AI Models Developed to Predict Domestic Abuse Risk in Patients

Researchers at Mass General Brigham (MGB) have achieved promising results in utilizing artificial intelligence to assess the risk of domestic abuse among patients. This innovative approach aims to identify individuals at risk of intimate partner violence, which encompasses physical, sexual, or psychological harm in romantic relationships. According to a study published in the journal NPJ Women’s Health on October 13, 2023, AI tools successfully detected potential victims up to four years before they sought assistance from domestic violence treatment centers.

One of the AI models, which incorporates data from a patient’s medical history, vital signs, and demographic information, demonstrated an impressive accuracy rate of 88 percent in predicting intimate partner violence. The study revealed that certain indicators, such as chest pain, analgesic usage, and an increased number of radiology tests, were correlated with a higher likelihood of abuse. Dr. Bharti Khurana, an emergency radiologist and one of the study’s authors, emphasized the importance of early intervention, stating, “The idea is to share resources sooner rather than later. This is something I call proactive screening, instead of waiting for them to disclose [abuse] and then offering services.”

The Centers for Disease Control and Prevention estimates that one in three women and one in six men experience intimate partner violence in their lifetimes. Despite its prevalence, many victims do not disclose their experiences to healthcare providers due to fears of judgment, concerns about their partner’s reaction, or dependency on the abuser. Dr. Khurana observed subtle patterns in the imaging results of patients suffering from intimate partner violence, but noted that radiologists typically do not have the time to analyze past medical records for additional signs of abuse. AI, on the other hand, could efficiently review electronic medical records to identify potential indicators of danger.

AI Models Enhance Early Detection

The study involved training AI models using data from nearly 850 women enrolled in MGB’s domestic abuse intervention and prevention center between 2017 and 2019, as well as from 2021 to 2022. Patients from 2020 were excluded due to the unique challenges presented by the COVID-19 pandemic. The models also drew from a control group of approximately 5,200 patients who had not experienced intimate partner violence but shared similar demographic characteristics with those who had.

The research team developed three distinct AI models: one assessed medications, vital signs, and demographics; another evaluated clinical and radiology notes; and the third combined both approaches. The integrated model achieved the highest accuracy in predicting instances of violence. As the conversation surrounding intimate partner violence is sensitive and complex, caution is advised in how these AI tools are implemented. Dr. Brigid McCaw, former medical director of the Kaiser Permanente Family Violence Prevention Program, emphasized the need for rigorous testing of any domestic violence screening tools and the inclusion of survivors’ perspectives in the development process.

Dr. McCaw highlighted the potential pitfalls of over-reliance on AI, stating, “We need to be very, very cautious about how [AI] information is used for clinicians so that they don’t become over-reliant on algorithms without understanding what the data are that drive the algorithms.”

Future Developments and Considerations

Dr. Khurana’s team continues to refine the models to ensure they accurately identify victims while minimizing false positives, as excessive inaccuracies could undermine trust in the system. “If there are too many false positives, then you lose trust and nobody’s using it,” she noted. The research team plans to extend its work through 2025 and engage with international researchers to enhance the effectiveness of this tool across diverse populations.

“My hope is to bring more institutions in so that we can learn from different ZIP codes, different areas, not only in the US,” Dr. Khurana expressed. As the integration of AI in healthcare evolves, its potential to transform the identification and support of domestic abuse victims represents a significant advancement in proactive medical care.