Clinical Report: AI to Identify Gender-Based Violence in Brazilian Health Records

Overview

ClarIA is an AI tool developed by Vital Strategies and the Federal University of Juiz de Fora to identify possible gender-based violence (GBV) in clinical notes and medical histories. In a pilot analysis of 13,000 documented GBV cases in Recife, women’s health records showed increased healthcare visits an average of 92 days before GBV was recorded, with patterns including headaches and mental health complaints.

Background

The article describes rising femicide and GBV in Brazil and reports that violence may be underreported or documented only after it becomes severe. Pedro de Paula of Vital Strategies estimates that only 20% of violence entering clinics or hospitals is reported, and that 80% of reported cases are recorded too late. The article also describes cases in which deaths were misclassified in national health data and notes that some patients did not explicitly disclose violence during healthcare visits. PAHO’s 2024 toolkit provides guidance for documenting violence against women and girls in health services and strengthening related administrative data.

Data Highlights

Key Findings

• ClarIA examines clinical notes and medical histories for patterns that may indicate unreported GBV and alerts healthcare professionals to possible cases.
• The Recife pilot involved 13,000 formally documented GBV cases; women’s clinic visits increased an average of 92 days before violence was recorded.
• Patterns identified in the records included increased complaints of headaches, mental health problems, and other issues.
• ClarIA uses semantic patterns across clinical narratives rather than triggering alerts from isolated terms such as “headache” or “depression.”
• Amber alerts indicate signs, symptoms, and complaints resembling those in records of women with a clear violence history; red alerts concern women with prior disclosure or professional identification of violence.
• PAHO’s 2024 toolkit offers step-by-step guidance for documenting intimate partner and sexual violence in health services and improving administrative datasets.

Clinical Implications

The article presents ClarIA as a means of flagging possible GBV when clinical records contain patterns that may not be apparent from individual symptoms or explicit disclosure. PAHO’s toolkit separately supports health services in documenting violence cases and recording their response; the supplied material does not report clinical outcome data for ClarIA.

Conclusion

The Recife pilot described in the article found that healthcare use often increased before GBV was formally recorded, and ClarIA was designed to detect patterns in those records. The source material describes the tool’s approach and alerts but does not establish its effect on intervention timing or patient outcomes.

Related Resources & Content

1. MDSpire News, Year not provided -- Using AI to Identify Gender-Based Violence in Brazilian Health Records and Enable Earlier Intervention
2. Pan American Health Organization, 2024 -- New toolkit on documenting cases of violence against women and girls and recording health sector response
3. Frontiers in Digital Health, 2026 -- Artificial intelligence in intimate partner violence risk pathways: a PRISMA-ScR review of femicide prevention and medico-legal accountability
4. Journal of Medical Internet Research (JMIR) — Prioritizing Patient Safety: The Impact of Pharmaceutical AI on Healthcare in Brazil
5. MDSpire News — AI May Help Close Women's Health Gap
6. ASCO Publications — Translating real-world data into actionable insights: An AI-powered model for prostate cancer survival prediction in a Brazilian cohort.
7. New toolkit on documenting cases of violence against women and girls and recording health sector response
8. Electronic Health Record–Based Screening for Intimate Partner Violence
9. From Reactive to Proactive Care: Using AI to Detect Gender-Based Violence in Brazil