By Clinical Informatics News Staff 

May 26, 2015 | Last fall, Mayo Clinic and IBM’s Watson group announced plans to use the Watson supercomputer to match Mayo oncology patients with the appropriate clinical trials. Earlier this month, Nicholas LaRusso, professor of medicine, biochemistry, and molecular biology at Mayo gave an update on the collaboration at the Clinical Research Informatics World event in Boston.*

The pilot program tests a new IBM product that uses Watson’s cognitive computing capabilities to scan a patient’s medical record and suggest the most appropriate clinical trials for that patient.

“Basically an individual—a physician or nurse provider—would provide information to Watson based on what they knew about the patient,” LaRusso told Clinical Informatics News. “Watson would then provide a list of trials a patient was eligible for and not eligible for.”

The program even prompts the clinician for more information—maybe particular tests that are not in the patient record—and updates the trial list as new data are added.

The pilot program focuses on oncology trials for two reasons, LaRusso said. Oncology trials account for 25% of the roughly 1,000 trials at Mayo Clinic, and Mayo oncologists were eager to find new solutions.

“I’m a big advocate… of looking for early adopters,” he said. “Colleagues in oncology were very excited about getting involved in this project.”

The pilot program began with lung, colon, and breast cancer patients and trials at Mayo’s Rochester location, but LaRusso hopes to expand the program to all Mayo sites next year and all disease categories over the next three years.

This is the first time Watson has matched patients with trials, but it wasn’t Watson’s first foray into cancer. Watson was “training” with oncologists at Memorial Sloan Kettering and MD Anderson Cancer Center as early as 2012. The artificial intelligence platform was learning vocabularies and terms associated with cancer, scanning oncology research literature, and ingesting data about cancer drugs.

The new effort links that vast body of knowledge with individual patients’ data stored in their electronic health records (EHRs).

It’s not a completely hands-off process, but it’s still a huge improvement, LaRusso says. “Compared to what it currently takes now to answer the question, ‘Is my patient eligible for a trial?’ this is vastly more efficient.”

And LaRusso predicts that as the offering is refined, the process will become more automated.

“Ultimately, the possibility exists that one could simply enter the patient’s name into Watson, and Watson would review all the information, [and make predictions.] We’re moving in that direction.”

He also says investigators could use the program to gauge feasibility of a trial: with a particular set of inclusion/exclusion criteria, how many patients would fit?

Mayo has been using Watson to match its own patients—over one million patients are seen at Mayo each year—with its own clinical trials. But LaRusso predicts that IBM and Mayo will jointly release the product in two to three months. Then, “any academic medical center is in a position to buy the product and utilize it in conjunction with its own set of patients,” he said.

In order to further facilitate the integration of Watson at medical centers, IBM announced a partnership with Epic, a hospital software provider serving over 350 hospital systems worldwide. IBM says both patients and providers will benefit from Watson’s rapid and thorough analysis of the medical data.

The pilot program is collecting data on how clinical trial matching has fared with Watson’s help. “One of the benefits is increased efficiency, meaning increased rapidity of identifying trials for a given patient,” LaRusso said. “We’re very confident that’s going to be the case. We’re collecting baseline data now… I’m very optimistic, though, that for most of the metrics we’re looking at, Watson is going to provide an advantage. That’s based on the feedback we’re already getting from our oncologists.”

* Clinical Research Informatics World, May 6-7, 2015, Boston, Mass.  The event is hosted by Cambridge Healthtech Institute, the parent company of Clinical Informatics News.