UHS, Eleos Health Launch AI Pilot to Improve Behavioral Outcomes

Universal Health Services (NYSE: UHS) has teamed up with Eleos Health, a digital behavioral health solutions provider, to launch a new artificial intelligence (AI) pilot study. 

The goal is to measure the effectiveness of a new behavioral health software that uses AI and voice analysis to improve behavioral health treatments and outcomes. If the small, six-week study is successful, UHS will roll the pilot out on a larger scale.

UHS is one of the nation’s largest operators of acute care and behavioral health facilities, with more than 400 locations across 37 U.S. states, Washington, D.C., Puerto Rico and the United Kingdom. Its behavioral health arm includes 328 facilities.


Meanwhile, Eleos Health is a digital health company that uses voice analysis, neuro-linguistic programming and artificial intelligence to help behavioral health providers improve treatment outcomes and workflow efficiency.

The new Eleos Health clinical decision-support software uses AI to analyze and process conversations between patients and clinicians during therapy. AI algorithms then help inform the implementation of evidence-based techniques and the therapeutic relationship. Additionally, the solution automatically generates administrative reports and translates the information into actionable clinical insights.

The idea is that the solution will supplement decision-making and help clinicians better target their care, while also allowing them to spend less time on paperwork and documentation.


UHS launched the six-week pilot earlier this month at one of its residential behavioral health facilities in Nashville. As part of the study, clinicians are using the new software tool in individual weekly sessions with consenting patients between the ages of 13 and 17.

The results of the pilot will guide the development of a larger, more comprehensive study involving a greater number of facilities later on, according to a press release announcing the news.

Companies featured in this article: