenues. ESC Heart Fail. 2025 Apr;12(2):1059-1080. doi: 10.1002/ehf2.14947. Epub 2024 Jul 22. Darvish M, Shakoor A, Feyz L, et al. Heart failure: assessment of the global econom-ic burden. Eur Heart J. 2025 Aug 14;46(31):3069-3078. doi: 10.1093/eurheartj/ehaf323. Roth GA, Forouzanfar MH, Moran AE, et al. Demographic and epidemiologic drivers of global cardiovascular mortality. N Engl J Med. 2015 Apr 2;372(14):1333-41. doi: 10.1056/NEJMoa1406656. Itchhaporia D. Artificial intelligence in cardiology. Trends Cardiovasc Med. 2022 Jan;32(1):34-41. doi: 10.1016/j.tcm.2020.11.007. Vandenberk B, Chew DS, Prasana D, Gupta S, Exner DV. Successes and challenges of artificial intelligence in cardiology. Front Digit Health. 2023 Jun 28;5:1201392. doi: 10.3389/fdgth.2023.1201392. James SK, Armstrong P, Barnathan E, et al. Troponin and C-reactive protein have dif-ferent relations to subsequent mortality and myocardial infarction after acute coronary syndrome: a GUSTO-IV substudy. J Am Coll Cardiol. 2003 Mar 19;41(6):916-24. doi: 10.1016/s0735-1097(02)02969-8. Scirica BM, Cannon CP, Sabatine MS, et al. Concentrations of C-reactive protein and B-type natriuretic peptide 30 days after acute coronary syndromes independently predict hospitalization for heart failure and cardiovascular death. Clin Chem. 2009 Feb;55(2):265-73. doi: 10.1373/clinchem.2008.117192. Han JH, Kwak JY, Lee SS, Kim HG, Jeon H, Cha RR. Markedly Elevated Aspartate Aminotransferase from Non-Hepatic Causes. J Clin Med. 2022 Dec 30;12(1):310. doi: 10.3390/jcm12010310. Myhre PL, Grupper A, Mebazaa A, et al. Changes in Liver Function Tests, Congestio-n, and Prognosis After Acute Heart Failure: The STRONG-HF Trial. JACC Adv. 2025 Mar;4(3):101607. doi: 10.1016/j.jacadv.2025.101607. El Iskandarani M, El Kurdi B, Murtaza G, Paul TK, Refaat MM. Prognostic role of albumin level in heart failure: A systematic review and meta-analysis. Medicine (Baltimore). 2021 Mar 12;100(10):e24785. doi: 10.1097/MD.0000000000024785. Bonilla-Palomas JL, Gámez-López AL, Moreno-Conde M, et al. Hypoalbuminemia in acute heart failure patients: causes and its impact on hospital and long-term mortality. J Card Fail. 2014 May;20(5):350-8. doi: 10.1016/j.cardfail.2014.01.016. Dastjerdi P, Mahalleh M, Shayesteh H, et al. Liver biomarkers as predictors of progn-osis in heart failure with preserved ejection fraction: a systematic review and meta-analysis. BMC Cardiovasc Disord. 2025 Apr 2;25(1):244. doi: 10.1186/s12872-025-04647-2. Ogawa S, Nagatomo Y, Takei M, et al. Impact of Left Ventricular Chamber Size on Outcome in Heart Failure with Preserved Ejection Fraction. Int Heart J. 2022;63(1):62-72. doi: 10.1536/ihj.21-486. Rowe SJ, Xiang R, Paratz ED, Takeuchi F, La Gerche A. Left ventricular size and heart failure: A cardiac MRI assessment of 38,129 individuals from the UK Biobank. Int J Cardiol. 2025 Jan 15;419:132687. doi: 10.1016/j.ijcard.2024.132687. Rengo G, Lymperopoulos A, Zincarelli C, et al. Blockade of β-adrenoceptors restores the GRK2-mediated adrenal α(2) -adrenoceptor-catecholamine production axis in heart failure. Br J Pharmacol. 2012 Aug;166(8):2430-40. doi: 10.1111/j.1476-5381.2012.01972.x. Bozic J, Kumric M, Ticinovic Kurir T, et al. Catestatin as a Biomarker of Cardiovas-cular Diseases: A Clinical Perspective. Biomedicines. 2021 Nov 25;9(12):1757. doi: 10.3390/biomedicines9121757. Xu Z, Yue Y, Xu M, et al. The role of cut-off values for creatinine, blood urea nitr-ogen, and uric acid in prognostic assessment of chronic heart failure: a retrospective cohort study. BMC Cardiovasc Disord. 2025 Mar 22;25(1):209. doi: 10.1186/s12872-025-04675-y. Duan S, Li Y, Yang P. Predictive value of blood urea nitrogen in heart failure: a sys-tematic review and meta-analysis. Front Cardiovasc Med. 2023 Jul 31;10:1189884. doi: 10.3389/fcvm.2023.1189884. Nichols GA, Hillier TA, Erbey JR, Brown JB. Congestive heart failure in type 2 dia-betes: prevalence, incidence, and risk factors. Diabetes Care. 2001 Sep;24(9):1614-9. doi: 10.2337/diacare.24.9.1614. Dei Cas A, Khan SS, Butler J, et al. Impact of diabetes on epidemiology, treatment, and outcomes of patients with heart failure. JACC Heart Fail. 2015 Feb;3(2):136-45. doi: 10.1016/j.jchf.2014.08.004. Kannel WB, Hjortland M, Castelli WP. Role of diabetes in congestive heart failure: the Framingham study. Am J Cardiol. 1974 Jul;34(1):29-34. doi: 10.1016/0002-9149(74)90089-7. Kannel WB, McGee DL. Diabetes and cardiovascular disease. The Framingham study. JAMA. 1979 May 11;241(19):2035-8. doi: 10.1001/jama.241.19.2035. Montaigne D, Marechal X, Coisne A, et al. Myocardial contractile dysfunction is asso-ciated with impaired mitochondrial function and dynamics in type 2 diabetic but not in obese patients. Circulation. 2014 Aug 12;130(7):554-64. doi: 10.1161/CIRCULATIONAHA.113.008476. Lowell BB, Shulman GI. Mitochondrial dysfunction and type 2 diabetes. Science. 2005 Jan 21;307(5708):384-7. doi: 10.1126/science.1104343. Liesa M, Shirihai OS. Mitochondrial dynamics in the regulation of nutrient utilization and energy expenditure. Cell Metab. 2013 Apr 2;17(4):491-506. doi: 10.1016/j.cmet.2013.03.002. Ritsinger V, Hagström E, Lagerqvist B, Norhammar A. Admission Glucose Levels an-d Associated Risk for Heart Failure After Myocardial Infarction in Patients Without Diabetes. J Am Heart Assoc. 2021 Nov 16;10(22):e022667. doi: 10.1161/JAHA.121.022667. Butler J, Hammonds K, Talha KM, et al. Incident heart failure and recurrent coronar-y events following acute myocardial infarction. Eur Heart J. 2025 Apr 22;46(16):1540-1550. doi: 10.1093/eurheartj/ehae885. Li X, Shang C, Xu C, Wang Y, Xu J, Zhou Q. Development and comparison of ma-chine learning-based models for predicting heart failure after acute myocardial infarction. BMC Med Inform Decis Mak. 2023 Aug 24;23(1):165. doi: 10.1186/s12911-023-02240-1. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 20 Jan, 2026 Reviews received at journal 15 Jan, 2026 Reviewers agreed at journal 15 Jan, 2026 Reviews received at journal 14 Jan, 2026 Reviewers agreed at journal 14 Jan, 2026 Reviewers invited by journal 19 Nov, 2025 Editor assigned by journal 15 Oct, 2025 Submission checks completed at journal 15 Oct, 2025 First submitted to journal 14 Oct, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7859492","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":549983252,"identity":"5ba64ba2-488e-4470-8b62-fcd094c68b4d","order_by":0,"name":"Shuang Liu","email":"","orcid":"","institution":"The Second Affiliated Hospital of Xuzhou Medical","correspondingAuthor":false,"prefix":"","firstName":"Shuang","middleName":"","lastName":"Liu","suffix":""},{"id":549983253,"identity":"b1463c6d-6dd1-4822-b206-195b177c6dfd","order_by":1,"name":"Xuejin Chen","email":"","orcid":"","institution":"The Second Affiliated Hospital of Xuzhou Medical","correspondingAuthor":false,"prefix":"","firstName":"Xuejin","middleName":"","lastName":"Chen","suffix":""},{"id":549983254,"identity":"161bca1f-691a-444b-94b2-b2c5fe81f300","order_by":2,"name":"Yanan Hu","email":"","orcid":"","institution":"The Second Affiliated Hospital of Xuzhou Medical","correspondingAuthor":false,"prefix":"","firstName":"Yanan","middleName":"","lastName":"Hu","suffix":""},{"id":549983255,"identity":"637c0918-c1f9-4c6a-85b6-2855037c37b0","order_by":3,"name":"Jingjing Jin","email":"","orcid":"","institution":"The Second Affiliated Hospital of Xuzhou Medical","correspondingAuthor":false,"prefix":"","firstName":"Jingjing","middleName":"","lastName":"Jin","suffix":""},{"id":549983256,"identity":"a5f272c4-2de4-43f8-a7b7-df1968266dd2","order_by":4,"name":"Chunmei Qi","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA5UlEQVRIie3RsWsCMRTH8RQh1yF46xPK9V8ICErBP+b3ELLVFlxucLihmEG7989w7KgEdEl3xxyCc7uUdpFKO1ruzs0hn/l94b1EiCi6QDKxm3c+UJYmBQfkk/qkrbwRQQ66nflyp4Nf1ycZoXcVpGH9wvtO+dRqsBjBBCgHrZzJuZAitTNUJ6rcaJAb9ZOp2fLrjSD/tqhOEoCg3fhu7o+Jl0LTfU0ioAlwvNg+fD7ytNUguUaPsDTHBEY0S5Qfai5+H3lI8GtVe8uttavyu/j7yo+vfJKl9rk6OaHOG4+iKIr+9QMvUlBDFi94nQAAAABJRU5ErkJggg==","orcid":"","institution":"The Second Affiliated Hospital of Xuzhou Medical","correspondingAuthor":true,"prefix":"","firstName":"Chunmei","middleName":"","lastName":"Qi","suffix":""},{"id":549983257,"identity":"1c1d2aae-baa7-41b9-a903-16507cbed635","order_by":5,"name":"Ji Hao","email":"","orcid":"","institution":"The Second Affiliated Hospital of Xuzhou Medical","correspondingAuthor":false,"prefix":"","firstName":"Ji","middleName":"","lastName":"Hao","suffix":""}],"badges":[],"createdAt":"2025-10-14 14:08:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7859492/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7859492/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":96979311,"identity":"effbaa55-40f3-4ac5-a249-061d72e4b03c","added_by":"auto","created_at":"2025-11-28 