Marianne Riksheim Stavseth

Artikkel Kunstig intelligens (KI) preger stadig mer av medisinens verden, til både glede og frustrasjon. Men hva skjer når vi slipper teknologien løs på statistikken? Språkmodeller kan skrive kode og forklare analyser på sekunder, men er de til å stole på? Illustrasjon: Tidsskriftet Du sitter med et datasett, men husker ikke syntaksen for en blandet modell i Stata eller hvordan du justerer for multippel testing i R. Tradisjonelt sett har løsningen vært tunge manualer eller timevis på Google.

ABSTRACT Scalable and evidence-based methods for risk assessment are increasingly required in the criminal justice system. The aim of this study was to externally validate the OxRec risk prediction tool using Norwegian data. Predictors were extracted from administrative registers. Outcomes included violent and any reoffending within 1 and 2 years after prison release.

Abstract Women with a clinical history of a severe mental illness (SMI) or multiple mental disorders involving substance use disorder (SUD) might be at increased risk of re-offending after release from prison. This retrospective cohort study merged data from the Norwegian prison release study (nPRIS) with other registry sources. All women released from a Norwegian prison between 2011–2019 were followed for two years after date of release.