Disease Prediction and Recommendation System: A Comprehensive Study with Multi-Model Comparison and Clustering Integrations

Abstract

Healthcare decision-support systems increasingly rely on machine learning to assist clinicians and patients. We present a comprehensive system that predicts diseases from symptom profiles and returns actionable recommendations (diet, medication, precautions, and workouts) using a structured health- care dataset. Five supervised learning models — Random Forest (RF), Support Vector Classifier (SVC), k-Nearest Neighbors (KNN), Naive Bayes (NB), and Decision Tree (DT) — were trained and evaluated for comparison. A Random Forest model achieved the highest accuracy of 99.05% and was selected as the final classifier. In addition, unsupervised K-Means clustering was implemented to reveal symptom similarity groups, while a Top-3 confidence mapping mechanism provided multi-disease probability outputs. We also report feature-importance ranking and interpretability analysis. The system provides interpretable, data-driven disease predictions with integrated healthcare rec- ommendations, bridging AI-based diagnosis with clinical decision support.