Journal of Research and Health

en Critical Appraisal of Artificial Intelligence-Based Prediction Models for Mortality in ST-Elevation Myocardial Infarction: A Systematic Review Using CHARMS and PROBAST Frameworks ● Artificial Intelligence مقاله مروری Review Article Background: Machine learning (ML) and artificial intelligence (AI) have shown promising potential for predicting mortality in patients with ST-elevation myocardial infarction. However, methodological quality, risk of bias, and clinical relevance of these models have not been systematically evaluated. This gap may hinder their adoption in clinical practice. This review exclusively examines AI-driven prediction models for mortality or mortality-inclusive MACE in STEMI patients, using the CHARMS and PROBAST frameworks. Methods: A comprehensive search was conducted across PubMed, Embase, Scopus, and Web of Science for studies published between 2014 and 2024. Seventeen eligible studies were identified and reviewed for methodological rigor, risk of bias, and clinical applicability. Two reviewers independently conducted data extraction and critical appraisal. Results: There was considerable variation in how models were developed, validated, and reported. Although many reported high predictive accuracy, 94% of the studies had a high risk of bias—mainly due to issues with participant selection, unclear definitions of predictors, and limited external validation. Moreover, the use of narrow inclusion criteria and lack of representative samples further limited the real-world applicability of these models. To address these limitations, we proposed potential refinements to existing appraisal checklists that can be utilized during model development for STEMI patients. Conclusion: AI models for STEMI mortality remain exploratory research tools, not deployable technologies. Despite high discrimination, substantial evidence gaps persist regarding calibration, validation, transportability, and clinical utility. Future studies must prioritize prospective multicenter validation and net benefit assessment before these models can be recommended for routine use in clinical practice.   Artificial Intelligence, ST-Elevation Myocardial Infarction (STEMI), Prediction Models, Machine Learning. 4 4 http://jrh.gmu.ac.ir/browse.php?a_code=A-10-2655-2&slc_lang=en&sid=1 Nafiseh Montazeri nafis_mont@yahoo.com 3256286135 No Cardiovascular Research Center, Health Policy and Promotion Institute, Kermanshah University of Medical Sciences, Kermanshah, Iran. Amirhossein kazemi Amirhossinkazmii@gmail.com 3360318374 No Student Research Committee, Kermanshah University of Medical Sciences, Kermanshah, Iran. Parto Ghahramani partoghahramani@gmail.com 3330405279 No Student Research Committee, Kermanshah University of Medical Sciences, Kermanshah, Iran. Ghobad Ramezani ramaznighobad@gmail.com 3230867939 No Education Development Center, Kermanshah University of Medical Sciences, Kermanshah, Iran. Farideh Akbarzadeh akbarzadehfar@ymail.com 3257379161 No Central Library Management of Scientific Resources and Information, Kermanshah University of Medical Sciences, Kermanshah, Iran. Mehdi Naderi m.naderi51@yahoo.com 3255422787 No Raya Salamat Nozohour company, Kermanshah University of Medical Sciences Kermanshah, I.R. Iran. Masoumeh Abbasi masoumeh.abbasi@kums.ac.ir 4559870098 Yes Department of Health Information Technology, School of Allied Medical Sciences, Kermanshah University of Medical Science, Kermanshah, Iran.