Haroz E, Grubin F, Goklish N, Pioche S, Cwik M, Barlow A, Waugh E, Usher J, Lenert MC, Walsh CG. Designing a Clinical Decision Support Tool That Leverages Machine Learning for Suicide Risk Prediction: Development Study in Partnership With Native American Care Providers. JMIR Public Health Surveill 2021;7(9):e24377. DOI: 10.2196/24377
Use of algorithms can be helpful as a method of risk detection for suicide. Researchers developed a machine learning algorithm to help identify people who are most at risk for suicide deaths or attempts in Native American reservation populations. Researchers explored how to implement the algorithm tool to inform care pathways in community-based suicide surveillance and case management systems. Researchers conducted qualitative in-depth interviews with 9 case managers from 3 communities (White Mountain Apache Tribe and two sites in Navajo Nation). Interviews included questions about staff perceptions and evaluation and response to risk as well as suggestions for implementation of risk algorithms into their care process. Participants highlighted the importance of current behavior, past history, and location to prioritize individuals. Participants agreed that algorithm-generated risk flags would be useful along with as much information as possible to respond to the flag. Researchers are now conducting an implementation pilot of the algorithm tool in the White Mountain Apache Tribe that flags people as high-risk or low-risk after an in-person follow-up.