Medical Big Data Analysis Using Machine Learning Algorithms in the Field of Clinical Pharmacy

Abstract

Industrial reforms utilizing artificial intelligence (AI) have been progressing remarkably around the world in recent years. In medical informatics, the application of medical big data analytics using AI is also being promoted, and it is expected to provide screening methods for predicting potential adverse drug reactions (ADRs) and discovering new eŠects. Previously, we developed a unique ADRs analysis system that incorporates Accord.NET, an open-source machine learning (ML) framework written in the programming language C#, and uses the Japanese Adverse Drug Event Report (JADER) database. By using this system to analyze ADRs and screening the cause and severity of ADRs, information can be obtained to evaluate efficacy as well as ADRs. Although both statistical methods and ML are commonly used for prediction, a characteristic diŠerence between them is that the former emphasizes causal relationships and the latter emphasizes prediction results. Therefore, it is important to distinguish between cases where decisions must be made with an emphasis on causality and those where decisions must be made by focusing on unknown risks, and statistical methods and ML should be selected and used as appropriate. Against this backdrop, this paper describes a use case and suggests that the proper use of AI tools to analyze medical big data will help clinical pharmacists practice optimal drug management for each patient.