Parole Review Individualized Support Model (PRISM)

Posters

Keyword: Criminal Justice

Abstract: This study develops a transparent, fairness-aware machine learning model to enhance parole decision-making. Prioritizing dynamic factors over criticized static approaches like COMPAS, the model promotes objectivity, equity, and rehabilitative goals through empirically outlined validation and openly disclosed methodology for external review and continuous improvement in parole outcomes.

• Drew Melancon (Presenting Author)
• Tierney Huppert (Author)
• Mahmood Mohamed (Author)
• Pevitr Bansal (Author)
• Tarika Daftary-Kapur (Author)