A Framework for Patient Specific Drug Recommendation and Side-Effect Prediction System
Morarjee Kolla, Vemula Ishitha Reddy, Rddhi Reddy · 2025
At present, systems for disease prediction, drug recommendation, and adverse drug reaction (ADR) prediction often operate independently, limiting their ability to provide a cohesive, personalized healthcare solution. These systems lack integration, struggle with real-time data adaptability, and fail to account for individual patient factors. This paper aims to propose a framework that can accurately predict the likelihood of various diseases based on a patient’s symptoms and medical history, as well as suggesting suitable medications for the predicted diseases, considering factors such as effectiveness, dosage, and patient-specific attributes, including any potential side effects in order to provide tailor-made recommendations that reduce risk. The system unifies three modules: disease prediction using patient symptoms and medical history, drug recommendation based on efficacy, dosage, and side effect profiles, and a sentiment analysis module to refine recommendations using patient feedback. Techniques like Random Forest and Neural Networks (NNs) are employed, with support from knowledge graphs and collaborative filtering for enhanced personalization. Results demonstrate significant improvements in accuracy, precision, and recall. By integrating disease prediction, drug recommendation, and side effect analysis, the proposed system provides a comprehensive approach that enhances outcomes and minimizes risk.