08:54:23","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":727043,"visible":true,"origin":"","legend":"","description":"","filename":"AMLModelforPredictingEarlyHFinPatientswithAMI.docx","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/1da40bdf93dfb4eeeb801605.docx"},{"id":96979278,"identity":"82e1043d-246b-44bc-acea-6551d709b0b9","added_by":"auto","created_at":"2025-11-28 08:54:20","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7644,"visible":true,"origin":"","legend":"","description":"","filename":"327df5ac6e41425f88786eb3f459cc71.json","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/959984f7774d189a2ca3831f.json"},{"id":96979273,"identity":"ee79c5c8-472a-40ad-be3d-20fee4540709","added_by":"auto","created_at":"2025-11-28 08:54:20","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":118842,"visible":true,"origin":"","legend":"","description":"","filename":"327df5ac6e41425f88786eb3f459cc711enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/cab06cf06c596d4f4e40a63b.xml"},{"id":96979286,"identity":"544106b1-141a-4635-b8ff-f3cffff6930f","added_by":"auto","created_at":"2025-11-28 08:54:21","extension":"jpeg","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":145502,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/5378643d7c35325334822a6d.jpeg"},{"id":96979303,"identity":"554d6d24-6ad6-4698-8a71-a9645da9f063","added_by":"auto","created_at":"2025-11-28 08:54:23","extension":"jpeg","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":122027,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/a5066cac8c8010acbeb01039.jpeg"},{"id":96979262,"identity":"66dae69c-9c6c-4231-8c38-0caf137647ba","added_by":"auto","created_at":"2025-11-28 08:54:18","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":59232,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/0bf2e023427fba23a90c88cc.jpeg"},{"id":96979280,"identity":"46331022-10cc-4616-a2b4-0b195f6c3824","added_by":"auto","created_at":"2025-11-28 08:54:21","extension":"jpeg","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":88020,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/4aee7d1073d9b528eb3d805b.jpeg"},{"id":96979322,"identity":"316531b8-a56c-4e69-8857-694f88f0eee6","added_by":"auto","created_at":"2025-11-28 08:54:24","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":94934,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/fd5e152d4b38f0257c3e36f1.jpeg"},{"id":96979261,"identity":"eead8912-7fb5-49fe-8621-dafc850f5d51","added_by":"auto","created_at":"2025-11-28 08:54:18","extension":"jpeg","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":78020,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/39de0ca0d6710d5078f6e9bb.jpeg"},{"id":96979267,"identity":"d901ad44-ca35-483f-b5af-b64227899e70","added_by":"auto","created_at":"2025-11-28 08:54:19","extension":"jpeg","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":85526,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/9b6bebfbdd9600cc1fc9610b.jpeg"},{"id":96979297,"identity":"4ad35e89-38dd-481d-b058-08f78fb1da70","added_by":"auto","created_at":"2025-11-28 08:54:23","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14449,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/1c56cc1bc8a6915f9f24f5f1.png"},{"id":96979274,"identity":"6787ef98-f312-4648-ba81-c530a3b10541","added_by":"auto","created_at":"2025-11-28 08:54:20","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13632,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/b0ebb838d592fc60ddc3e77e.png"},{"id":96979253,"identity":"22d6d208-8b50-4d05-9710-1a0ba4bdef2b","added_by":"auto","created_at":"2025-11-28 08:54:17","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":16592,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/e01c948853e2bd37d37ac74e.png"},{"id":96979281,"identity":"e2479c3a-4d5a-4c4d-ad0d-d389239d421d","added_by":"auto","created_at":"2025-11-28 08:54:21","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":22852,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/3195171dafeb637e53806859.png"},{"id":96979288,"identity":"2af243a4-d893-49e8-83fe-32772e97f6ea","added_by":"auto","created_at":"2025-11-28 08:54:21","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":29414,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/6b3a87c4819e0569e88aac06.png"},{"id":96979264,"identity":"29984ed7-1eda-4713-8513-b98647fb042d","added_by":"auto","created_at":"2025-11-28 08:54:18","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13082,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/5c944893b083b7f3262a147b.png"},{"id":96979268,"identity":"e4b90ee2-edb5-4236-b2de-a557cb6f03c6","added_by":"auto","created_at":"2025-11-28 08:54:19","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":11892,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/e510b99f0da591df02049b07.png"},{"id":96979318,"identity":"bab67bc3-33af-4073-89d0-21e128e6a711","added_by":"auto","created_at":"2025-11-28 08:54:23","extension":"xml","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":117252,"visible":true,"origin":"","legend":"","description":"","filename":"327df5ac6e41425f88786eb3f459cc711structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/25d50339fb7269f7eca0fe5d.xml"},{"id":96979276,"identity":"5e9739a2-a601-4da8-839d-4d6ed4aabd23","added_by":"auto","created_at":"2025-11-28 08:54:20","extension":"html","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":126399,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/6d48f309891e5ab2ccc4cbff.html"},{"id":96979244,"identity":"09e7266d-9744-41d9-930b-b1166bf4a3b5","added_by":"auto","created_at":"2025-11-28 08:54:16","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":145502,"visible":true,"origin":"","legend":"\u003cp\u003eCoefficient path diagram of LASSO regression model.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/7a5178b8ddcf86db52324eec.jpeg"},{"id":96979295,"identity":"f857449f-b089-4deb-a6e5-11f12271759f","added_by":"auto","created_at":"2025-11-28 08:54:22","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":122027,"visible":true,"origin":"","legend":"\u003cp\u003eLASSO regression cross-validation plot.\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/bdc603f5c9f9786ade32b061.jpeg"},{"id":97138931,"identity":"969358b2-229a-48cd-ad73-a5bd9c621364","added_by":"auto","created_at":"2025-12-01 09:59:26","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":59232,"visible":true,"origin":"","legend":"\u003cp\u003eForest plots of 8 variable effects after logstic regression screening.\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/27dc28844a206becf8ce9303.jpeg"},{"id":96979258,"identity":"41c5f386-7c2e-4abb-9f0b-64cf74b9db7b","added_by":"auto","created_at":"2025-11-28 08:54:18","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":382108,"visible":true,"origin":"","legend":"\u003cp\u003ea: ROC curve of the training set.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb\u003c/strong\u003e: ROC curve of the testing set.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/2417320146c17902947a0eca.png"},{"id":96979325,"identity":"c5c472ba-8e95-4792-8788-f2a82663944a","added_by":"auto","created_at":"2025-11-28 08:54:24","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":78020,"visible":true,"origin":"","legend":"\u003cp\u003eFeature importance ranking of the 8 features in the XGBoost.\u003c/p\u003e","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/2bb95f90909601063ec50e3a.jpeg"},{"id":96979360,"identity":"7a6e88f8-dec6-4d59-92bd-e62ac418226c","added_by":"auto","created_at":"2025-11-28 08:54:24","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":85526,"visible":true,"origin":"","legend":"\u003cp\u003eThe DCA curve of the XGBoost.\u003c/p\u003e","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/caf618be6a0de72aa87056a4.jpeg"},{"id":97144742,"identity":"182b6dd1-6cff-4015-8555-3895efb81260","added_by":"auto","created_at":"2025-12-01 10:11:50","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1565787,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7859492/v1/087df767-0234-478e-8b4c-aa24202734cc.