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

Abstract

Recent evolution in the field of data science has revealed the potential utility of machine learning (ML) applied to criminal justice. Hence, the literature focused on finding better techniques to predict criminal recidivism risk is rapidly flourishing. However, it is difficult to make a state of the art for the application of ML in recidivism prediction. In this systematic review, out of 79 studies from Scopus and PubMed online databases we selected, 12 studies that guarantee the replicability of the models across different datasets and their applicability to recidivism prediction. The different datasets and ML techniques used in each of the 12 studies have been compared using the two selected metrics. This study shows how each method applied achieves good performance, with an average score of 0.81 for ACC and 0.74 for AUC. This systematic review highlights key points that could allow criminal justice professionals to routinely exploit predictions of recidivism risk based on ML techniques. These include the presence of performance metrics, the use of transparent algorithms or explainable artificial intelligence (XAI) techniques, as well as the high quality of input data.

When do drugs trigger criminal behavior? a machine learning analysis of offenders and non-offenders with schizophrenia and comorbid substance use disorder

“…On that note, the advanced statistical approach we chose reached a level of complexity difficult to encapsulate in detail and clarity within the boundaries of a non-statistical publication. However, readers interested in further detail are encouraged to contact the authors, as the brevity of the methods section should not hamper reproducibility of results ( 32 ).…”

Model Building in Forensic Psychiatry: A Machine Learning Approach to Screening Offender Patients with SSD

“…Thus far, the application of ML has been rare in the field of forensic psychiatry. Previous studies have mainly explored heterogenous forensic populations, e.g., for purposes of recidivism risk prediction, and have not focused on patients with SSD in particular [ 41 , 42 ]. The authors’ former publications, which evaluated a more homogenous population of offender patients with SSD exclusively, mainly focused on providing a better understanding of complex, multifactorial phenomena, such as stress, criminal recidivism, migration experience, self-harm, and aggressive behavior [ 6 , 18 , 30 , 32 , 37 ].…”

Artificial Intelligence in Violence Risk Assessment: Addressing Racial Bias and Inequity

“…The prediction of recidivism in the criminal justice system has been a popular target for AI. In a recent paper, Travaini et al9 performed a meta-analysis of studies using machine learning to predict recidivism risk among criminal defendants, with generally positive results. The area under the curve (AUC) for general recidivism for all studies examined was 0.74, representing good predictive validity and an improvement over traditional risk-assessment tools 9…”

“…The area under the curve (AUC) for general recidivism for all studies examined was 0.74, representing good predictive validity and an improvement over traditional riskassessment tools. 9 Algorithms that predict violence or recidivism may also be used in sentencing or parole hearings. The Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) system is widely used to advise judges in sentences in the United States and has been upheld by the Wisconsin Supreme Court.…”