Using Machine Learning Prediction to Create a 15-question IPV Measurement Tool.

Area of Science:

• Public Health
• Data Science
• Sociology

Background:

• Domestic violence, particularly intimate partner violence (IPV), is a significant global health concern, with notable prevalence in India.
• Many survivors face barriers in reporting abuse or accessing necessary social support systems.
• Identifying at-risk individuals is crucial for timely intervention and support.

Purpose of the Study:

• To develop and validate machine learning models for predicting IPV using large-scale survey data.
• To create a practical field tool for frontline health workers to identify women at high risk of IPV.
• To facilitate access to support services for survivors of intimate partner violence.

Main Methods:

• Utilized National Family Health Survey data comprising 66,013 women.
• Developed and trained machine learning models to predict instances of IPV.
• Identified the top 15 predictive questions, focusing on non-sensitive inquiries, to form the basis of a field tool.

Main Results:

• Machine learning models achieved a prediction recall of 78% for IPV.
• A set of 15 carefully selected questions demonstrated high predictive accuracy.
• The developed tool enables frontline health workers to screen for IPV risk effectively.

Conclusions:

• Machine learning offers a promising approach for identifying IPV, even with non-sensitive questions.
• The created field tool can empower health workers to proactively support women at risk of IPV.
• This initiative can improve access to care and social support for domestic violence survivors.