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Machine Learning Model for Predicting the Occurrence of Early Heart Failure in Patients with Acute Myocardial Infarction","fulltext":[{"header":"Introduction","content":"\u003cp\u003eHeart failure (HF) is a complex clinical syndrome resulting from structural or functional cardiac abnormalities that impair ventricular filling or ejection. Key clinical manifestations include dyspnea, fatigue, and fluid retention (e.g., pulmonary and systemic congestion, peripheral edema). As a major global public health challenge, HF affects over 64\u0026nbsp;million individuals worldwide, with an estimated prevalence of 1\u0026ndash;2% \u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e. The condition imposes a substantial economic burden on healthcare systems, driven by high rates of morbidity, hospital readmissions, and mortality, as well as significant impairments in functional capacity and health-related quality of life \u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e. In developed countries, HF accounts for an estimated 1\u0026ndash;2% of total annual healthcare expenditures \u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e. Indirect costs\u0026mdash;such as productivity loss, early retirement, and informal caregiving\u0026mdash;further amplify its socioeconomic impact \u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eIschemic heart disease(IHD) remains the leading cause of HF. Following acute myocardial infarction (AMI), the irreversible loss of cardiomyocytes triggers a reparative fibrotic response. This process activates neurohumoral pathways, leading to ventricular remodeling, cardiac dilation, and impaired contractility. Subsequent hemodynamic overload often results in pulmonary congestion, manifesting as dyspnea and chest tightness, and poses a serious threat to patient survival. Against the backdrop of population aging and rising prevalence of comorbidities such as obesity, both the incidence and prevalence of HF are expected to increase. Missed or delayed diagnosis can worsen clinical outcomes and elevate treatment costs, underscoring the critical importance of early screening and risk stratification. Timely identification of high-risk patients not only saves lives but also enables personalized treatment strategies that can improve quality of life \u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eRecent advances in artificial intelligence (AI) have created new opportunities for enhancing clinical decision-making. Machine learning (ML), a core subset of AI, leverages computational models to identify patterns in large datasets and supports tasks such as diagnosis, risk prediction, and treatment optimization \u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/sup\u003e. ML approaches can be broadly categorized into supervised, unsupervised, and reinforcement learning. In clinical settings, supervised learning algorithms\u0026mdash;such as logistic regression, k-nearest neighbors, decision trees, support vector machines, and naive Bayes\u0026mdash;are frequently used for classification tasks, including binary outcome prediction \u003csup\u003e[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/sup\u003e. For instance, the extreme gradient boosting (XGBoost) algorithm has been employed to predict in-hospital mortality in acute HF \u003csup\u003e[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/sup\u003e, and random forest models have been applied to HF classification \u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e. These methods can uncover complex, non-linear relationships in multidimensional clinical data, offering valuable insights for disease prediction and diagnosis.\u003c/p\u003e\u003cp\u003eIn this study, we aimed to develop and compare six ML algorithms for early prediction of HF following AMI. Using routinely collected clinical data, we constructed prediction models and evaluated their performance to identify the most effective approach. We anticipate that a robust ML-based tool will assist clinicians in stratifying post-AMI risk, facilitating early intervention, and ultimately improving patient outcomes.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003e\u003cstrong\u003eStudy Design and Patient Selection\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis single-center, retrospective study enrolled consecutive patients diagnosed with acute myocardial infarction (AMI) at the Second Affiliated Hospital of Xuzhou Medical University between June 1, 2022, and December 31, 2024.\u003c/p\u003e\n\u003cp\u003ePatients were stratified into two groups based on the occurrence of in-hospital heart failure (HF): the HF group and the non-HF group. \u0026nbsp;The diagnosis of AMI (including ST-segment elevation MI [STEMI] and non-ST-segment elevation MI [NSTEMI]) was established upon admission according to contemporary guidelines for acute coronary syndrome\u0026nbsp;\u003csup\u003e[11]\u003c/sup\u003e. \u0026nbsp;Similarly, HF events during hospitalization were identified based on the latest clinical guidelines\u003csup\u003e\u0026nbsp;\u003c/sup\u003e\u003csup\u003e[12]\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInclusion and Exclusion Criteria\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInclusion criteria were as follows: (1) age \u0026ge;18 years; (2) initial diagnosis of AMI at admission; and (3) no evidence of HF at the time of admission.\u003c/p\u003e\n\u003cp\u003eExclusion criteria comprised: (1) HF attributable to valvular heart disease or cardiomyopathy; (2) non-obstructive myocardial infarction; (3) concomitant rheumatic or autoimmune diseases; (4) history of malignant tumor; and (5) in-hospital mortality.\u003c/p\u003e\n\u003cp\u003eThe study protocol was approved by the Ethics Committee of the Second Affiliated Hospital of Xuzhou Medical University (Approval No. 2024071501), which waived the requirement for informed consent due to the retrospective nature of the analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Collection and Preprocessing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDemographic, clinical, laboratory, and echocardiographic data were extracted from the hospital\u0026apos;s electronic medical record system for all eligible participants and linked to a unique admission identifier. \u0026nbsp; Data preprocessing involved cleaning and standardization: records with extensive missing data (\u0026gt;30%) were excluded, while remaining missing values in continuous variables were imputed using the mean or median after assessing normality with the Shapiro-Wilk test. \u0026nbsp; The robustness of this imputation was evaluated by comparing model performance pre- and post-imputation. \u0026nbsp; Finally, all laboratory values were converted to international standard units to ensure consistency.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eML Algorithms and Model Development\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo develop the prediction model, a two-step feature selection process was employed. \u0026nbsp;First, Least Absolute Shrinkage and Selection Operator (LASSO) regression with 5-fold cross-validation was applied to the initial 61 variables, identifying 24 non-zero coefficients as potentially significant predictors. \u0026nbsp;These 24 variables were subsequently subjected to a multivariate logistic regression analysis, which further refined the feature set to 8 statistically significant indicators. \u0026nbsp;These final 8 features were used as inputs for the following six machine learning algorithms: backpropagation neural network (BPNN), decision tree (DT), logistic regression (LR), random forest (RF), support vector machine (SVM), and XGBoost. \u0026nbsp;The dataset was randomly partitioned into a training set (70%) for model construction and a validation set (30%) for performance evaluation. \u0026nbsp;The models were trained on the training set, and their predictive performance was assessed on the validation set. \u0026nbsp;Furthermore, 5-fold cross-validation was implemented during the training phase of each algorithm to optimize hyperparameters and ensure model robustness.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData management and statistical analyses were performed using Excel 2024 and SPSS software (version 23.0). \u0026nbsp;Continuous variables were assessed for normality using the Shapiro\u0026ndash;Wilk test. \u0026nbsp;Normally distributed data are presented as mean \u0026plusmn; standard deviation and were compared between groups using the independent samples t-test. \u0026nbsp;Non-normally distributed data are expressed as median (interquartile range, IQR) and were compared using the Mann\u0026ndash;Whitney U test. \u0026nbsp; Categorical variables are summarized as frequencies (percentages) and were compared using Pearson\u0026rsquo;s chi-square test or Fisher\u0026rsquo;s exact test, as appropriate. \u0026nbsp;The predictive performance of the developed models was evaluated by the area under the receiver operating characteristic curve (AUC). \u0026nbsp;The clinical utility of the optimal model was assessed using decision curve analysis (DCA). \u0026nbsp;A two-sided p-value of less than 0.05 was considered statistically significant.