Decision-Making Modeling for Treating Intimate Partner Violence

点击翻译按钮获取中文摘要

Project Summary Intimate partner violence (IPV) is defined as any physical, sexual, or emotional abuse of a current or former intimate partner. Approximately one in four women will experience some form of severe partner violence during their lifetime, and many of these situations result in serious injury or death. Although men can also be victims of IPV, most cases involve female victimization. Gender specific group therapy is widely considered as the standard treatment for IPV, but some participants of these groups do not experience a decrease in violence in response to these treatments. Reports on the effectiveness of standard treatments, as well as research findings suggest that different treatments may be more effective in reducing violence recidivism in certain situations. Many factors influence how participants respond to treatment. These factors include demographics, types of violence, and treatment delivery. Standard IPV treatment does not reflect this variability, and does not provide equal opportunity for recovery to all who are struggling with IPV. If we can determine which subgroups of the population respond similarly to treatment, and which treatments lead to the best outcomes for each subgroup, we will be able to reduce treatment inequalities and improve the quality of life for people suffering from IPV. This study will address this problem in three aims: Aim 1 – We will conduct a systematic review and meta-analysis of existing evidence to characterize treatment outcomes in response to different treatment models. We will examine data from pre-existing research studies to assess levels of violence and relationship satisfaction. This will reveal which treatment is most effective in reducing violence recidivism for each subgroup. Aim 2 – We will use a data-driven approach to systematically investigate patterns of violence to identify subgroups of individuals who respond similarly to treatment. Demographic, socioeconomic, cultural, and age related factors will be considered during subgroup identification. This will involve latent class analysis. Aim 3 – We will develop a decision making tool for clinicians to help them choose between evidence based treatments for each situation. Results from Aims 1 and 2 will be used in the selection of features. This will involve Bayesian and regression networks. We will compare the resulting decision making models to models that are built using traditional features. The outcome of this research will reduce the inequality faced by many individuals who are currently only offered generic treatment for this complex problem, although their circumstances call for tailored solutions.