AI combs insurance databases to find women at highest risk of preterm delivery

An AI model developed by a startup in Kentucky is analyzing insurance claims databases to flag expectant mothers at the greatest risk of giving birth to a preterm baby, Wired reported March 26.

Louisville startup Lucina Health began mining records from Passport Health, a nonprofit Medicaid insurer, last year in an attempt to identify more women at a high risk of preterm delivery. Lucina CEO Kevin Bramer told Wired it’s difficult for systems like Passport to comprehensively track patients’ pregnancies—or even identify them in the first place.

“The truth is that moms are in the system, they just can’t find them,” he said. “You have big, chaotic systems that fail to talk to each other.”

Lucina’s AI technology combs through insurance claims to look for less-obvious pregnancy reports, such as those that occur when a woman presents to the emergency room for an unrelated purpose but leaves with an incidental positive pregnancy test. It also delves into patient records and claims and vets that information against other data like prescription medications. Wired reported the approach more than tripled Passport’s identification rate in the first trimester, to around 68 percent.

Once a woman is confirmed as pregnant, the system matches data from her health records and demographics to generate a preliminary risk score. If a mother’s risk is high she’s referred to a nurse to learn more about lowering her chances of a tricky delivery, and Lucina’s AI algorithm accommodates daily updates from the patient’s health records and interviews, generating new risk scores constantly.

Read the full report from Wired below: