Metaheuristic-based hyperparameter optimization for multi-disease detection and diagnosis in machine learning

Metaheuristic algorithms with machine learning techniques have become popular because it works so well for problems like regression, classification, rule mining, and clustering in health care. This paper's primary purpose is to design a multi-disease prediction system using AI-based metaheuristic approaches. Initially, the data is collected in the form of diverse classes, which include Id, gender, date of birth, etc. The data has been preprocessed, normalized, and graphically represented to improve its quality and detect any errors. Later, machine learning models, such as decision tree, extra tree classifier, extreme gradient boosting classifier, light gradient boosting machine classifier, random forest, and artificial neural network, are initially trained without optimizing hyperparameters and then fine-tuned by integrating various hyperparameter optimizers such as grid search CV, random search, hyperband, and genetic search. During experimentation, it is found that optimizing the models using random search optimizer computed the highest accuracy of 100% as compared to the rest of the hyperparameter optimizers. In the context of 'Service Oriented Computing and Applications,' our multi-disease prediction system offers valuable innovation. It aligns with the goal of enhancing healthcare services, patient outcomes, and healthcare efficiency. Our pioneering integration of metaheuristic algorithms and machine learning introduces intelligent healthcare solutions, with the study's focus on hyperparameter optimization and achieving 100% accuracy demonstrates practical significance in SOC and its applications.