Will This Substance-Abuse Patient Relapse? Ask IBM's Watson

May 17, 2017

A company specializing in population health management plans to harness the artificial intelligence of Watson, the IBM supercomputer, to better predict when a person in treatment for substance abuse is going to start using again.

With a digital head's-up about a looming relapse, an addiction counselor or case manager can pick up the phone and guide the patient back to healthier choices — like getting back to the gym.

Managing the risk for relapse is the name of the game for MAP Health Management, a company in Austin, Texas, that gives health insurers, health systems, and clinicians high-tech tools to treat patients with substance-use and chronic behavioral disorders. About six in 10 patients treated for substance abuse disorder (SAD) will fall back into their old habits at least once in the first year, MAP CEO Jacob Levenson told Medscape Medical News. For 20% to 25%, however, the backsliding is severe enough to require a higher and more expensive level of care, such as hospitalization.

MAP has been putting patients on a spectrum of risk for relapse since its founding in 2011. Structured data in the patient chart provides some obvious clues. Are patients keeping their appointments? Are they faithfully using a breathalyzer for remote monitoring?

However, there are even more clues in the case notes, narratives, and other types of long-form information that don't lend themselves to discrete data fields on a computer screen. That's where Watson helps out. The cloud-based computer system can make sense of all this shaggy, unstructured data and extract "actionable nuggets," said Levenson.

For example, a case note for an office visit may reveal that the patient had lost not only a job that week, but also a sibling — a big dose of bad news that he or she may not handle constructively. Reading the case note, Watson can tease out these sources of stress and feed that information into MAP's algorithm for relapse risk. A counselor or case manager, in turn, can take the initiative and help the patient cope with these life jolts.

"For people recovering from substance abuse, keeping stress levels low is an absolute must," said Levenson.

Watson, the Multispecialist Computer

IBM describes Watson as a cognitive computer that understands natural language, continuously learns, and mimics human reasoning. In the healthcare field, it is proving to be a multispecialist. IBM's Watson Health division has collaborated with other companies to help oncologists develop treatment plans for various types of cancer, predict hypoglycemic episodes in patients with diabetes, and detect intracranial bleeding.

The supercomputer's foray into addiction medicine comes in the midst of an opioid abuse epidemic that entails prescription painkillers as well as the heroin that people turn to when they can't afford the legal drugs anymore. Levenson estimates that opioids are the primary drug of choice for about 40% of the 40,000 patients that his company has helped manage over the years.

The first customer to use the combination of Watson and MAP's technology platform to better predict substance-abuse relapses will be insurance giant Aetna and its network of addiction specialists, according to Levenson. The ultimate goal, he said, is to make Watson's risk-management insights available to every clinician — including those in primary care — who sees a SAD patient.  The technology fits right in with the much-discussed patient-centered medical home.

"We want to give them a much better tool set to help that patient with a substance abuse disorder," said Levenson. "We have a public health crisis."

Follow Robert Lowes on Twitter @LowesRobert

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as:

processing....