Application of Machine Learning Approach for Medicine Recommendation in Healthcare System

Abstract

Healthcare is a very crucial domain that should evolve to meet the growing demands of the population and unpredictable global crisis. In challenging situations, such as pandemics, natural disasters, or other medical emergencies, patients may not receive proper treatment due to high demand. Recognising this shift, we propose a machine learning-based medicine recommendation system that allows users to input symptoms to detect possible diseases, receive basic descriptions, and obtain corresponding medicine suggestions. In cases where direct medication is not available, the system provides essential instructions to manage the condition. This model aims to serve as a trustworthy platform for early diagnosis, especially when traditional healthcare access is limited. By bridging the gap between online health inquiries and clinical insight. Specifically, it focuses on improving healthcare accessibility and reliability in both everyday and emergencies. The results support the use of Random Forest, which achieved perfect classification performance, closely followed by K-Nearest Neighbour. In contrast, SVM underperformed due to its sensitivity to high-dimensional data. The system also incorporates safety features, such as precautionary measures to consult a physician when the prediction confidence is low, ensuring ethical reliability.