Medication recommender system for healthcare solutions

Abstract

An ever-increasing number of individuals hear about the well-being and clinical determination issues. Nonetheless, as indicated by the organization’s report, over 200 thousand individuals in China, even 100 thousand in the USA, bite the dust every year due to prescription mistakes. Specialists makeovers 42% of medicine mistakes since they compose the remedy as indicated by their very restricted encounters. Advancements such as information mining and recommender innovations provide prospects to investigate likely information from determination history records and assist specialists with endorsing drugs accurately to diminish prescription mistakes adequately. This paper plans and actualizes a general medication recommender system that applies information mining to suggest a better prescription to hospitalized patients. Our medication recommender system consists of constructing feature modules, data splitting modules, count vectorizers, and sentiment classification modules. The classifiers such as logistic regression, Naive Bayes, and random forest are used for the model evaluation. Our classification results show that the random forest-based medication recommender system has the highest prediction results than other classifiers. The random forest-based medication recommender system achieved 86.6% accuracy on the UCI ML Drug Review dataset.

Subject Classification:

• 68N99 Computer science

Keywords:

• Deep neural network (DNN)
• Count vectorizer
• Artificial intelligence (AI)
• Sentiment analysis
• Recommender systems