Abstract: Clinical decision support (CDS) suffers low provider acceptance which potentially could be improved with machine learning. We extracted repeat imaging decision support data between 2007 and 2014, and built multiple classifiers to predict the acceptance of repeat imaging alerts. Our best classifier (Random Forest) achieved an accuracy of 75.1%. Machine learning could be utilized to improve provider acceptance of CDS alerts by targeting instances with high likelihood of acceptance.
Learning Objective 1: After participating in this session, the learner should be better able to:
- Recognize the potential role of machine learning in improving provider acceptance of clinical decision support (CDS) alerts.
Wasim Al Assad (Presenter)
Harvard Medical School
Ivan Ip, Brigham and Women's Hospital
Li Zhou, Harvard Medical School
Ramin Khorasani, Brigham and Women's Hospital