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eStudy Population and Baseline Characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA total of 664 patients with AMI were included in the final analysis. Based on the occurrence of in-hospital HF, 421 patients (63.4%) were classified into the HF group and 223 (33.6%) into the non-HF group. The baseline characteristics of the two groups are summarized in Table 1. Significant differences were observed in age, MI type (coded as 1 for STEMI and 2 for NSTEMI), history of diabetes, and sex between the two groups. Consequently, these variables were included as candidate features for subsequent model development. In contrast, no significant differences were found in the prevalence of hypertension or history of alcohol consumption between the groups.\u003c/p\u003e\n\u003cp\u003eTable 1：Demographic characteristics of the participants.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"497\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 66px;\"\u003e\n \u003cp\u003eVariable\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 69px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 234px;\"\u003e\n \u003cp\u003eGroup\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 64px;\"\u003e\n \u003cp\u003eZ/\u0026chi;\u0026sup2;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 64px;\"\u003e\n \u003cp\u003ep\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003eNHF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003eHF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 64px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 64px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e61.0（52.5, 70.0）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e71.0（61.0, 79.0）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 64px;\"\u003e\n \u003cp\u003e-7.9692\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 64px;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 66px;\"\u003e\n \u003cp\u003eMI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003eSTEMI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e116(18.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e264(41.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 64px;\"\u003e\n \u003cp\u003e6.887\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.09\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003eNSTEMI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e107(16.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e157(24.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 66px;\"\u003e\n \u003cp\u003eHypertension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003eNO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e107(16.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e180(28.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 64px;\"\u003e\n \u003cp\u003e1.612\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.212\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003eYES\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e116(18.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e241(37.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 66px;\"\u003e\n \u003cp\u003eDiabetes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003eNO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e152(23.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e249(38.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 64px;\"\u003e\n \u003cp\u003e5.044\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.026\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003eYES\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e71(11.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e172(26.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 66px;\"\u003e\n \u003cp\u003eSex\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e183(28.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e293(45.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 64px;\"\u003e\n \u003cp\u003e11.75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e40(6.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e128(19.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 66px;\"\u003e\n \u003cp\u003eSmoke\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003eNO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e108(16.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e243(37.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 64px;\"\u003e\n \u003cp\u003e5.073\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.025\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003eYES\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e115(17.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e178(27.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 66px;\"\u003e\n \u003cp\u003eAlcohol\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003eNO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e176(27.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e344(53.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.728\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.402\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 69px;\"\u003e\n \u003cp\u003eYES\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 116px;\"\u003e\n \u003cp\u003e47(7.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e77(12.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eNHF: non-Heart failure group; HF: Heart failure group；MI：type of AMI.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel Construction and Feature Selection\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA total of 61 variables were initially extracted from patient baseline characteristics, initial laboratory tests, and echocardiographic examinations as potential predictors for model development. Missing quantitative data were imputed using the median value of each variable. Feature selection was performed in two stages: first, LASSO regression with 5-fold cross-validation was applied, which reduced the feature set to 24 non-redundant variables (Figures 1 and 2). These variables included type of myocardial infarction (MI), diabetes, sex, age, diastolic blood pressure (DBP), neutrophil count (NEUT), albumin (ALB), globulin (GLB), aspartate aminotransferase (AST), alkaline phosphatase (ALP), blood urea nitrogen (BUN), serum creatinine (Scr), glucose (GLU), high-density lipoprotein cholesterol (HDLC), lipoprotein(a) (Lpa), residual cholesterol (RC), high-sensitivity C-reactive protein (hsCRP), prothrombin time (PT), high-sensitivity cardiac troponin I (hscTnI), creatine kinase-MB (CKMB), left atrial anterior\u0026ndash;posterior diameter (LAAPD), left ventricular anterior\u0026ndash;posterior diameter (LVAPD), left ventricular posterior wall thickness (LVPW), and right ventricular anterior\u0026ndash;posterior diameter (RVAPD). Subsequently, logistic regression was used to further refine the predictor set, identifying the eight most clinically meaningful features: MI type, age, ALB, AST, BUN, GLU, PT, hsCRP, and LVAPD, which were used to construct the final prediction model for post-infarction heart failure (Figure 3).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel Selection and Validation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe cohort of 664 AMI patients was randomly divided into a training set (70%) and a validation set (30%). \u0026nbsp;Six machine learning algorithms\u0026mdash;BPNN, DT, LR, RF, SVM, and XGBoost\u0026mdash;were trained and evaluated using these datasets. \u0026nbsp;The sensitivity, specificity, and AUC values for both the training and validation sets are summarized in Table 2a, 2b and illustrated in Figures 4a and 4b.