The model is built based off 200k reviews from different patients on different drugs recommended to them.
The aim is to assist medical practitioners in post marketing surveillance of drugs after clinical trials by collecting information of patients and reviews of drugs recommended to them.
This is to enable the medical practitioners to note the drug that works for a particular set of people and same drug that produces negative effects on another set of people.
More forms will be added soon to collect more information. Once the target number of patient information is collected, the new dataset generated will then be used to build a drug recommender system.
It's a Natural Language Processing model trained with a support vector classifier algorithm with returned an accuracy of 95%.