AI Enabled Community Supervision for Criminal Justice Services

This project involves a researcher-practitioner partnership between a research team led by Purdue University and Tippecanoe County, IN. The project has three goals: (1) develop a novel AI-based system — AI-SMS (AI-based support and monitoring system) — to support offender reentry; (2) deploy that system with Tippecanoe County Community Corrections (TCCC); and (3) analyze the use and impact of AI-SMS and modify the system based on that analysis.

AI-SMS will provide supervising officers and offenders early warning of risky behavior on an offender’s part and the means and mechanisms for reducing that behavior. Offenders will be provided a smartphone and a tracking-health-related wearable device or bracelet. Officers and practitioners (e.g., clinicians) and caseworkers assigned to that offender will be provided smartphones/tablets with specific dashboards (user interfaces) that would be used to communicate with the offender.

AI-SMS will apply AI and Machine Learning (ML) techniques to data provided by the offender’s wearable device and smartphone to identify risky behaviors and to recommend interventions. A total of 250 offenders will be randomly selected to participate in the study analyzing AI-SMS deployment of which 125 will be assigned to the control group and 125 will be assigned to the treatment group.

To be selected for the study, an individual must be (1) age 18 or older; (2) released from a correctional facility within the past 30 days; (3) under supervision for a minimum of one year; (4) assessed as being moderate to high risk using the Indiana Risk Assessment tool; (5) capable of conversational English; and (6) willing and able to use a smartphone and wearable health monitor. Data variables will include (1) technical violations, (2) revocations, (3) re-arrest for new crime, (4) quality of office-participant interactions, (5) participant goal attainment, and (6) community stability (i.e., securing employment, stable housing, connection to treatment and services support, involvement in positive social networks, etc.)

Both parametric and nonparametric tests will be used in the analysis. The first 18 months of the project will focus on design and development. The next 12-18 months will focus on deployment and assessment. The final 12-18 months will focus on data analysis, revising and upgrading AI-SMS. The AI-SMS code will be made available in an online repository using either Bitbucket or GitHub. The applicant proposes to upload the final version of the code and its final report to National Archive of Criminal Justice Data (NACJD).

Note: This project contains a research and/or development component, as defined in applicable law," and complies with Part 200 Uniform Requirements - 2 CFR 200.210(a)(14). ca/ncf