\u003c/p\u003e\n\u003cp\u003eTable 2a: Diagnostic efficacy of Six classifiers（Train set）\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"507\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 96px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSPE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSEN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eACC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRecall\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eF1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAUC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 96px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eBPNN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.503\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.913\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.765\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.765\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.913\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.832\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.830\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 96px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDT\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.795\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.878\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.849\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.885\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.878\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.881\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.886\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 96px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLogistic\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.609\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.854\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.766\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.791\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.854\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.821\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.846\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 96px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRF\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 96px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSVM\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.826\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.948\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.904\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.907\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.948\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.927\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.868\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 96px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eXGBoost\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.702\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.896\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.826\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.843\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.896\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.869\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.879\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 2b: Diagnostic efficacy of Six classifiers（Test set）\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"519\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSPE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSEN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eACC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRecall\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eF1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAUC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eBPNN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.450\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.910\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.767\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.786\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.910\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.843\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.794\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDT\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.565\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.759\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.697\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.789\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.759\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.774\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.720\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLogistic\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.710\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.767\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.749\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.850\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.767\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.806\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.795\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRF\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.629\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.707\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.682\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.803\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.707\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.752\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.786\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSVM\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.565\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.865\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.769\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.810\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.865\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.837\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.776\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eXGBoost\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.677\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.820\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.774\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.845\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.820\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.832\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 64px;\"\u003e\n \u003cp\u003e0.818\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eSPE = True Negative/( True Negative + False Positive); SEN = True Positive /( True Positive + False Negative); ACC = (True Positive + True Negative)/( Positive + Negative); PR = True Positive/( True Positive + False Positive); Recall = True Positive /( True Positive + False Negative); F1= 2*Precision*Recal / (Precision + Recal)\u003c/p\u003e\n\u003cp\u003eBased on a comprehensive assessment of sensitivity, specificity, and AUC\u0026mdash;while also accounting for potential overfitting or underfitting\u0026mdash;the XGBoost model was identified as the best-performing predictor. \u0026nbsp; Feature importance analysis within the XGBoost model revealed the following predictors in descending order of contribution: hs-CRP, age, AST, LVAPD, BUN, albumin, glucose, and MI type (Figure 5). \u0026nbsp; DCA further demonstrated that the XGBoost model provided substantial net clinical benefit across both the training and validation sets (Figure 6).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe growing global burden of AMI is closely linked to contemporary shifts in lifestyle and increasing exposure to psychosocial stressors, contributing to its rising incidence. As a leading cause of HF, AMI imposes a substantial economic and healthcare burden worldwide. This challenge is further compounded by population aging, which has been widely recognized as an independent risk factor for HF. Epidemiological studies indicate that adults over 60 years of age face a markedly elevated risk of developing HF compared to younger individuals \u003csup\u003e[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]\u003c/sup\u003e, and aging is strongly associated with the pathogenesis and progression of the disease \u003csup\u003e[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]\u003c/sup\u003e. In our study, the mean age of both the HF and non-HF groups exceeded 60 years, with patients in the HF group being significantly older. Those affected by HF frequently experience recurrent hospitalizations, long-term pharmacotherapy, and severely impaired functional capacity, leading to considerable loss of productivity and diminished quality of life. These factors underscore the critical importance of early prediction of HF following AMI \u003csup\u003e[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eThe widespread adoption of electronic medical record (EMR) systems in healthcare institutions has generated vast repositories of clinical data, which\u0026mdash;while requiring rigorous information security measures\u0026mdash;provide a valuable foundation for applying ML in clinical research. Conventional statistical methods often face limitations in handling high-dimensional and complex multimodal health data, which can restrict the predictive performance of resulting models. In contrast, ML techniques offer a powerful alternative by leveraging computational algorithms to automatically learn patterns from data without relying solely on pre-specified hypotheses.