Drug Recommendations Using Support Vector Machine

As the online usability of health forums online has increased and user provided information, content is about the health is voluminous, we propose a unique hybrid approach for the recommendation of drug that considers the sentimental analysis and review score using vector-based approach. The main objective of the approach is to provide fine-tuned decisions about various drug choices. We use natural language processing methods that convert the reviews into numerical vectors which capture sentiment and semantic information. We have used Support Vector Machine (SVM) to process the vectored data. This is the first approach to cover the qualitative aspects of the user experience and emotional content. We have collected 200,000 datasets from UCI ML Drug Review dataset to implement the proposed classifier. Our results demonstrate that our vector-based approach to drug recommendations outperforms traditional methods in terms of accuracy and user satisfaction. SVM reports an accuracy of 96.98%, with precision and f1 score of 0.926 and 0.936, respectively.