Abstract:Objective There is currently inconclusive evidence regarding the relationship between recidivism and mental illness. This retrospective study aimed to use rigorous machine learning methods to understand the unique predictive utility of mental illness for recidivism in a general population (i.e.; not only those with mental illness) prison sample in the United States. Method Participants were adult men (n = 322) and women (n = 72) who were recruited from three prisons in the Midwest region of the United States. Three model comparisons using Bayesian correlated t-tests were conducted to understand the incremental predictive utility of mental illness, substance use, and crime and demographic variables for recidivism prediction. Three classification statistical algorithms were considered while evaluating model configurations for the t-tests: elastic net logistic regression (GLMnet), k-nearest neighbors (KNN), and random forests (RF). Results Rates of substance use disorders were particularly high in our sample (86.29%). Mental illness variables and substance use variables did not add predictive utility for recidivism prediction over and above crime and demographic variables. Exploratory analyses comparing the crime and demographic, substance use, and mental illness feature sets to null models found that only the crime and demographics model had an increased likelihood of improving recidivism prediction accuracy. Conclusions Despite not finding a direct relationship between mental illness and recidivism, treatment of mental illness in incarcerated populations is still essential due to the high rates of mental illnesses, the legal imperative, the possibility of decreasing institutional disciplinary burden, the opportunity to increase the effectiveness of rehabilitation programs in prison, and the potential to improve meaningful outcomes beyond recidivism following release.

Sacrificing Accuracy for Transparency in Recidivism Risk Assessment: The Impact of Classification Method on Predictive Performance

Interpretable Classification Models for Recidivism Prediction

Predicting Recidivism With Machine Learning: An Analysis of Risk Factors and Proposal of Preventions

An algorithm for removing sensitive information: application to race-independent recidivism prediction

A Comparative User Study of Human Predictions in Algorithm-Supported Recidivism Risk Assessment

In Pursuit of Interpretable, Fair and Accurate Machine Learning for Criminal Recidivism Prediction

The Application of Machine Learning to a General Risk–Need Assessment Instrument in the Prediction of Criminal Recidivism

Predicting Criminal Recidivism in Adult Male Offenders

Reducing race-based bias and increasing recidivism prediction accuracy by using past criminal history details

Achieving Fairness through Adversarial Learning: an Application to Recidivism Prediction

Accuracy, Fairness, and Interpretability of Machine Learning Criminal Recidivism Models

An assessment of racial disparities in pretrial decision-making using misclassification models

An impact assessment of machine learning risk forecasts on parole board decisions and recidivism

Almost Politically Acceptable Criminal Justice Risk Assessment

Can nonexperts really emulate statistical learning methods? A comment on “The accuracy, fairness, and limits of predicting recidivism”

The age of secrecy and unfairness in recidivism prediction

Clarifying the relationship between mental illness and recidivism using machine learning: A retrospective study

Fair prediction with disparate impact: A study of bias in recidivism prediction instruments

Case Study: Predictive Fairness to Reduce Misdemeanor Recidivism Through Social Service Interventions

Recidivism early warning model based on rough sets and the improved K-prototype clustering algorithm and a back propagation neural network

Predicting Successful Placements for Youth in Child Welfare with Machine Learning