Integrating AI Recommendations into the Pharmacologic Management of Major Depressive Disorder

AI predictions provide an important opportunity to support clinicians during complex decision-making processes. One such process is selecting treatments for major depressive disorder (MDD). Towards the goal of implementing AI models that make MDD treatment recommendations, we have designed a factorial vignette study to assess how recommendations and explanations may influence clinician's treatment decisions. We report on our initial data analysis, evaluating the influence of incorrect predictions on antidepressant selection. We found that recommendation correctness had a significant effect on treatment selection accuracy.

• Authors' version (pdf)

Maia Jacobs, Roy H. Perlis, Melanie F. Pradier, Finale Doshi-Velez, Elizabeth Mynatt, and Krzysztof Z. Gajos. Integrating ai recommendations into the pharmacologic management of major depressive disorder. In Proceedings of the workshop on Identifying Challenges and Opportunities in Human–AI Collaboration in Healthcare at CSCW 2019, 2019.

BibTeX