{"title":"心肌梗死预后中的人工智能工具:评估机器学习和深度学习模型的性能。","authors":"Cyntia Szymańska, Artur Baszko","doi":"10.2174/011573403X380162250706094642","DOIUrl":null,"url":null,"abstract":"<p><p>In clinical practice, mortality risk assessment in patients with myocardial infarction often relies on scales such as GRACE and TIMI. However, these scales were developed based on cohorts assembled many years ago. Since then, numerous changes have occurred, ranging from shifts in MI patient profiles to the introduction of new antiplatelet medications and the adoption of more restrictive lipid therapy targets. To address this issue, researchers are working to develop new stratification tools. Artificial intelligence (AI), which finds applications in nearly every area of medicine, also presents solutions to this problem. This review includes sixteen papers that contain machine learning and deep learning models used to prognosticate mortality risk at different points. Machine learning (ML) models, such as random forest, gradient boosting, and support vector machines, have demonstrated good to excellent performance. However, no single algorithm appears to be top-performing. Although artificial neural networks are considered one of the most promising algorithms, they do not invariably outperform other ML methods. The adaptability of AI models to various scenarios and their ability to handle complex datasets reassures us of their potential in cardiology. Concerning variables that influence the risk of mortality, most are well-established factors, such as age, left-ventricular ejection fraction, lipid parameters, and B-type natriuretic peptide. Additionally, less apparent indicators include platelet parameters, neutrophil count, and blood urea nitrogen. In conclusion, utilizing AI-based models in myocardial infarction risk stratification presents a significant opportunity to develop effective and tailored tools.</p>","PeriodicalId":10832,"journal":{"name":"Current Cardiology Reviews","volume":" ","pages":""},"PeriodicalIF":2.2000,"publicationDate":"2025-07-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Artificial Intelligence Tools in Myocardial Infarction Prognosis: Evaluating the Performance of Machine Learning and Deep Learning Models.\",\"authors\":\"Cyntia Szymańska, Artur Baszko\",\"doi\":\"10.2174/011573403X380162250706094642\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><p>In clinical practice, mortality risk assessment in patients with myocardial infarction often relies on scales such as GRACE and TIMI. However, these scales were developed based on cohorts assembled many years ago. Since then, numerous changes have occurred, ranging from shifts in MI patient profiles to the introduction of new antiplatelet medications and the adoption of more restrictive lipid therapy targets. To address this issue, researchers are working to develop new stratification tools. Artificial intelligence (AI), which finds applications in nearly every area of medicine, also presents solutions to this problem. This review includes sixteen papers that contain machine learning and deep learning models used to prognosticate mortality risk at different points. Machine learning (ML) models, such as random forest, gradient boosting, and support vector machines, have demonstrated good to excellent performance. However, no single algorithm appears to be top-performing. Although artificial neural networks are considered one of the most promising algorithms, they do not invariably outperform other ML methods. The adaptability of AI models to various scenarios and their ability to handle complex datasets reassures us of their potential in cardiology. Concerning variables that influence the risk of mortality, most are well-established factors, such as age, left-ventricular ejection fraction, lipid parameters, and B-type natriuretic peptide. Additionally, less apparent indicators include platelet parameters, neutrophil count, and blood urea nitrogen. In conclusion, utilizing AI-based models in myocardial infarction risk stratification presents a significant opportunity to develop effective and tailored tools.</p>\",\"PeriodicalId\":10832,\"journal\":{\"name\":\"Current Cardiology Reviews\",\"volume\":\" \",\"pages\":\"\"},\"PeriodicalIF\":2.2000,\"publicationDate\":\"2025-07-11\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Current Cardiology Reviews\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.2174/011573403X380162250706094642\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"CARDIAC & CARDIOVASCULAR SYSTEMS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Current Cardiology Reviews","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.2174/011573403X380162250706094642","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"CARDIAC & CARDIOVASCULAR SYSTEMS","Score":null,"Total":0}
Artificial Intelligence Tools in Myocardial Infarction Prognosis: Evaluating the Performance of Machine Learning and Deep Learning Models.
In clinical practice, mortality risk assessment in patients with myocardial infarction often relies on scales such as GRACE and TIMI. However, these scales were developed based on cohorts assembled many years ago. Since then, numerous changes have occurred, ranging from shifts in MI patient profiles to the introduction of new antiplatelet medications and the adoption of more restrictive lipid therapy targets. To address this issue, researchers are working to develop new stratification tools. Artificial intelligence (AI), which finds applications in nearly every area of medicine, also presents solutions to this problem. This review includes sixteen papers that contain machine learning and deep learning models used to prognosticate mortality risk at different points. Machine learning (ML) models, such as random forest, gradient boosting, and support vector machines, have demonstrated good to excellent performance. However, no single algorithm appears to be top-performing. Although artificial neural networks are considered one of the most promising algorithms, they do not invariably outperform other ML methods. The adaptability of AI models to various scenarios and their ability to handle complex datasets reassures us of their potential in cardiology. Concerning variables that influence the risk of mortality, most are well-established factors, such as age, left-ventricular ejection fraction, lipid parameters, and B-type natriuretic peptide. Additionally, less apparent indicators include platelet parameters, neutrophil count, and blood urea nitrogen. In conclusion, utilizing AI-based models in myocardial infarction risk stratification presents a significant opportunity to develop effective and tailored tools.
期刊介绍:
Current Cardiology Reviews publishes frontier reviews of high quality on all the latest advances on the practical and clinical approach to the diagnosis and treatment of cardiovascular disease. All relevant areas are covered by the journal including arrhythmia, congestive heart failure, cardiomyopathy, congenital heart disease, drugs, methodology, pacing, and preventive cardiology. The journal is essential reading for all researchers and clinicians in cardiology.