OPTIMAL HEALTH DIAGNOSIS RECOMMENDATION SYSTEM

Abstract

Healthcare navigation remains a pressing challenge in growing urban centres like Hyderabad, where patients must choose among hundreds of hospitals without reliable cost or distance guidance. This paper presents an AI-powered hospital recommendation system that integrates Google Dialogflow for conversational intent parsing, a Random Forest regression model for medical cost prediction, and the Geoapify geospatial API for proximity scoring. Given six patient parameters— name, age, symptom, severity, budget, and location—the system returns a ranked shortlist of up to ten hospitals with budgetfitness labels. Trained on a synthetic dataset of 15,000 records covering eight medical categories, the Random Forest model achieves an overall R² of 0.953 and MAE of ₹412, demonstrating strong cost-prediction accuracy suitable for patient-facing guidance.