MINING SERENDIPITOUS DRUGS FROM REVIEWS USING MACHINE LEARNING TECHNIQUES

The word serendipity means ‘happy accident’. Making discoveries by accident has contributed a lot to medical history. Serendipitous drug use is when a patient takes a prescription for a separate known indication and unintentionally experiences relief from comorbid illnesses or symptoms. The discovery of numerous new medication indications has benefited greatly from serendipity throughout history. Drug-repositioning hypotheses might be created and validated if patient-reported serendipitous drug usage in social media could be computationally discovered. We looked into deep neural network models for social media mining of accidental drug use. We contrasted Regression and KNN with our support vector machine, random forest, RNN, and LSTM algorithms. We used machine learning and natural language processing techniques in a web application to mine social media and data reviews for accidental drug use. An essential algorithm is sentiment analysis. We decided to employ Natural Language Processing for our project since it can be used to identify sentiment in text. Upon reviewing reviews of various pharmaceuticals that have been rated on a scale of 1 to 10 and have been reviewed as texts. This data set was collected from the UCI machine learning repository, which contained the train and test data sets (divided as 75–25%). In general, we categorize the drug's numerical rating into three categories: positive (7–10), negative (1-4), or neutral (4-7). We chose to look into how the ratings of the drugs are affected by the inclusion of different words in reviews for ailments with many reviews for drugs that are used to treat those conditions. Our main goal was to construct supervised machine learning classification algorithms that use textual reviews to predict the rating class. Last but not least, we used machine learning and natural language processing techniques to mine data reviews for drug usage