Hoping to bolster participation in clinical trials of drugs that treat rare diseases, Real Chemistry’s IPM.ai has upgraded its machine learning-based recruitment system to more efficiently locate eligible patients.
The problem it aims to solve is not a new one. For years, researchers have struggled to enroll an adequate number of ideal candidates in trials for drugs set to treat rare diseases.
The number of participants in these trials sometimes don’t reach triple-digits, given that much of the potential patient population is either undiagnosed or misdiagnosed. By comparison, Moderna’s Phase 3 trial for its COVID-19 vaccine included over 30,000 participants, while Pfizer’s topped 40,000.
The disparity remains one of pharma’s most vexing headaches. According to Ron Elwell, the founder and CEO of IPM.ai, the average clinical trial for rare diseases takes 68% longer to conduct than the average trial for non-rare diseases simply because more time is needed to find participants. An astonishing 90% of these trials don’t meet recruitment goals; 81% of patients who want to be included are deemed ineligible, for a variety of reasons.
So as pharma trends towards treatments for increasingly smaller patient populations, companies have been struggling to figure out exactly who to turn to when recruiting trial participants. Although rare disease trials require far fewer people that non-rare trials do, rare trials have to enroll six times as many trial sites because they don’t know where to look, which increases costs.
To begin to fix this, Elwell said the key lies in “expanding the recruiting pool by looking beyond just diagnosed patients.” Specifically, he believes pharma (and its research and tech partners) need to more effectively identify pockets of patient populations.
“If you don’t put the sites where the people are, you’re never going to recruit the people,” he said. “It sounds silly, but it’s actually a huge problem.”
Elwell said the solution to this issue will ultimately be found in artificial intelligence – specifically, machine learning. Like many organizations, IPM.ai has access to anonymous health data, which identify the physicians with whom a patient has interacted as well as show affiliations between those physicians and larger medical institutions. By mapping this data against information they have about the rare disease, IPM.ai believes its clinical trial recruitment platform can expand the number of potential trial participants for rare diseases.
In the past, IPM.ai has worked with Massachusetts company X4 Pharmaceuticals to project the existence of undiagnosed WHIM Syndrome patients in the U.S. By combining its database of 300 million U.S. patients with X4’s genetic lab data – and by overlaying the positive genetic marker for WHIM Syndrome with the database – IPM.ai was able to identify the “positive class” of people to pursue.
“What the machine learning does is look at that positive class across all of the health data and see what the commonalities are and what the treatment patterns are,” Elwell explained. “Then it literally predicts everyone in the U.S. on the likelihood of them having the disease.”