Advanced Drug Recommendation System Using Multi-Layer Perception with TF-IDF Features

Abstract

This work proposes the introduction of a sentiment and machine learning-based drug recommendation system. This system will accept disease names from patients, after which it will recommend drug and simultaneously display sentiment rating based on reviews given by previous users based on their experiences. If the patient's expected rating is high, then the patient may believe the recommendation and take the medicine. The study that has been suggested has made use of several methods to extract features, such as TF-IDF (term frequency – inverse document frequency). This technique has been used in order to analyse the frequency of terms inside pre-processed dataset. The collected features will be used and trained with multilayer perceptron classifier (MLP). The drug review dataset, which can be found on the UCI machine learning website, was used so that this study could be carried out. When compared to other machine learning models, the MLP classifier that utilises TF-IDF feature extraction will provide results that are better in terms of performance