Researchers at the Centre for Social and Behaviour Change at Ashoka University have developed a unique machine learning (ML) model to can predict intimate partner violence (IPV). It is based on utilizing large datasets and algorithms like random forest. It aims to identify individuals who are at higher risk of domestic violence. It is capable in providing a critical step toward better prevention and intervention strategies.
The model is basically designed to assist health and social workers in identifying vulnerable women. Sneha Shashidhara, cognitive neuroscientist and senior research fellow involved in the project, said that the tool can help field workers to pinpoint such women who are more likely to experience intimate partner violence. The initiative was led by Centre Director Pavan Mamidi. It was supported by senior data researchers Shardul Vaidya and Ishank Daral.
The ML approach offers a distinct advantage by analyzing multiple factors simultaneously. It captures the complexities of psychosocial phenomena like domestic violence. Traditional statistical methods often fail to account for the intricate interactions between various social and environmental factors, but ML algorithms can highlight these relationships more effectively. Shashidhara added that the process does not require strict adherence to traditional rules and allows for a broader analysis of contributing factors such as socioeconomic status, past violence and addiction.
The study titled ‘Using Machine Learning Prediction to Create a 15-question IPV Measurement Tool’ was published in the Journal of Interpersonal Violence earlier this year in January. The model was trained on data from the National Family Health Survey (NFHS) 2015-16 and it included 66,013 women. The survey revealed that nearly 30% of Indian women aged 15-49 experienced domestic violence within a year. The ML model achieved a prediction accuracy of 78% based on key indicators like a husband’s alcohol use, controlling behaviour and a history of violence in the woman’s family.
