Atlantic Health System, one of the largest nonprofit healthcare networks in New Jersey, is deploying a new artificial intelligence tool to improve medication management and avert medication safety events. The FeelBetter AI-based precision population health software platform identifies high-risk patients, suggests clinical interventions and oversees care journeys for polypharmacy patients (those on at least five medications).
“As we brought clinical pharmacists into our accountable [care] organizations [ACOs], we realized a significant return on our investment in their work toward achieving our triple-aim goals of quality, experience and affordability,” said Jim Barr, MD, the chief medical officer of Optimus Healthcare Partners and Atlantic Health ACOs. “We can now use FeelBetter to scale medication management efforts for 500,000 patients without the need for additional [full-time employees].”
The system went live in early June, and the algorithm now identifies which patients are at the highest risk and most in need of medication management. “Those are the patients that our two ACO clinical pharmacists will be specifically dedicated to,” Dr. Barr said. “And for our physicians, instead of trying to digest list after list of reports, they will know that this is the next patient they need to act on. It takes away the burden and delivers timely, actionable insights.”
Beyond Atlantic Health, the FeelBetter system is already being used in the management of more than 350,000 patients in the United States and Israel. “Within Leumit, a leading [health maintenance organization] in Israel, we have seen a 24% reduction in hospitalizations for high-risk patients and 46% reduction in days spent in the hospital in our intervention group, along with a fivefold increase in pharmacist capacity,” said FeelBetter CEO Liat Primor.
In a retrospective study presented at the Society of General Internal Medicine 2023 Annual Meeting (poster PS4-187) and the 2023 Annual Research Meeting of Academy Health, investigators from Brigham and Women’s Hospital (BWH), in Boston, examined data on 108,817 of the hospital’s patients 65 years of age and older who were taking an average of nine medications to treat multiple chronic conditions, and used the FeelBetter algorithm to stratify these patients into risk groupings. Patients in successive risk strata incurred incrementally more emergency department visits and hospitalizations at one, three and six months after the prediction. Patients in the highest risk percentile (99%) incurred a mean of 0.531 hospitalizations three months post-prediction; randomly, across the overall population, there is a hospitalization rate of 0.04%. Among that cohort categorized as highest risk by FeelBetter, 22.9% of hospitalizations were considered possibly preventable.
“Looking randomly at the BWH population, they would have needed to meet with 25 patients in order to address one who would be hospitalized,” Ms. Primor said. “FeelBetter identified more than half of high-risk patients who would be hospitalized within the next three months, so they only needed to meet with two patients to address one who would be hospitalized.”
The study also estimated $4.6 million in potential savings from avoidable healthcare utilization in the top 5% of high-risk patients in the three months following the use of FeelBetter for risk stratification.
The sources reported no relevant financial disclosures beyond their stated employment.
A previous version of this article included an incorrect title for Dr. Barr; the article has been updated.
This article is from the September 2024 print issue.