Optimized Machine Learning Model using Binary Grasshopper Algorithm for Heart Disease Diagnosis

Heart disease appears to be a significant worldwide medical concern and alarmingly, a leading cause of death. Accurate prediction and rapid diagnosis are important in providing timely medical attention and improved patient results. Machine learning (ML) provides a possible solution to this issue by examining vast volumes of patient data to discover patterns and forecast the risk of heart disease. However, the quality and utility of input features significantly influence the performance of ML models. This study aims to improve the effectiveness of heart disease prediction by selecting the most relevant features using a feature selection algorithm and to compare the classification performance of machine learning models using the Cleveland CVD dataset. The Binary Grasshopper Optimization Algorithm (BGOA) is used to identify the significant features in the dataset, which chooses nine important attributes. This work analyses the performance of Naive Bayes (NB), Support Vector Machine (SVM), and Logistic Regression (LR) methods with and without incorporating BGOA. The study made use of the Cleveland Cardiovascular Disease (CVD) Dataset obtained from Kaggle, which includes 303 occurrences and 14 attributes. The Naive Bayes method outperformed other classification algorithms in terms of all evaluation metrics used in this study.