AI shows success in reducing premature births

Kentucky has one of the highest rates of preterm birth in the nation, according to the Centers for Disease Control and Prevention.

High rates of smoking, substance abuse, obesity, diabetes and hypertension have all been major factors for keeping the preterm birth rate within the state, 11.4% well above the national average of 9.85%.

A big challenge has been identifying mothers at risk early enough in their pregnancy in order to design and implement effective intervention strategies, said Dr. Larry Griffin, an OB-GYN and medical director for women's health for Passport Health Plan, a state-based Medicaid managed-care insurer.

Out of an average of 7,000 pregnancies a year, Griffin said Passport Health had missed about 1,500 to 2,000 pregnancies until the time of delivery, meaning the women likely did not receive care and the insurer was unable to respond to those patients' needs.

So, in May, Passport Health started a pilot that uses artificial intelligence to identify mothers' at-risk factors. The insurer then deploys a team of nurse care managers who create a personalized care plan for the mother to follow throughout her pregnancy. Nurse care managers also partner with community resources to address any social determinants for the mother, such as housing, transportation or food issues.

The plan uses a maternity analytics platform developed by Kentucky-based data management solutions firm Lucina Health to stratify medical and socio-economic risk factors to allow Passport to pinpoint how best to help those at greatest risk.

The platform uses multiple data sources to identify which patients are pregnant, including medical and pharmacy claims not related to pregnancy as well as public data on community socio-demographics. That information is then aggregated and analyzed daily to update the plan if a patient's situation changes.

Griffin said initial results have been positive in the program's first six months. So far the platform was able to identify 85% of pregnant mothers within their first two trimesters. Up to 700 mothers are in case management for being at high risk. Griffin said the platform has helped Passport develop interventions that have been led to a 13% reduction in preterm births for that group.

The preliminary results are encouraging, Griffin said, adding that there are early signs that the program can reduce lengths-of-stay for babies in the neonatal intensive care unit, as well as lower healthcare costs. According to the March of Dimes, the average daily cost of a stay in the NICU was between $2,500 and $3,000 a day with an average length of stay of about 13 days; the average cost per patient was $76,000.

Griffin expects the next two years of the pilot to yield information on longer-term healthcare utilization of both mothers and infants. That is also likely to affect overall healthcare costs as preterm birth is tied to higher rates of both infant and maternal hospitalizations throughout the first year of life.

Passport plans to expand the use of AI in improving interventions for other risk categories such as substance use disorder.

"If this works the way I believe it will I'm hoping it will be deployed nationwide and everyone can benefit from the technology we're developing," Griffin said. "I would like to see the preterm birth rate down to 7%—to numbers we haven't even dreamed of before."