AI Tool Detects Intimate Partner Violence Years Before Victims Speak Up

Researchers at Mass General Brigham have developed an AI tool to detect intimate partner violence (IPV) years before victims seek help. The system, called HAIM, analyses medical records to flag high-risk individuals with remarkable accuracy. Current data shows no other hospital worldwide has tested a similar AI-supported screening method for early IPV detection.

The Holistic AI in Medicine (HAIM) model was designed by a team led by Dr. Bharti Khurana, an emergency radiologist. Her goal was to help clinicians identify and intervene in IPV cases far earlier than usual. The tool processes both structured data, such as diagnoses, and unstructured data, like clinical notes, to improve predictions.

In testing, HAIM achieved an 88% accuracy rate in identifying at-risk individuals. On average, it detected potential IPV cases nearly four years before victims engaged with specialised treatment services. The study also found that patients with chronic pain, mental health disorders, or frequent emergency visits showed higher risk profiles. Conversely, those who regularly attended preventive care, such as mammograms or vaccinations, appeared less likely to experience IPV. The research received funding from the National Institute of Biomedical Imaging and Bioengineering and the National Institutes of Health Office of the Director. Statistics highlight the scale of the issue: over one-third of women and 10% of men will face IPV in their lifetimes. Yet many survivors never disclose abuse during medical visits, making early detection even more challenging.

The HAIM model offers a new way to spot IPV risks long before traditional methods. Its high accuracy and early warning capability could change how healthcare systems address domestic violence. No other facility has implemented a comparable AI tool, leaving this approach as a potential first in the field.