A hybrid machine learning approach for heart disease prediction using hyper parameter optimization

Abstract

Heart disease is a serious terminal condition in most parts of the world. The acute lack of medical professionals, expertise, and technology to identify important signs. So a smart and efficient model and technology is required to lead early diagnosis of heart disease. The current study proposes a new experience-based method namely HSPUCD (Heart stage prediction using clinical data) to forecast cardiac disorders utilizing hybrid machine learning. Type 4 protocols Cross Validation is used to assess a model's competence on unknown data. Hyper parameters were modified to regulate the suggested model's behavior and performance. The suggested system is based on extensive research. Data pre-processing was utilized to discover the most relevant network traffic and generate the initial candidate collection of feature characteristics. The suggested model additionally uses automated feature selection to extract features from a candidate set of features and selects the most relevant subset of features from this candidate set. Crow search algorithm is used to extract data characteristics from the pre-processed data. Finally, an online training classifier is utilized to categorize the data and forecast the result. The proposed strategy greatly increases the system's illness prediction accuracy with less noise and more advanced regression approaches using existing medical datasets. Mathematical results depicts the effectiveness of the proposed approach in comparison with others.