\u003c/p\u003e\u003cp\u003eML models are trained by establishing mappings between input features and output labels, enabling the prediction of outcomes through classification or regression mechanisms \u003csup\u003e[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]\u003c/sup\u003e. These methods can capture both linear and non-linear relationships, but their performance heavily depends on appropriate feature selection and model design. Consequently, model development is an iterative, empirically driven process that involves continuous adjustment of algorithms and hyperparameters based on performance feedback \u003csup\u003e[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eCurrent diagnosis of heart failure relies on comprehensive assessment of patient symptoms, physical signs, laboratory tests, imaging findings, and medical history. In this study, we developed and validated an XGBoost-based machine learning model using routinely collected clinical data\u0026mdash;including baseline characteristics, laboratory results, and echocardiographic parameters\u0026mdash;to predict early-onset heart failure following AMI.\u003c/p\u003e\u003cp\u003eFeature importance analysis revealed that hsCRP was the most influential predictor in our model, with significantly higher levels observed in the HF group compared to the non-HF group. As a sensitive inflammatory marker, hsCRP reflects underlying chronic inflammation, which contributes to atherosclerotic plaque vulnerability and progression of coronary artery disease. Elevated hsCRP levels have been documented in patients with non-ST-segment elevation acute coronary syndrome (NSTEMI-ACS) and have shown predictive value for new-onset HF \u003csup\u003e[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]\u003c/sup\u003e. Similarly, elevated hsCRP in stable ACS patients has been associated with a two-fold increased risk of new or worsening HF within two years \u003csup\u003e[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eAST and ALB were also identified as significant predictors in our model. Elevated AST following AMI may result from oxidative stress, cardiomyocyte necrosis, or ischemia-reperfusion injury, although direct evidence linking AST to HF progression remains limited \u003csup\u003e[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]\u003c/sup\u003e. Thus, AST may serve as an auxiliary predictive marker in this context. Hypoalbuminemia, frequently observed in HF patients (affecting up to 40% in some cohorts \u003csup\u003e[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/sup\u003e), is associated with higher NYHA functional class, reduced renal function, and increased comorbidity burden \u003csup\u003e[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]\u003c/sup\u003e. Higher serum albumin levels have been correlated with improved outcomes in HF with preserved ejection fraction (HFpEF) \u003csup\u003e[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]\u003c/sup\u003e, supporting ALB\u0026rsquo;s relevance as a prognostic indicator.\u003c/p\u003e\u003cp\u003eLVAPD also contributed to model performance. Following AMI, compensatory ventricular dilation may initially maintain stroke volume but can progress to adverse remodeling and functional deterioration. Previous studies have linked LVAPD to adverse outcomes in both HFpEF and HF with reduced ejection fraction (HFrEF) \u003csup\u003e[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]\u003c/sup\u003e, consistent with our findings.\u003c/p\u003e\u003cp\u003eBUN emerged as another relevant predictor. In HF, reduced cardiac output activates neurohormonal systems such as the sympathetic nervous system and renin\u0026ndash;angiotensin\u0026ndash;aldosterone system, promoting renal sodium retention and elevated BUN. These pathways are associated with worse cardiac function and prognosis \u003csup\u003e[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]\u003c/sup\u003e, and BUN has been consistently linked to HF outcomes \u003csup\u003e[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eDiabetes and elevated admission blood glucose are well-established risk factors for HF, with diabetic patients facing a two-fold increase in HF risk in men and up to five-fold in women after age adjustment \u003csup\u003e[\u003cspan additionalcitationids=\"CR33 CR34\" citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]\u003c/sup\u003e. Hyperglycemia at admission may reflect impaired myocardial energy metabolism and oxidative stress, further aggravating cardiac injury \u003csup\u003e[\u003cspan additionalcitationids=\"CR37\" citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]\u003c/sup\u003e and contributing to HF onset \u003csup\u003e[\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eFinally, MI type (STEMI vs. NSTEMI) was incorporated into the model. STEMI, typically resulting from transmural ischemia, often leads to more extensive myocardial damage and higher HF incidence compared to NSTEMI \u003csup\u003e[\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]\u003c/sup\u003e. Although MI type had the lowest feature importance in our model, it still provided complementary predictive value, consistent with prior studies \u003csup\u003e[\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003eStrengths and Limitations\u003c/h2\u003e\u003cp\u003eThis study possesses several notable strengths. First, the use of initial patient data obtained upon admission enabled the early identification of key predictors, creating a valuable time window for preventive strategies against post-infarction heart failure. Through a rigorous two-step feature selection process\u0026mdash;incorporating LASSO regression followed by logistic regression\u0026mdash;we distilled 8 clinically meaningful predictors from an initial set of 61 variables, achieving an optimal balance between model performance and simplicity. Furthermore, the model relies exclusively on routinely collected electronic medical record data, requiring no additional costs or specialized examinations, which enhances its potential for real-world implementation. Finally, by systematically comparing six machine learning algorithms and selecting XGBoost based on its AUC performance, robustness against overfitting and underfitting, and net clinical benefit, this study offers a reliable prediction tool.\u003c/p\u003e\u003cp\u003eSeveral limitations should also be acknowledged. As a single-center retrospective analysis, the findings may reflect local patient characteristics and clinical practices, and missing data, though handled, may introduce bias. Future multicenter prospective studies are needed to validate and generalize the results. The relatively limited sample size may also constrain the model\u0026rsquo;s predictive power; expanding the cohort in subsequent research could improve accuracy and stability. In addition, this study incorporated basic demographic, laboratory, and echocardiographic variables but did not include electrocardiographic, coronary CTA, or angiographic data. Integrating these parameters in future models could further enhance predictive comprehensiveness.\u003c/p\u003e\u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn this study, we developed and compared six machine learning models for predicting in-hospital heart failure following acute myocardial infarction. The XGBoost algorithm demonstrated superior performance, forming the basis of a final prediction model that incorporates eight key clinical features: hs-CRP, age, AST, LVAPD, BUN, albumin, glucose, and MI type. This model exhibits strong predictive capability and clinical utility, offering a reliable, data-driven tool for early risk stratification that may support clinical decision-making and improve patient management.