Privacy Preserving Multi Disease Prediction Framework in Healthcare Using Federated Learning

Abstract

The rapid adoption of machine learning in healthcare has enhanced disease prediction capabilities. However, traditional centralized approaches pose significant risks to patient data privacy and security. To overcome these limitations, this paper presents a privacy-preserving multi-disease prediction framework based on Federated Learning, a decentralized approach that enables collaborative model training without sharing raw data. The proposed system supports the prediction of multiple diseases, including heart disease, lung disease, Parkinson’s disease, and diabetes. In this framework, datasets are distributed across multiple simulated hospital clients, where local models are trained independently using Logistic Regression and Neural Networks. The locally trained model parameters are securely aggregated at a central federated server using the Federated Averaging algorithm to generate a global model. Performance evaluation is conducted using metrics such as accuracy, precision, recall, and F1-score, and the model with superior performance is selected for final prediction. The system is integrated into a Streamlit-based web application to provide real-time, user-friendly disease prediction. Experimental results demonstrate that the proposed approach achieves high predictive accuracy while ensuring data privacy, scalability, and efficient model collaboration. This work highlights the potential of Federated Learning as a secure and practical solution for multi-disease prediction in modern healthcare systems.