The advancements in machine learning have created a scope for accurate disease prediction and personized drug recommendation. However, the sensitive and private nature of medical data become a challenge in developing robust models for ensuring patient privacy. In this paper, we propose a novel approach that involves federated learning to address the challenges, enabling collaborative model training and knowledge sharing across multiple healthcare entities. The proposed methodology involves establishing a network of healthcare providers, each contributing their local disease prediction and drug recommendation models. Through federated learning, these models are aggregated and refined without sharing raw data, ensuring patient privacy. This project contributes to the advancement of healthcare technology by offering a secure and privacy-preserving solution for disease prediction and drug recommendation. The proposed federated learning approach shows a way for healthcare entities to collaborate and improve models without compromising the data privacy. Through rigorous experimentation and validation, we anticipate that the federated learning framework will prove instrumental in driving the future of data-driven healthcare advancements.