Montefiore Health System is working with Intel on an artificial intelligence initiative to predict potential complications for patients.

While early efforts are focused on identifying emergency department and ICU patients at risk for respiratory failure as much as 48 hours before problems become apparent, the health system has bigger plans for the effort.

The delivery system is hoping to expand the initiative to include patients outside the hospital. The eventual goal is to use Intel AI software, called the Patient-centered Analytical Learning Machine, or PALM, to analyze genomic and socioeconomic data to determine who will develop chronic conditions nearly two years before they manifest.

Being able to identify at-risk patients is becoming more important as patients are not being admitted to the hospital as often as they previously would have. “Medicine is evolving and we move more patients to home and the clinic, so only the sickest are admitted,” says Michelle Gong, MD, director of critical care research at Montefiore.

Because of this, remote monitoring of patients in their home using AI-supported software is being supplemented with self-collected information from patients, such as vital signs transmitted several times a day, and data from electronic medical records that can show if a patient is doing better or not responding to treatment. If a patient appears to be faltering, a physician can immediately be alerted, Gong explains.

In one case, an AI-based alert saved a life at Montefiore. A patient was admitted to the hospital, but clinicians couldn’t detect any issues with the patient’s condition. However, an AI alert prompted physicians to order another CAT scan, which found bleeding in the liver—if that had not been detected, it could have caused the patient’s death.

Intel is supporting Montefiore with a large supply of data stored on the AI platform and also has worked with the hospital on improving workflows, says Jennifer Esposito, health and life sciences worldwide general manager at Intel.

PALM brings together data sets that otherwise would be separate, such as administrative and pharmacy data, and now Montefiore can bring that data and other sources together for analysis. “PALM is the organizer of the data,” Esposito says.

Montefiore also is receiving decision-support services from Epic, its electronic health record vendor, to identify high-risk patients.

In another initiative, the organization is using an algorithm called APPROVE that identifies severe sepsis and other conditions. The algorithm finds patients with the highest risk for sepsis, separates out false positives and enables physicians to focus on the most acute patients.

“Less people trigger, but those who trigger do better, as we send a doctor to the patient who needs an extra look,” Gong says.

Using the machine learning tools in PALM, Montefiore also has found it is possible to run algorithms in real-time to identify persons at risk, Gong notes. “But that’s only the first step. The second step is to use decision support to change patient behavior.”

Gong recently received a grant from the National Institutes of Health to test how artificial intelligence impacts outcomes, with the research likely taking a couple years to complete.

Register or login for access to this item and much more

All Health Data Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access