A machine-learning algorithm detected potential signs of colorectal cancer (CRC) in patients identified as high-risk who had missed a routine colonoscopy, according to a study led by integrated health system Geisinger and vendor partner Medial EarlySign.
The findings, published this month in NEJM Catalyst Innovations in Care Delivery, present a noninvasive method to increase screening among those who may have CRC, according to Pennsylvania-based Geisinger.
Despite evidence of the benefits of regular CRC screening and significant efforts among providers and healthcare systems to increase screenings, approximately 32 percent of age-eligible adults in the United States do not follow current CRC screening guidelines, according to the National Cancer Institute. Serious illness and death from CRC can be prevented if asymptomatic polyps and other early-stage cancers are detected and treated early.
In the study, Geisinger identified a group of 25,610 patients who were overdue for CRC screening, and used a machine-learning algorithm to flag those at highest risk for developing cancer. The algorithm, developed by EarlySign, identified patients as high-risk by analyzing age, gender, and a recent outpatient complete blood count (CBC). A nurse then called the patients to inform them of their risk and offer to schedule a colonoscopy.
Of the patients flagged as high-risk, 68 percent were scheduled for a colonoscopy, and of those, approximately 70 percent had a significant finding.
“When carefully implemented and supported by healthcare providers, machine learning can be a low- cost, noninvasive supplement to other colorectal cancer screening efforts,” said Keith Boell, D.O., chief quality officer for population initiatives at Geisinger and a co-author of the study, in a statement. “This technology can act as a safety net, potentially preventing missed or delayed diagnosis among some patients who may already have undiagnosed signs of disease.”
Israel-based EarlySign said its AlgoMarkers and predictive solutions allow for early detection of complications from serious disease and help clients more accurately identify and prioritize patients for multiple conditions for interventions to halt or prevent the serious complications from the onset of disease.
