Every provider knows that finding enough patients to take part in clinical trials is tough and involves a painstaking search.
An estimated 40 percent of clinical trials close early because they do not have enough patients, according to Bobbie Rimel, MD, a gynecologic, oncologist and associate director for Gynecological Clinical Trials at Cedars-Sinai Medical Center.
Rimel designs the clinical studies and finds the patients.
Clinical trials are necessary to test cutting-edge treatment and for improving patient outcomes. But with close to half of trials fizzing out prematurely, the return on investment is not always there.
Rimel found a way through artificial intelligence to search through records for patient matching for much better recruitment and ROI.
Working with a vendor, Deep 6 AI, Rimel uses AI software to read patient charts for the correct terms needed to meet study criteria.
“It’s designed to be able to look into the record and pull the language,” she said. “It’s designed to take inclusion and exclusion criteria and pull out the bits of language.”
AI finds patients as close to the same as possible. This is difficult because not all doctors document in the same way. Many use abbreviations, and not all of these are the same.
The software cuts through the different spellings and abbreviations to pull out the information. Not only that, the AI software learns from past queries. It can then search using the same specific references for breast cancer and the type of cancer.
“That’s what makes it incredibly powerful,” Rimel said. “It knows ‘I think this is what you mean here.’ That’s incredibly powerful. Say it nine different ways, in a different font; it’s close enough.”
The correct inclusion of all patients with the same type of cancer also ensures that those who fit the criteria are given the opportunity to take part in a clinical trial — and that may the biggest reason of all to apply AI in this case.
“This is a pretty powerful opportunity for hospitals,” she said.
