Using Machine Learning to Improve Patient Safety in the Home or Remote Setting for Adults

Feske-Kirby K, Whittington J, McGaffigan P. Using Machine Learning to Improve Patient Safety in the Home or Remote Setting for Adults. Boston: Institute for Healthcare Improvement; 2022. (Available at www.ihi.org)

The primary aim of the IHI innovation project described in this report was to assess the use of predictive analytics, specifically machine learning, to improve patient safety through emerging and existing approaches to predict risk, such as technologies and decision support tools. Specific attention was given to how predictive analytics and machine learning can assist in monitoring patient deterioration in the home setting for adults ages 18 and older.

This report discusses key reasons why efforts to develop and implement new predictive analytic technologies in health care encounter numerous barriers such as complications of data mining and protection, daily workflow and a lack of interoperability, concerns about accuracy, workforce burden and lack of relevant expertise, and related health equity issues like biased data.

Recommendations to address these barriers and other safety considerations are also presented, including the need for a clear purpose for new technologies, an emphasis on daily workflow and interoperability, and the development of quality and safety guardrails to support the development and integration of machine learning tools or remote patient monitoring systems.