An Advanced Model to Predict Heart Disease Applying Random Forest Classifier and Whale Optimization Algorithm

Abstract

Background: In this article a model for heart disease prediction has been proposed using machine learning to address the significant concern of lives lost due to this disease, especially in remote and underprivileged areas where access to proper medical support is lacking. Identifying the disease at an early stage is crucial for preventing unnecessary fatalities. Methods: The dataset has been collected from UCI data repository. The raw data set consists of 14 features, which are age, sex, cp, tresbps, chol, fbs, restecg, thalach, exang, oldpeak, slope, thal, ca, target. The collected dataset has been standardized applying StandardScaler and then feature selection has been carried out applying SelectKBest to select 9 most competent features. The selected features are age, sex, cp, thalach, exang, oldpeak, slope, ca, thal and target. 80% of the dataset has been used for training and remaining for testing. For classification Random Forest has been applied and the model performance has been increased applying Whale Optimization Algorithm. Findings: The WOA, inspired by the social behavior of humpback whales, has been utilized to optimize the parameters and configurations of the model. The proposed model has achieved 86.53% accuracy. Novelty: The proposed model combines the Random Forest classifier, with the SelectKBest to select optimal number of features through experiments which ensures that the model is trained only with effective features. This dimensionality reduction makes the model computationally efficient. More over application of Whale Optimization has increased the performance of model with 0.93 AUC score. This model's novelty lies in its targeted focus on undeserved populations, aiming to improve early identification of heart disease and reduce fatalities in resource constrained areas.