Mayo Clinic finds IBM Watson increases enrollment of clinical trials

By Laura Lovett

The Mayo Clinic has just released the results of a new study, which show an 80 percent increase in enrollment of clinical trials for breast cancer when using IBM’s Watson for Clinical Trial matching system. 

The Watson system uses artificial intelligence to analyze unstructured information and pull out insights from the data. 

“Watson is able to give us faster, better matching of patients to potential clinical trials that our oncologists wouldn’t have otherwise be able to see — and I sit with our oncologists who work on this kind of thing,” Christopher Ross, CIO at the Mayo Clinic told MobiHealthNews. “[The oncologist's] day, every day, is to sit with stage four breast cancer patients who have no hope and I don’t know how she does it. But she’s excited about that work because of the true potential to get better results from those patients every day.”

In addition to the uptake in participants, the trial showed that the technology could significantly reduce the time it takes to screen an individual patient for clinical trial matches. The system has the ability to generate a ranked list of relevant trials for each patient without making clinicians read through EMRs or long lists of eligibility criteria, according to the IBM webpage. 

“I’m much more a fan of little AI than big AI. I’m much more interested in simple utilities than moonshots,” Ross said. “Hopefully the best and most frequent use of what we’re using AI for is natural language processing, and then we use it for data enrichment, so we pour all this data out of real-time monitoring into this data lake that we can do interesting stuff with.”

The technology is able to integrate clinical notes, pathology reports, labs, and more data about how to include or exclude a participant. As a result of this trial, the Mayo Clinic has announced a new plan to continue their partnership with the Watson system while targeting trials for additional cancers, including breast, lung colon, and GI cancers. 

"We’re working on clinical trial matching and not clinical trial managing,” Ross said. “It involves recruitment and the matching process. We’re also working with IBM looking for opportunities. We’ve started with the one patient at a time thing: I have a patient with these characteristics, let me match that up against what’s available. And so Watson reads the patient record, Watson reads all of, Watson has read all of the literature on the planet and then tries to figure out a match on a probabilistic basis between all those things.”

Clinical trial management is not something that has historically been a top priority for the Mayo Clinic, but that could be changing, according to Ross. 

“Clinical trial management is not an area we’ve prioritized with AI,” Ross said. “There’s so much lower-hanging fruit, AI might be a little bit of a cannon where a fly swatter will do for that particular case.”

Earlier this week, Ross discussed the future of AI at HIMSS18. He said that when the Mayo Clinic did a review of all of their AI projects they found upwards of 130 applications — most of them for smaller initiatives and not what Ross referred to as “moon shot” projects.