Predictive computer models that tap into electronic health records highlight subtle patterns of domestic violence that are easy to miss.
Predictive computer models that tap into electronic health records could help doctors diagnose domestic abuse 10-30 months earlier by highlighting subtle patterns that are easy to miss, according to a study from Children's Hospital Boston.
Domestic abuse accounts for half of all murders of women every year, and is the most common cause of nonfatal injury to U.S. women, according to a statement by Children's Hospital. With patient encounters typically lasting under 10 minutes, doctors often lack time and resources to carefully review and interpret information from multiple visits over many years, which might otherwise tip doctors off to the problem.
In the study, researchers at Children's Hospital analyzed six years of anonymized insurance claims for hospitalizations and emergency-department visits by more than 560,000 patients over age 18. All the patients had visits recorded over at least a four-year period, and 1-3% had an abuse diagnosis on the record, depending on the case definition used.
"Using data from two-thirds of the patients, a computer model was trained to differentiate those who ultimately received a diagnosis of abuse from those who didn't, based solely on their history of visits. The variables associated with abuse (such as a higher number of annual visits, mental health diagnoses, and visits for injury) were used to create a predictive model, which was tested on the remaining third of the patients," according to the statement from the hospital.
"By providing doctors with this additional safety net, we're hoping to minimize the chances that a high-risk patient falls through the cracks," said Ben Reis, PhD, of the Children's Hospital informatics program and division of emergency medicine, in a statement. He led the research, collaborating with colleagues Kenneth Mandi, MD, MPH, and Isaac Kohane, MD, PhD.
"The analysis also found some gender differences in the statistical signs associated with abuse. Alcoholism, poisoning, and injuries from external causes were more predictive of abuse in women than in men, while psychoses, affective disorders, and other mental conditions were more predictive in men than in women," according to the statement.
The researchers developed a color-coded display to help physicians quickly process large amounts of information, presenting an easy-to-read view of the patient's entire diagnostic history and displaying an alert if the history is suggestive of abuse. The goal is to communicate an overview of the patient's history in 10 seconds.
Researchers plan to extend the model to other environments, including HMO and pediatric patients, which could help the researchers spot child abuse. The researchers also plan to develop similar models to predict other conditions that sometimes go undiagnosed, such as depression, or diseases such as diabetes that can be missed in their early stages.
The study was published online September 29 in the British Medical Journal, and funded by the U.S. Centers for Disease Control and Prevention and National Library of Medicine. More information is available at intelligenthistories.org.
InformationWeek has published an in-depth report on e-health and the federal stimulus package. Download the report here (registration required).
We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.
Cybersecurity Strategies for the Digital EraAt its core, digital business relies on strong security practices. In addition, leveraging security intelligence and integrating security with operations and developer teams can help organizations push the boundaries of innovation.