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eAI = artificial intelligence\u003c/p\u003e\n\u003cp\u003eALB = albumin\u003c/p\u003e\n\u003cp\u003eALP = alkaline phosphatase\u003c/p\u003e\n\u003cp\u003eAMI = acute myocardial infarction\u003c/p\u003e\n\u003cp\u003eAST = aspartate aminotransferase\u003c/p\u003e\n\u003cp\u003eAUC = an area under the ROC curve\u003c/p\u003e\n\u003cp\u003eBPNN = backpropagation neural network\u003c/p\u003e\n\u003cp\u003eBUN = blood urea nitrogen\u003c/p\u003e\n\u003cp\u003eCKMB = creatine kinase-MB\u003c/p\u003e\n\u003cp\u003eDBP = diastolic blood pressure\u003c/p\u003e\n\u003cp\u003eDCA = ecision curve analysis\u003c/p\u003e\n\u003cp\u003eDT = decision tree\u003c/p\u003e\n\u003cp\u003eEMR = electronic medical record\u003c/p\u003e\n\u003cp\u003eGLB = globulin\u003c/p\u003e\n\u003cp\u003eGLU = glucose\u003c/p\u003e\n\u003cp\u003eHDLC = high-density lipoprotein cholesterol\u003c/p\u003e\n\u003cp\u003eHF = heart failure\u003c/p\u003e\n\u003cp\u003eHFpEF = heart failure with preserved ejection fraction\u003c/p\u003e\n\u003cp\u003eHFrEF = heart failure with reduced ejection fraction\u003c/p\u003e\n\u003cp\u003ehsCRP = high-sensitivity C reactive protein\u003c/p\u003e\n\u003cp\u003ehscTnI = high-sensitivity cardiac troponin I\u003c/p\u003e\n\u003cp\u003eIHD = Ischemic heart disease\u003c/p\u003e\n\u003cp\u003eLAAPD = left atrial anterior\u0026ndash;posterior diameter\u003c/p\u003e\n\u003cp\u003eLASSO Regression = Least Absolute Shrinkage and Selection Operator regression\u003c/p\u003e\n\u003cp\u003eLpa = lipoprotein\u003c/p\u003e\n\u003cp\u003eLR = logistic regression\u003c/p\u003e\n\u003cp\u003eLVAPD = left ventricular anterior\u0026ndash;posterior diameter\u003c/p\u003e\n\u003cp\u003eLVPW = left ventricular posterior wall thickness\u003c/p\u003e\n\u003cp\u003eML = type of AMI\u003c/p\u003e\n\u003cp\u003eNEUT = neutrophil count\u003c/p\u003e\n\u003cp\u003eNSTEMI = non-ST-segment elevation MI\u003c/p\u003e\n\u003cp\u003eNSTEMI-ACS = non-ST-segment elevation acute coronary syndrome\u003c/p\u003e\n\u003cp\u003ePT = prothrombin time\u003c/p\u003e\n\u003cp\u003eRC = residual cholesterol\u003c/p\u003e\n\u003cp\u003eRF = random forest\u003c/p\u003e\n\u003cp\u003eROC = receiver operating characteristic\u003c/p\u003e\n\u003cp\u003eRVAPD = right ventricular anterior\u0026ndash;posterior diameter\u003c/p\u003e\n\u003cp\u003eScr = serum creatinine\u003c/p\u003e\n\u003cp\u003eSTEMI = ST-segment elevation MI\u003c/p\u003e\n\u003cp\u003eSVM = support vector machine\u003c/p\u003e\n\u003cp\u003eXGBoost = extreme gradient boosting\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNone.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eShuang Liu conceptualized and designed the study. Xuejin Chen and Yanan Hu collected the data. Shuang Liu, Xuejin Chen and Jingjing Jin performed the analyses and produced the results. Shuang Liu, Xuejin Chen and Yanan Hu analysed the results and wrote the manuscript. Chunmei Qi and Ji Hao provided funding and reviewed the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNone.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used and analysed during the current study available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA statement to confirm that all methods were carried out in accordance with relevant guidelines and regulations. The study protocol was approved by the Ethics Committee of the Second Affiliated Hospital of Xuzhou Medical University (Approval No. 2024071501), which waived the requirement for informed consent due to the retrospective nature of the analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.\u003c/p\u003e"},{"header":"References","content":"\u003col class=\"decimal_type\"\u003e\n \u003cli\u003eSavarese G, Becher PM, Lund LH, et al. Global burden of heart failure: a comprehensive and updated review of epidemiology. Cardiovasc Res. 2023 Jan 18;118(17):3272-3287. doi: 10.1093/cvr/cvac013.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eKhan MS, Shahid I, Bennis A, et al. Global epidemiology of heart failure. Nat Rev Cardiol. 2024 Oct;21(10):717-734. doi: 10.1038/s41569-024-01046-6.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eHessel FP. Overview of the socio-economic consequences of heart failure. Cardiovasc Diagn Ther. 2021 Feb;11(1):254-262. doi: 10.21037/cdt-20-291.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eOng SC, Low JZ, Yew WY, et al. Cost analysis of chronic heart failure management in Malaysia: A multi-centred retrospective study. Front Cardiovasc Med. 2022 Nov 2;9:971592. doi: 10.3389/fcvm.2022.971592.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eMohammad MA, Olesen KKW, Koul S, et al. Development and validation of an artif-icial neural network algorithm to predict mortality and admission to hospital for heart failure after myocardial infarction: a nationwide population-based study. Lancet Digit Health. 2022 Jan;4(1):e37-e45. doi: 10.1016/S2589-7500(21)00228-4.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eLi F, Wang Y, Xu T, et al. Deep learning-based automated detection for diabetic reti-nopathy and diabetic macular oedema in retinal fundus photographs. Eye (Lond). 2022 Jul;36(7):1433-1441. doi: 10.1038/s41433-021-01552-8.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eSki CF, Thompson DR, Brunner-La Rocca HP. Putting AI at the centre of heart failu-re care. ESC Heart Fail. 2020 Oct;7(5):3257-3258. doi: 10.1002/ehf2.12813.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eYasmin F, Shah SMI, Naeem A, et al. Artificial intelligence in the diagnosis and det-ection of heart failure: the past, present, and future. Rev Cardiovasc Med. 2021 Dec 22;22(4):1095-1113. doi: 10.31083/j.rcm2204121.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eTanaka M, Kohjitani H, Yamamoto E, et al. Development of interpretable machine le-arning models to predict in-hospital prognosis of acute heart failure patients. ESC He-art Fail. 2024 Oct;11(5):2798-2812. doi: 10.1002/ehf2.14834.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eChulde-Fern\u0026aacute;ndez B, Enr\u0026iacute;quez-Ortega D, Guevara C, et al. Classification of Heart Fai-lure Using Machine Learning: A Comparative Study. Life (Basel). 2025 Mar 19;15(3):496. doi: 10.3390/life15030496.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eRao SV, O\u0026apos;Donoghue ML, Ruel M, et al. 2025 ACC/AHA/ACEP/NAEMSP/SCAI Gui-deline for the Management of Patients With Acute Coronary Syndromes: A Report \u0026nbsp;of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2025 Apr;151(13):e771-e862. doi: 10.1161/CIR.0000000000001309.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eChinese Society of Cardiology, Chinese Medical Association; Chinese College of Card-iovascular Physician; Chinese Heart Failure Association of Chinese Medical Doctor Association; Editorial Board of Chinese Journal of Cardiology. [Chinese guidelines for \u0026nbsp;the diagnosis and treatment of heart failure 2024]. Zhonghua Xin Xue Guan Bing \u0026nbsp;Za Zhi. 2024 Mar 24;52(3):235-275. Chinese. doi: 10.3760/cma.j.cn112148-20231101-00405.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eGoyal P, Maurer MS, Roh J. Aging in Heart Failure: Embracing Biology Over Chron-ology: JACC Family Series. JACC Heart Fail. 2024 May;12(5):795-809. doi: 10.1016/j.jchf.2024.02.021. Epub 2024 Apr 8.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eFang Z, Raza U, Song J, et al. Systemic aging fuels heart failure: Molecular mechani-sms and therapeutic avenues. ESC Heart Fail. 2025 Apr;12(2):1059-1080. doi: 10.1002/ehf2.14947. Epub 2024 Jul 22.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eDarvish M, Shakoor A, Feyz L, et al. Heart failure: assessment of the global econom-ic burden. Eur Heart J. 2025 Aug 14;46(31):3069-3078. doi: 10.1093/eurheartj/ehaf323.\u003c/li\u003e\n \u003cli\u003eRoth GA, Forouzanfar MH, Moran AE, et al. Demographic and epidemiologic drivers of global cardiovascular mortality. N Engl J Med. 2015 Apr 2;372(14):1333-41. doi: 10.1056/NEJMoa1406656.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eItchhaporia D. Artificial intelligence in cardiology. Trends Cardiovasc Med. 2022 Jan;32(1):34-41. doi: 10.1016/j.tcm.2020.11.007.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eVandenberk B, Chew DS, Prasana D, Gupta S, Exner DV. Successes and challenges of artificial intelligence in cardiology. Front Digit Health. 2023 Jun 28;5:1201392. doi: 10.3389/fdgth.2023.1201392.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eJames SK, Armstrong P, Barnathan E, et al. Troponin and C-reactive protein have dif-ferent relations to subsequent mortality and myocardial infarction after acute coronary syndrome: a GUSTO-IV substudy. J Am Coll Cardiol. 2003 Mar 19;41(6):916-24. doi: 10.1016/s0735-1097(02)02969-8.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eScirica BM, Cannon CP, Sabatine MS, et al. Concentrations of C-reactive protein and B-type natriuretic peptide 30 days after acute coronary syndromes independently predict hospitalization for heart failure and cardiovascular death. Clin Chem. 2009 Feb;55(2):265-73. doi: 10.1373/clinchem.2008.117192.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eHan JH, Kwak JY, Lee SS, Kim HG, Jeon H, Cha RR. Markedly Elevated Aspartate Aminotransferase from Non-Hepatic Causes. J Clin Med. 2022 Dec 30;12(1):310. doi: 10.3390/jcm12010310.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eMyhre PL, Grupper A, Mebazaa A, et al. Changes in Liver Function Tests, Congestio-n, and Prognosis After Acute Heart Failure: The STRONG-HF Trial. JACC Adv. 2025 Mar;4(3):101607. doi: 10.1016/j.jacadv.2025.101607.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eEl Iskandarani M, El Kurdi B, Murtaza G, Paul TK, Refaat MM. Prognostic role of albumin level in heart failure: A systematic review and meta-analysis. Medicine (Baltimore). 