Using machine learning to identify individuals at risk for intimate partner violence
Using machine learning to identify individuals at risk for intimate partner violence
Publish Date: 2026-03-13 08:36:00
Source Domain: www.news-medical.net
- Researchers from Mass General Brigham have developed AI tools using machine learning to identify individuals at risk for intimate partner violence (IPV) by analyzing electronic medical records (EMRs).
- The AI tools were able to detect IPV up to four years before individuals sought care at a domestic violence treatment center, suggesting the potential for proactive screening.
- The research highlights the importance of earlier identification and intervention in preventing significant mental and physical health consequences associated with IPV.
- The study employed three machine-learning models: a tabular model, a notes model, and a fusion model combining both data types called Holistic AI in Medicine (HAIM), with the fusion model achieving the highest accuracy in predicting IPV risk.
- The fusion model could identify 80.5% of IPV cases on average more than 3.7 years before patients sought care, demonstrating high accuracy in its predictions.
- The study identified additional risk factors for IPV, including mental health disorders, chronic pain, and frequent emergency department visits, while noting that preventive services like mammograms and immunizations were linked to a lower risk.
- The research team acknowledges potential limitations such as the AI tools being developed with data from those who disclosed IPV and the possibility of false negatives in the control group, suggesting room for improvement with larger and more diverse datasets.
