Using Artificial Intelligence to Detect Risk of Family Violence: Protocol for a Systematic Review and Meta-Analysis

Area of Science:

• Computational social science
• Public health informatics
• Artificial intelligence in behavioral science

Background:

• Family violence remains a global issue despite existing prevention strategies.
• Innovative approaches are needed, with artificial intelligence (AI) showing promise.
• AI utilizing voice or text data to detect family violence risk is a growing research area.

Purpose of the Study:

• To systematically review and meta-analyze the accuracy of AI models.
• To differentiate individuals at risk of family violence perpetration from those not at risk.
• To utilize textual or voice signal data for risk assessment.

Main Methods:

• Comprehensive database searches (IEEE Xplore, PubMed, PsycINFO, EBSCOhost, etc.) and grey literature.
• Inclusion of studies on human adults using machine learning for family violence risk assessment (voice or text data).
• Data screening, full-text review, quality assessment (risk of bias), narrative synthesis, and random-effects meta-analysis (Area Under the Curve).

Main Results:

• The literature search has not yet commenced (as of October 2024).
• The review will synthesize current research on AI accuracy in detecting family violence risk.
• Results will be presented via narrative synthesis, tables, and forest plots.

Conclusions:

• This study will clarify the accuracy of machine learning models in identifying high-risk individuals for family violence.
• Findings will inform the development of AI and machine learning tools for violence prevention.
• The research aims to enhance the effectiveness of AI in mitigating family violence.