A team of researchers from Indiana University and the Regenstrief Institute, both based in Indianapolis, developed predictive models to identify primary care patients in need of social services.
To create the models, the researchers integrated clinical data with 48 socioeconomic and public health indicators from community-level data sources, according to study results published in the Journal of the American Medical Informatics Association. They used this data to build decision models that predict the need for mental health, dietician and social work referrals among patients receiving care at a safety-net hospital.
Here are four things to know about the project.
1. The models that predicted the need for any social service referrals reported positive predictive values between 65 and 73 percent.
2. The models that predicted the need for mental health or dietitian referrals yielded sensitivity, specificity and accuracy measures ranging between 60 and 75 percent.
3. The models that predicted the need for social work and other social determinants of health service referrals reported specificity and accuracy measures between 67 and 77 percent, while sensitivity scores were between 50 and 63 percent.
4. The researchers argued that early intervention into community-level factors that influence health — such as food availability and adequate housing — will help to prevent costlier medical issues later on.
"The need for various social service referrals can be predicted with considerable accuracy using a wide range of readily available clinical and community data that measure socioeconomic and public health conditions," the study authors concluded. "While the use of [social determinants of health] did not result in significant performance improvements, our approach represents a novel and important application of risk predictive modeling."