Assessing Risk Among Correctional Community Probation Populations: Predicting Reoffense With Mobile Neurocognitive Assessment Software

Abstract

We seek to address current limitations of forensic risk assessments by introducing the first mobile, self-scoring, risk assessment software that relies on neurocognitive testing to predict reoffense. This assessment, run entirely on a tablet, measures decision-making via a suite of neurocognitive tests in less than 30 minutes. The software measures several cognitive and decision-making traits of the user, including impulsivity, empathy, aggression, and several other traits linked to reoffending. Our analysis measured whether this assessment successfully predicted recidivism by testing probationers in a large urban city (Houston, TX, United States) from 2017 to 2019. To determine predictive validity, we used machine learning to yield cross-validated receiveroperator characteristics. Results gave a recidivism prediction value of 0.70, making it comparable to commonly used risk assessments. This novel approach diverges from traditional self-reporting, interview-based, and criminal-records-based approaches, and can also add a protective layer against bias, while strengthening model accuracy in predicting reoffense. In addition, subjectivity is eliminated and time-consuming administrative efforts are reduced. With continued data collection, this approach opens the possibility of identifying different levels of recidivism risk, by crime type, for any age, or gender, and seeks to steer individuals appropriately toward rehabilitative programs. Suggestions for future research directions are provided.

Machine Learning and Criminal Justice: A Systematic Review of Advanced Methodology for Recidivism Risk Prediction

“…Finally, it is important to mention the paper of Haarsma and his working group [ 24 ] in which the authors used the NeuroCognitive Risk Assessment (NCRA), a neurocognitive test-based risk assessment software able to measure key criminogenic factors related to recidivism. In this study, 730 participants in the Harris County Department of Community Supervision and Corrections self-administered the NCRA.…”

“…Four studies used pre-processing to improve the datasets. Two of them preferred a feature selection technique [ 24 , 26 ]. One study used a generic data standardization and then applied feature selection [ 21 ].…”

“…For a better comparison of the studies, it is possible to perform a sorting by the type of recurrence they aim to predict. The sorting leads to four categories: general [ 22 , 23 , 24 , 28 , 31 , 32 ], sexual [ 31 , 32 ], violent [ 22 , 25 , 31 ] and all other recidivism. The last category includes studies that considered a specific type of crime [ 21 ] or referred only to males [ 27 ] or youth [ 26 , 29 , 30 ].…”

“…In this review, for each study, we considered only the ML models that performed better, according to the author’s observation ( Table 2 ). The most used ML model is the logistic regression [ 21 , 30 , 31 , 32 ] and two of its variants, the LogitBoost [ 22 ] and the generalized linear models with ridge and lasso regularization (Glmnet) [ 24 ]. The second most popular model is the random forest [ 23 , 26 , 27 , 33 ].…”

Assessing the Cognitive Contributors to Violence

“…Mean severity of all violent and aggressive incidents and offences combined over 6 months. Codes range from 0 (completely non-violent) to 4 (severe violence) (Gunn & Robertson, 1976;Robertson et al, 1987).…”