Health Mantra: Machine Learning based Solution for Usage of Drugs from Social Network

  • Shridevi Soma, Supriya B D


             Medication through Social Network is a need of hour in the situation like COVID19 and for the people placed away from the health centers.  In a social  media when somebody finds some good medicines by accident, such types of drugs are considered as serendipitous drugs.  Finding such medicines is based on the health condition, mental status of the patient, cost and reviews of the medicines.  This requires a kind of intelligence to fulfill the need of the patient.  Hence, Health Mantra: A Machine Learning based solution proposed in this work provides an optimal solution for serendipitous drugs.  This work contains various modules such as add patient details, view patient details, search drug and get drug details, view all types of drug reviews, view drug score result, drug review count.  The total of 161297 drug data are considered from a standard webMD data set for experimentation, which includes the features like drugID, drug name, condition, review, ratings, useful count, date of manufacture the drug.  Apriori algorithm is used to perform association analysis on the characteristics of drugs and balancing the data set.  A Naives Bayes classifier is used in the proposed system and Optimal results are obtained from Naives Bayes classifier.  Over 53766 drug data are tested against whole data set from standard dataset i.e. 161297 that resulted with an Accuracy of 100% for Naives Bayes classifier against existing system[1] where the authors experimented with 15714 dataset with Deep neural network classifiers and obtained an accuracy of 93%.  The future scope of this work is to design a system for differentiating Drugtraffickers and Consumers