AI-DRIVEN PREDICTIVE ANALYSIS FOR HEALTHCARE MANAGEMENT

Abstract

The integration of machine learning (ML) in healthcare has opened new possibilities for improving diagnostic
accuracy and accessibility, particularly in resource-constrained environments. This study presents a robust framework for
symptom-based disease prediction using machine learning techniques, with a focus on the Random Forest Classifier (RFC) and
Multi-Layer Perceptron (MLP) models. The proposed approach highlights the importance of data preprocessing, feature
engineering, and model evaluation while addressing key challenges such as missing data, overlapping symptoms, and ethical
considerations. The system utilizes datasets sourced from Kaggle to train and validate the models. Experimental results indicate
that the Random Forest Classifier outperforms the MLP model, achieving an accuracy of 99%. In addition, an interactive web
application has been developed using Streamlit, allowing users to perform disease prediction, update datasets, and retrain models
dynamically. This solution offers a scalable and reliable tool for early disease detection, especially in underserved regions,
helping to reduce diagnostic errors and improve access to healthcare services