Drug Recommendation System Based on Sentiment Analysis of Drug Reviews using Machine Learning

Abstract

People are self-medicating more during the COVID-19 pandemic because they can't get to good medical tools. This makes their health situations worse. This study suggests a drug suggestion system that uses machine learning and emotion analysis of patient reviews to make the job of healthcare workers easier. We use different vectorization methods, like Bag of Words (BoW), TF-IDF, Word2Vec, and Manual Feature Analysis, to guess how people feel about certain diseases and suggest the best drugs for them. We use classification methods, such as LinearSVC, to rate emotions based on their accuracy, F1-score, precision, and AUC score. The results show that LinearSVC with TF-IDF vectorization works well, as it achieved 93% accuracy, which was better than other models. By making drug suggestions, this system aims to make it easier for people to get medical care when they need it most, especially during emergencies.