Multimodal AI Model Can Predict Intimate Partner Violence

TUESDAY, March 17, 2026 (HealthDay News) -- A multimodal artificial intelligence (AI) model can identify patients at risk of intimate partner violence (IPV) years before they seek help, according to a study published online March 13 in npj Women's Health.Jiayi Gu, from Brigham and Women's Hospital in Boston, and colleagues present machine learning models for early detection of IPV in clinical settings. The models were developed using a dataset of female patients who sought help at a domestic abuse intervention and prevention center of a major U.S. hospital. The training set included 673 women with IPV and 4,169 control patients, and the testing set included 168 women with IPV and 1,043 control patients. Single-modality and multimodal models were built for different data availability scenarios utilizing tabular clinical data and unstructured clinical notes.The researchers found that all three models had high accuracy, with the Holistic AI in Medicine multimodal fusion model achieving the highest accuracy (area under the receiver operating characteristic curve, 0.88). The multimodal fusion model predicted 80.6 percent of cases in advance, with an average lead time of 3.68 years. The model was validated on patients who did not seek help at the intervention center and among patients from another hospital, with comparable performance."Our research offers proof of concept that AI can support clinicians in flagging possible abuse earlier," principal author Bharti Khurana, M.D., from Mass General Brigham in Boston, said in a statement. "Earlier identification of intimate partner violence and future risk may enable clinicians to intervene sooner and help prevent significant mental and physical health consequences."Abstract/Full Text.Sign up for our weekly HealthDay newsletter

Multimodal AI Model Can Predict Intimate Partner Violence

Related Stories