A new study from Purdue University and the University of Chicago introduces an innovative framework combining machine learning (ML) and queueing theory to enhance the management of incarceration-diversion programs. These programs aim to reduce recidivism by addressing root causes such as substance use and mental health issues. The research focuses on optimizing program size and staffing through a Decision Support System (DSS) that predicts program census and outcomes while offering insights into program dynamics and counterfactual scenarios.
At the core of this effort is a user-friendly web app designed to help program managers visualize census data by counties and demographic groups. The app enables real-time simulations for adjusting admission criteria and scaling programs to new locations. This tool has already been deployed in Illinois, where simulation times were reduced from 8 minutes to just 15 seconds, streamlining decision-making processes.
Learn more about this transformative approach: https://ojs.aaai.org/index.php/AAAI/article/view/30330/32355
Leave a Reply