👨 AI Model Predicts Risks of Domestic Violence

According to the WHO, nearly one in three women has experienced sexual or physical violence from a partner at least once.

Even so, these women often do not seek help from either the police or social services. And when they end up in the hospital, many also hide the real cause of their injuries.

American scientists have developed an AI tool that identifies signs of domestic violence based on a patient's complaints and medical history—in more than three out of four women who are actually at risk.

The model was trained on medical records of more than 800 women who sought hospital care due to intimate partner violence between 2017 and 2023. The dataset included their radiology reports, medical histories, and prescriptions. This data was compared with the medical histories of more than 5,200 women from a control group.

❗️Researchers found that victims of domestic violence more often report depression and other psychological and emotional disorders, as well as chest pain; they get X-rays and CT scans for injuries to the face, neck, and upper extremities more frequently, and are also prescribed painkillers.

🔍 The model was tested on real women hospitalized due to episodes of violence. It can identify signs of physical and sexual abuse in intimate relationships 3–5 years earlier than when, statistically, patients would be expected to seek help on their own.

Should such systems be used in hospitals?

❤️ — Yes, definitely
🔥 — No, it's a violation of privacy...