While preventive medicine has been around for decades, it has recently matured into an era of personalized and predictive medicine in which doctors tailor medical interventions on an individual’s personal genetic risks.
Now researchers at HMS are taking predictive medicine a step further.
“We’re trying to extend the vision of predictive medicine to include all types of information, not just genetic information,” said Ben Reis, HMS assistant professor of pediatrics in the Children’s Hospital Boston informatics program.
Reis, along with HMS professor Isaac Kohane and associate professor Kenneth Mandl, recently investigated whether medical histories could be used to predict a future diagnosis of domestic abuse.
They found that by turning medical records into “intelligent histories,” they were able to predict future abuse on average two years before actual first diagnosis. The British Medical Journal published their results online Sept. 29.
To create the intelligent histories, the team analyzed more than 500,000 medical records, developing a statistical model that weights the type and frequency of doctor visits. In their model, the risk of a future abuse diagnosis increases with the number of visits per year. Visits for injuries or mental health issues weigh more heavily than other complaints.
To make the data comprehensible in the short time doctors have, the team conceived of a “risk gel” (see diagram). They hope to present risk gels to doctors on the opening screen of a patient’s electronic medical history as an array of at-a-glance risk assessments for a variety of conditions including diabetes, depression and abuse.
The intent of this visualization tool, said Reis, is to empower the doctor. A risk gel that predicts a likely future diagnosis should encourage a doctor to take the time to screen her patients carefully. “This is an opportunity for identifying conditions that might fall through the cracks,” he said.e
Students may contact Ben Reis at ben_reis@harvard.edu for more information.
Conflict Disclosure: The authors declare no conflicts.
Funding Sources: The U.S. Centers for Disease Control and Prevention and the National Library of Medicine. The authors are solely responsible for the content of this work.
