A Prediction Model for the Early Detection of Angina Pectoris by Using Machine Learning Approaches

Abstract

Angina Pectoris remains one of the leading causes of mortality worldwide, and early detection is essential for effective intervention. Recent advancements in data mining and machine learning have enabled the development of intelligent diagnostic systems capable of analyzing large volumes of clinical data. This study proposes an enhanced prediction model for early detection of Angina Pectoris using multiple data mining approaches, including Decision Tree, Random Forest, Naïve Bayes, CN2 Rule Inducer, Support Vector Machine, and a Majority Voting baseline. The model employs a systematic pipeline comprising data cleaning, normalization, feature selection, classifier optimization, and cross-validation. Experiments conducted on a benchmark Angina Pectoris dataset demonstrate that the combined use of optimized classifiers significantly improves predictive performance. Among the evaluated algorithms, Naïve Bayes and Random Forest achieve the highest accuracy and AUC, indicating their suitability for clinical decision support. The proposed framework can assist healthcare professionals by providing automated, reliable risk assessment and can be integrated into practical diagnostic systems for continuous model improvement. of the patient's data and provide a qualitative assessment of the patient's risk of Angina Pectoris. A natural ordinal number between 0 and 4, where 0 indicates that the patient is healthy and 4 indicates that there is a substantial possibility that he is, is used to express this probability. If the classifier can correctly forecast the patient's state of health in 85% of the cases or if its accuracy in absolute terms surpasses 85%, the findings will be considered remarkable.