Advanced Deep Learning Approach for Predicting Heart Disease Through Comprehensive Analysis of Clinical Features and Data Science Techniques

Abstract

This study develops and evaluates a deep learning model for predicting heart disease using the UCI Cleveland clinical dataset, which contains 303 patient records and 14 commonly used diagnostic features. After data cleaning, imputation of missing values, feature encoding, and normalization, a neural network with multiple dense layers, batch normalization, and dropout was trained and tested on a stratified train?test split. The model achieved an accuracy of about 82 percent with balanced precision, recall, and F1 scores for both classes. Traditional machine learning models, particularly Support Vector Machine achieved slightly higher performance, but the neural network remained competitive. These findings highlight the potential of deep learning as a decision-support tool for early identification of heart disease risk when combined with structured clinical data.

Ähnliche Arbeiten