Clinical Scorecard: Exploring the Role of Artificial Intelligence in Assessing Risks of Intimate Partner Violence: A PRISMA-ScR Review on Femicide Prevention and Legal Accountability
At a Glance
Category
Detail
Condition
Intimate Partner Violence (IPV)
Key Mechanisms
Artificial intelligence (AI), machine learning, natural language processing for risk assessment and management.
Target Population
Individuals experiencing intimate partner violence and related risks.
Care Setting
Clinical, legal, law enforcement, and social service environments.
Key Highlights
AI techniques are used for detection, classification, and risk stratification in IPV.
The review emphasizes the importance of human oversight in AI applications.
AI cannot predict individual femicide but may assist in risk recognition.
Existing data sources for AI modeling are often incomplete and under-reported.
The review proposes a conceptual model for integrating AI in risk management frameworks.
Guideline-Based Recommendations
Diagnosis
Utilize AI to analyze unstructured data for IPV risk indicators.
Management
Implement human-driven risk-recognition frameworks alongside AI tools.
Monitoring & Follow-up
Ensure continuous assessment of AI applications in IPV contexts.
Risks
Address concerns regarding privacy, discrimination, and systemic inequality in data.
Patient & Prescribing Data
Survivors of intimate partner violence and related forms of gender-based violence.
Focus on survivor-centered approaches and inter-agency collaboration.
Clinical Best Practices
Prioritize documentation and communication across institutional frameworks.
Encourage safe inquiry and confidentiality in healthcare settings.
Facilitate referrals to appropriate services for survivors.
A JAMA Internal Medicine Viewpoint urges clinicians and health systems to verify risk-model inputs before acting on automated breast cancer screening recommendations.
