Adaptive Federated Learning for Mortality Risk Assessment in Multimorbid Patients

The healthcare domain faces numerous challenges, particularly in assessing the mortality risk of multimorbid patients. Multimorbidity, defined as the coexistence of multiple chronic diseases within a patient, complicates the prognosis and decision-making process for healthcare providers. Traditional risk assessment methods often fail to account for the complex interactions between various chronic conditions. In recent years, machine learning (ML) has emerged as a powerful tool for predicting patient outcomes, but challenges such as data privacy, the limited availability of quality data, and the need for personalized healthcare solutions remain. This project introduces an Adaptive Federated Learning (AFL) framework for mortality risk assessment in multimorbid patients. The core concept of federated learning (FL) lies in its ability to train machine learning models on decentralized data, preserving patient privacy. AFL enhances this by adapting the learning process to improve the model’s accuracy over time based on the patient’s medical history, thus enabling more personalized and accurate risk prediction. In this study, machine learning models such as Random Forest Classifier, Logistic Regression, Decision Tree Classifier, and KNeighbors Classifier are applied to predict the likelihood of mortality in patients with multiple chronic diseases. The adaptive federated learning model is evaluated against traditional centralized models, focusing on factors like accuracy, privacy, and data security. The proposed system also leverages a Flask-based web interface for real-time prediction and monitoring, The integration of federated learning ensures that the system can operate across multiple institutions, utilizing their local data without compromising patient confidentiality. Ultimately, the AFL approach presented in this study aims to enhance the healthcare system’s ability to accurately predict mortality risks in multimorbid patients while maintaining high standards of privacy and security.