A group of researchers have begun a project in which they use EMRs to identify pregnant mothers who may be at high risk for medical complications.
The researchers, who are being supported by Johns Hopkins University’s Center for Population Health IT (CPHIT), are conducting a pilot using predictive modeling and natural language processing to find indicators of possible risk in the text of records for pregnant Medicaid beneficiaries, according to an article in iHealthBeat.
Maryland, where the pilot is taking place, has had a statewide HIE in place since 2009. The HIE data is useful for spotting trends in the medical histories of individual patients, as it ensures that doctors have the whole story, but obviously, the data doesn’t analyze itself.
That’s where CPHIT comes in. Its job is to find ways to improve public health using existing sources of data.
To find high-risk moms, the researchers are working with CPHIT to find such hints such as whether the mother smokes or lives in an abusive environment. Historically, those beneficiaries don’t receive regular follow-up care, the story notes.
The team of researchers and CPHIT learned which beneficiaries should be considered a risk, in part, by taking a trip to a Johns Hopkins campus in East Baltimore, where a nurse shared warning signs for complicated pregnancies and along the way, shared different phrases which could confuse the search (such as ‘former tobacco user’ or ‘this patient is not a tobacco user’ or ‘this patient lives with a tobacco user.’)
Now armed with this information — and a difficult-to-obtain link between OB, primary care charts and insurance files — the pilot is slowly moving forward. When researchers find mothers who could be at risk for complicated pregnancy, they contact those mothers about receiving care needed to increase the odds of their having a safe, normal pregnancy and delivery.