Multimodal prediction of behavioral and mental health outcomes in incarcerated women - Project Summary/Abstract Research in psychology and neuroscience has helped deepen our understanding of relationships between mental health issues, social-environmental factors, and maladaptive behaviors contributing to risk for incarceration. Such research holds significant economic implications, as the societal cost of crime in the United States is a staggering $3.2 trillion annually, a sum that exceeds annual expenditures across all national health care. While most of these costs are due to men, incarcerated women represent the single fastest growing segment of the criminal legal population over recent decades. Relative to men, women in the legal system display higher rates of mental health problems, including with comorbid substance use, those of which result in a multitude of negative outcomes that impact women and their overall psychosocial functioning, as well as social networks and public health systems, broadly. Despite this, there has been little modern research dedicated to female forensic populations. Indeed, we need prospective work in forensic samples of women aimed at understanding adaptive and maladaptive changes in mental health functioning over time, including risk and protective factors that influence these changes. This type of work is far reaching, as it can be translated to treatment options to provide better long-range outcomes for this population. In prior NIH-supported work, we have examined psychosocial and brain functioning in 588 incarcerated women and female youth. Here we propose to utilize this sample to create a unique longitudinal study in order to develop algorithms that help predict mental health functioning and related outcomes (e.g., recidivism, substance use relapse). Our algorithms will utilize the latest in machine learning techniques and will include multimodal (psychosocial, forensic, and neuroimaging) variables. Utilizing these methods, we intend to identify specific brain measures, in combination with psychosocial and forensic variables, that accompany long-range outcomes of resilience (desistance) and risk (poor global mental health, recidivism, substance use relapse). The translational value of this work will be to clarify informative patterns of data that may indicate positive outcomes. Furthermore, neural measures indicative of specific vulnerability will be identified as targets for treatment and novel intervention strategies. By identifying vulnerabilities and the changes that accompany positive outcomes, we will be closer to understanding ways to improve outcomes for vulnerable women.