2021 Mar 12;100(10):e24785. doi: 10.1097/MD.0000000000024785.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eBonilla-Palomas JL, G\u0026aacute;mez-L\u0026oacute;pez AL, Moreno-Conde M, et al. Hypoalbuminemia in acute heart failure patients: causes and its impact on hospital and long-term mortality. J Card Fail. 2014 May;20(5):350-8. doi: 10.1016/j.cardfail.2014.01.016.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eDastjerdi P, Mahalleh M, Shayesteh H, et al. Liver biomarkers as predictors of progn-osis in heart failure with preserved ejection fraction: a systematic review and meta-analysis. BMC Cardiovasc Disord. 2025 Apr 2;25(1):244. doi: 10.1186/s12872-025-04647-2.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eOgawa S, Nagatomo Y, Takei M, et al. Impact of Left Ventricular Chamber Size on Outcome in Heart Failure with Preserved Ejection Fraction. Int Heart J. 2022;63(1):62-72. doi: 10.1536/ihj.21-486.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eRowe SJ, Xiang R, Paratz ED, Takeuchi F, La Gerche A. Left ventricular size and heart failure: A cardiac MRI assessment of 38,129 individuals from the UK Biobank. Int J Cardiol. 2025 Jan 15;419:132687. doi: 10.1016/j.ijcard.2024.132687.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eRengo G, Lymperopoulos A, Zincarelli C, et al. Blockade of \u0026beta;-adrenoceptors restores the GRK2-mediated adrenal \u0026alpha;(2) -adrenoceptor-catecholamine production axis in heart failure. Br J Pharmacol. 2012 Aug;166(8):2430-40. doi: 10.1111/j.1476-5381.2012.01972.x.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eBozic J, Kumric M, Ticinovic Kurir T, et al. Catestatin as a Biomarker of Cardiovas-cular Diseases: A Clinical Perspective. Biomedicines. 2021 Nov 25;9(12):1757. doi: 10.3390/biomedicines9121757.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eXu Z, Yue Y, Xu M, et al. The role of cut-off values for creatinine, blood urea nitr-ogen, and uric acid in prognostic assessment of chronic heart failure: a retrospective cohort study. BMC Cardiovasc Disord. 2025 Mar 22;25(1):209. doi: 10.1186/s12872-025-04675-y.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eDuan S, Li Y, Yang P. Predictive value of blood urea nitrogen in heart failure: a sys-tematic review and meta-analysis. Front Cardiovasc Med. 2023 Jul 31;10:1189884. doi: 10.3389/fcvm.2023.1189884.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eNichols GA, Hillier TA, Erbey JR, Brown JB. Congestive heart failure in type 2 dia-betes: prevalence, incidence, and risk factors. Diabetes Care. 2001 Sep;24(9):1614-9. doi: 10.2337/diacare.24.9.1614.\u003c/li\u003e\n \u003cli\u003eDei Cas A, Khan SS, Butler J, et al. Impact of diabetes on epidemiology, treatment, and outcomes of patients with heart failure. JACC Heart Fail. 2015 Feb;3(2):136-45. doi: 10.1016/j.jchf.2014.08.004.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eKannel WB, Hjortland M, Castelli WP. Role of diabetes in congestive heart failure: \u0026nbsp;the Framingham study. Am J Cardiol. 1974 Jul;34(1):29-34. doi: 10.1016/0002-9149(74)90089-7.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eKannel WB, McGee DL. Diabetes and cardiovascular disease. The Framingham study. JAMA. 1979 May 11;241(19):2035-8. doi: 10.1001/jama.241.19.2035.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eMontaigne D, Marechal X, Coisne A, et al. Myocardial contractile dysfunction is asso-ciated with impaired mitochondrial function and dynamics in type 2 diabetic but not in obese patients. Circulation. 2014 Aug 12;130(7):554-64. doi: 10.1161/CIRCULATIONAHA.113.008476.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eLowell BB, Shulman GI. Mitochondrial dysfunction and type 2 diabetes. Science. 2005 Jan 21;307(5708):384-7. doi: 10.1126/science.1104343.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eLiesa M, Shirihai OS. Mitochondrial dynamics in the regulation of nutrient utilization and energy expenditure. Cell Metab. 2013 Apr 2;17(4):491-506. doi: 10.1016/j.cmet.2013.03.002.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eRitsinger V, Hagstr\u0026ouml;m E, Lagerqvist B, Norhammar A. Admission Glucose Levels an-d Associated Risk for Heart Failure After Myocardial Infarction in Patients Without Diabetes. J Am Heart Assoc. 2021 Nov 16;10(22):e022667. doi: 10.1161/JAHA.121.022667.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eButler J, Hammonds K, Talha KM, et al. Incident heart failure and recurrent coronar-y events following acute myocardial infarction. Eur Heart J. 2025 Apr 22;46(16):1540-1550. doi: 10.1093/eurheartj/ehae885.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eLi X, Shang C, Xu C, Wang Y, Xu J, Zhou Q. Development and comparison of ma-chine learning-based models for predicting heart failure after acute myocardial infarction. BMC Med Inform Decis Mak. 2023 Aug 24;23(1):165. doi: 10.1186/s12911-023-02240-1.\u0026nbsp;\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Myocardial infarction, Heart failure, Machine learning, Predict","lastPublishedDoi":"10.21203/rs.3.rs-7859492/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7859492/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eAims\u003c/h2\u003e\u003cp\u003eHeart failure (HF) remains a frequent and burdensome complication of acute myocardial infarction (AMI), posing a substantial challenge to global healthcare systems. This study aimed to develop and compare six machine learning (ML) algorithms to identify the optimal model for the early prediction of HF following AMI.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eWe retrospectively enrolled patients admitted for AMI at the Second Affiliated Hospital of Xuzhou Medical University between June 1, 2022, and December 31, 2024. Participants were categorized into HF and non-HF groups based on the occurrence of in-hospital heart failure. The cohort was randomly split into a training set (70%) and a validation set (30%) for model development and internal validation, respectively. Model performance was assessed using the receiver operating characteristic (ROC) curve, and clinical utility was evaluated via decision curve analysis (DCA).\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eAmong the six ML models evaluated, the extreme gradient boosting (XGBoost) algorithm demonstrated superior predictive performance. Feature importance analysis within the XGBoost model identified the top eight predictors, in descending order of contribution: high-sensitivity C reactive protein (hsCRP), age, aspartate aminotransferase (AST), left ventricular anterior\u0026ndash;posterior diameter (LVAPD), blood urea nitrogen (BUN), albumin (ALB), glucose (GLU), and myocardial infarction type(MI). In the validation cohort, the model achieved an area under the ROC curve (AUC) of 0.818. DCA further confirmed its favourable net clinical benefit.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e\u003cp\u003eAn XGBoost model incorporating eight readily available clinical features was developed and validated for the early prediction of HF after AMI, showing promising discriminative ability and clinical utility. This tool may assist clinicians in stratifying risk and guiding early intervention.\u003c/p\u003e","manuscriptTitle":"A Machine Learning Model for Predicting the Occurrence of Early Heart Failure in Patients with Acute Myocardial Infarction","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-28 08:53:47","doi":"10.21203/rs.3.rs-7859492/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-01-20T08:29:44+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-01-15T12:37:06+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"154876088460185266993425481600146393976","date":"2026-01-15T11:52:54+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-01-14T09:48:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"28467572548388913740513983529688880358","date":"2026-01-14T09:32:26+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-11-19T13:41:10+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-10-15T05:20:40+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-10-15T05:20:15+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Informatics and Decision Making","date":"2025-10-14T13:54:53+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"c1df1fb1-e448-4969-8ebf-60f668537c0f","owner":[],"postedDate":"November 28th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-02-12T07:23:56+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-28 08:53:47","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7859492","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7859492","identity":"rs-7859492","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}