Construction of a risk predictor application for domestic violence against women

Objective: to build a risk predictor application for domestic violence against women. Method: This is a population-based, applied methodological research. The research consisted of two distinct phases: the elaboration of a predictive model based on the Neural Network model. For the implementation of the second phase, mobile devices were used (tablet/cell phones), where, through the sampling plan, 58 primary care professionals were selected and instructed to participate in a pilot project, applying the predictor model to women assisted in the service. Results: using as a basis the questionnaire that infers quality of life and the questionnaire that assesses violence against women, the significant variables were obtained through the neural network model and multiple logistic regression. Thus, the probability of a woman suffering domestic violence was calculated. This numerical expression was transcribed into the VCMulher application software. Conclusion: The construction of the risk predictor model proved to be a powerful tool to identify risks/cases of domestic violence against women in the context of primary care.