New AI models could support early detection of domestic violence risks.
US researchers have developed an AI system designed to help doctors identify patients who may be at risk of intimate partner violence. The tool analyzes hospital data to detect patterns associated with abuse, potentially allowing health professionals to intervene early.
Intimate partner violence refers to abuse by a current or former partner that can cause serious injuries, chronic pain, and long-term mental health problems. According to the European Commission, 18% of women with a partner reported experiencing physical or sexual violence by their partner in 2021.
The study, published in the journal Nature, looked at the hospital records of about 850 women who had experienced intimate partner violence and a control group of more than 5,200 similar patients. The researchers used the data to train three different machine learning systems to detect patterns associated with abuse.
One model analyzed structured hospital data such as age and medical history. The second model looked at written clinical records, such as physician observations and radiology reports. The third system combined both data types and achieved the strongest results, accurately identifying risks in 88% of cases.
Researchers found that the system could alert some patients to potential abuse more than three years before they later entered a hospital-based intervention program. By analyzing large datasets, the tool detects patterns of physical trauma associated with abuse and alerts clinicians so they can approach the problem sensitively and provide support.
Want to know more about AI, technology and digital diplomacy? If so, Ask the Diplo chatbot a question!
