A personalized drug recommender system using machine learning algorithms

Abstract

In today's digital world rapid advances in healthcare and the rise of the internet have enabled us to look online at various medicines available in the market for a particular disease. The internet helps us to make an informed decision when buying medicines and also look for alternatives. However, when looking online, patients often encounter irrelevant medicines or medicines which are not suited to their needs. The medicine may even contain side effects which we may not even know about. This project aims to propose a machine learning-based drug recommender system that considers the patient’s profiles when recommending a drug. Unlike traditional systems, our system accounts for factors that come when deciding to recommend a medicine like age, gender, side effects, user review etc. We have combined an ensemble model as well as a K-Means cluster model to ensure the accuracy of our recommendations. The model has an accuracy of 87.34% and a precision of 88.88%, which ensures that the drug our model has recommended is truly what the person requires while also accounting for side effects.