Predicting acute myocardial infarction from haematological markers utilizing machine learning and explainable artificial intelligence

Myocardial infarction (MI) is the leading cause of human death globally. It occurs when a blockage in an artery prevents blood and oxygen from reaching the heart muscle, causing tissues in the heart muscle to die. This leads to a necessity to develop a method to diagnose MI's early, preventing further complications such as irregular heart rhythm, heart failure or even cardiac arrest. This research aims to develop a more accurate machine learning (ML) model to help predict acute myocardial infarction (AMI) with a greater degree of accuracy without invasive procedures using additional explainable artificial intelligence (XAI) techniques which will help medical practitioners to better diagnose AMI more precisely. According to the results, the random forest classifier model gave the highest accuracy of 86%. XAI techniques were used to visualize the data and results, and determined white blood cell (WBC) count to be the most crucial feature in classification, followed by neutrophil (NEU) count, neutrophil-lymphocyte (NEU/LY) ratio, platelet width of distribution (PDW) values and basophil (BA) counts. The developed model can help medical practitioners make a more accurate early diagnosis of AMI using readily available hematological parameters, enabling practitioners to provide superior care to a diverse range of individuals.