Comparative Analysis of Machine Learning Models for Early Heart Disease Diagnosis

Abstract

Heart disease remains among the leading causes of death worldwide, and its early detection ability can be the difference between life and death. In this research, we investigate the capability of machine learning—namely Support Vector Machines (SVM)—to predict the occurrence of heart disease based on regular clinical information. We used the Cleveland Heart Disease dataset, which contains critical patient data like age, gender, blood pressure, cholesterol level, type of chest pain, and other crucial health factors. Prior to creating our model, we pre-processed and cleaned the data by dealing with missing values, changing categorical variables into numerical form, and scaling the features for uniformity. We then optimized the SVM model using grid search and cross-validation to make it run at its optimal level. The resulting model had an accuracy of 86.41% in the test set and performed better than other popular models such as logistic regression and random forest. The significant about this work is the potential for applying it in practical situations. An SVM-based program such as this could be a second opinion for physicians or integrated into early diagnostic tools—most helpful in clinics with limited access to specialists. It's progress toward smarter, data-driven healthcare that enables faster and more precise diagnoses. There's still potential for expansion, using bigger, more varied datasets or incorporating real-time patient information could further enhance the model. But this research demonstrates that with the proper data and methodology, machine learning can be a useful tool in the early diagnosis of heart disease.

Ähnliche Arbeiten