سامانه همایش های دانشگاه آزاد اسلامی واحد نجف آباد

صفحه اصلی / سومین کنفرانس بین المللی مهندسی و علوم کامپیوتر و اولین کنفرانس ملی سامانه های هوشمند، کلان داده و رایانش پیشرفته

Optimizing Heart Disease Screening: A Hybrid Random Forest and Genetic Algorithm Approach for Feature Selection

نویسندگان :

Majid Khalilian¹ Sepehr Eslami²

1- دانشگاه آزاد کرج 2- دانشگاه آزاد تهران مرکز

کلمات کلیدی :

Heart Disease Prediction،Genetic Algorithm،Feature Selection،Random Forest،Big Data Analytics

چکیده :

The rapid growth of healthcare data has created a significant challenge in Big Data analytics: processing high-dimensional medical records efficiently. While large datasets offer opportunities for better diagnosis, they often suffer from redundancy and high computational costs. This paper proposes a hybrid framework for heart disease prediction that integrates a Genetic Algorithm (GA) with a Random Forest classifier. The primary objective is to identify a minimal subset of features without compromising diagnostic sensitivity. Unlike conventional methods that prioritize overall accuracy, this study introduces a fitness function specifically designed to balance sensitivity and accuracy, thereby minimizing false negatives. The proposed method was evaluated using the UCI Cleveland dataset. Experimental results indicate that the feature space can be reduced from 13 to only 4 essential attributes (Age, Chest Pain Type, Exercise-Induced Angina, and Number of Colored Vessels). The optimized model achieved a sensitivity of 96.43%, equal to the baseline model utilizing all features, with a comparable accuracy of 88.52%. These findings suggest that intelligent feature selection can significantly enhance the scalability of diagnostic systems for real-time health monitoring.