Machine Learning-Based Outcome Prediction for Data-Driven Personalized Pharmacotherapy in Type-2 Diabetes Mellitus

Wataru Takeuchi¹, Shinji Tarumi², Michael Flynn³, Hideyuki Ban⁴, David Shields⁵, Kensaku Kawamoto⁵

¹Hitachi Ltd.
²Research and Development Group, Hitachi, Ltd.
³Departments of Internal Medicine and Pediatrics, Univ. of Utah
⁴Central Research Lab., Hitachi, Ltd.
⁵Department of Biomedical informatics, Univ. of Utah

Details

12:00 - 13:45 | Mon 6 Nov | Auditorium Foyer, E1/E2, Upper Atrium Space | MLunch_Break.29

Session: Lunch, Posters and POC Technologies Demonstrations – Session I

Full Text

Abstract

In order to help clinicians to provide personalized pharmacotherapy for patients with type-2 diabetes mellitus, machine learning-based models were developed which predict the probability of achieving treatment targets such as a hemoglobin A1c level under 7% at 90 days after prescription. The best prediction model using gradient boosting achieved 0.85 AUC in both 5-fold cross validation and external validation.