Cardiovascular Health Analysis: Machine Learning and Explainable AI to Predict Heart Attacks

Abstract

A heart attack, medically termed myocardial infarction, occurs due to a blockage of blood flow to the heart, resulting in tissue damage. To enable early diagnosis, machine learning models can play a pivotal role in analyzing data related to heart attacks. This advancement significantly improves patient outcomes and enhances the overall efficiency of healthcare systems. The objective of this study is to reduce heart attack mortality by enabling early detection and risk prediction through advanced machine learning techniques, thereby improving patient outcomes globally and reducing the burden of cardiac diseases. This paper employs statistical methods to predict the occurrence of heart attacks. Based on the outcomes of hypothesis testing, machine learning models such as Random Forest and XGBoost were trained. For XGBoost, the accuracy rate was 98.68%, the precision rate was 98.45%, the recall (sensitivity) rate was 98.71%, and the F1-score was 98.58%. Furthermore, explainable AI techniques such as SHAP (SHapley Additive exPlanations) were utilized to identify the influencing factors in the machine learning model. In conclusion, certain cardiovascular indicators, such as systolic blood pressure, differ significantly between patients who have experienced heart attacks and those who have not. Understanding heart attack risks through cardiovascular analysis and AI models advances personalized medicine and early intervention strategies.

Ähnliche Arbeiten