Investigation of the Diseases Prediction Rate using Specified Heuristic Method

Abstract

The mining of healthcare data is an important aspect for predicting and estimating critical diseases based on previous records. Various tools and technologies are employed in healthcare data mining, with data mining algorithms playing a central role. There has been increasing interest in gathering nontraditional digital information to perform disease surveillance, including datasets from social media, internet searches, and environmental data. With the growth of big data in biomedical and healthcare communities, accurate analysis of medical data supports early disease detection, improved patient care, and enhanced community services. In this paper, we propose a new model that combines classification methods such as k-nearest neighbor (KNN) classification and decision tree classification with optimization techniques from the swarm intelligence family, specifically particle swarm optimization (PSO). The proposed optimization methods enhance classification accuracy for datasets related to disease prediction. We evaluate the model using datasets such as the Heart dataset, Cleveland dataset, and Diabetes dataset, all sourced from the UCI Machine Learning Repository. The simulations are conducted using MATLAB software, and results demonstrate improved classification performance for disease prediction tasks.