Limits...
Predicting general criminal recidivism in mentally disordered offenders using a random forest approach.

Pflueger MO, Franke I, Graf M, Hachtel H - BMC Psychiatry (2015)

Bottom Line: Psychiatric expert opinions are supposed to assess the accused individual's risk of reoffending based on a valid scientific foundation.With our statistical approach we were able to correctly identify 58-95% of all reoffenders and 65-97% of all committed offences (AUC = .90).This approach might serve not only for expert opinions in court, but also for risk management strategies and therapeutic interventions.

View Article: PubMed Central - PubMed

Affiliation: Department of Forensic Psychiatry, University Psychiatric Clinics, Wilhelm Klein-Str. 27, CH-4012, Basel, Switzerland. marlon.pflueger@upkbs.ch.

ABSTRACT

Background: Psychiatric expert opinions are supposed to assess the accused individual's risk of reoffending based on a valid scientific foundation. In contrast to specific recidivism, general recidivism has only been poorly considered in Continental Europe; we therefore aimed to develop a valid instrument for assessing the risk of general criminal recidivism of mentally ill offenders.

Method: Data of 259 mentally ill offenders with a median time at risk of 107 months were analyzed and combined with the individuals' criminal records. We derived risk factors for general criminal recidivism and classified re-offences by using a random forest approach.

Results: In our sample of mentally ill offenders, 51% were reconvicted. The most important predictive factors for general criminal recidivism were: number of prior convictions, age, type of index offence, diversity of criminal history, and substance abuse. With our statistical approach we were able to correctly identify 58-95% of all reoffenders and 65-97% of all committed offences (AUC = .90).

Conclusions: Our study presents a new statistical approach to forensic-psychiatric risk-assessment, allowing experts to evaluate general risk of reoffending in mentally disordered individuals, with a special focus on high-risk groups. This approach might serve not only for expert opinions in court, but also for risk management strategies and therapeutic interventions.

Show MeSH

Related in: MedlinePlus

Average partial OOB ensemble mortality (OOBEM, i.e. a hazard-rate) as function of A) prior convictions, B) kind of index offences committed, C) age at index offence, D) diversity of criminal history, and E) substance abuse (according to WHO ICD-10). The variables are ordered by importance. The average partial ensemble mortality is predicted by the Random Survival Forest. All effects are adjusted for one another. dngr: danger to public safety, drugs: violation of narcotic laws, freedm: illegal restraints, life.limb: Offence against life and limb other than homicide, misc: miscellaneous offences, harm: assault, kill: homicide, prpty: property crimes, robb: robbery, sex: sex offence.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4384374&req=5

Fig2: Average partial OOB ensemble mortality (OOBEM, i.e. a hazard-rate) as function of A) prior convictions, B) kind of index offences committed, C) age at index offence, D) diversity of criminal history, and E) substance abuse (according to WHO ICD-10). The variables are ordered by importance. The average partial ensemble mortality is predicted by the Random Survival Forest. All effects are adjusted for one another. dngr: danger to public safety, drugs: violation of narcotic laws, freedm: illegal restraints, life.limb: Offence against life and limb other than homicide, misc: miscellaneous offences, harm: assault, kill: homicide, prpty: property crimes, robb: robbery, sex: sex offence.

Mentions: The number of prior criminal offences was dropped from the model since it was highly correlated with the number of prior convictions as indicated by the cross-validation procedure. The resultant five most important predictors contributed differentially and in a non-linear way to the average partial OOB ensemble mortality (OOBEM, i.e. hazard-rate) (see Figure 2). The number of prior convictions contributed most to OOBEM. In fact, OOBEM increased as a square root function of the number of prior convictions by a slope of 19.4% and an intercept of 23.6% (F1,14 = 135.6, p < .001). Basically, offenders committing rather less serious index offences, such as violation of narcotics law (59%), property crimes (54%), assault (53%), and robbery (49%), were tightly associated with an elevated risk for reoffending (remaining index offences entailed an OOBEM between 30% and 40%). Other predictors such as age below 30 years (<30ys OOBEM ~ 54%, > 30ys OOBEM ~ 30%; F2,22 = 150.5, p < .001), substance abuse (increased OOBEM by 4%), and the diversity of criminal history (cyclic deflection of ±2.9% from baseline at 44.2% OOBEM; F2,3 = 9.1; p = .053) contributed, likewise, significantly. According to time at risk, a 25%, 50%, 75%, and 100% recidivism (hazard) rate was observed after approximately 7, 10, 12, and 14.5 year’s observation period.Figure 2


Predicting general criminal recidivism in mentally disordered offenders using a random forest approach.

Pflueger MO, Franke I, Graf M, Hachtel H - BMC Psychiatry (2015)

Average partial OOB ensemble mortality (OOBEM, i.e. a hazard-rate) as function of A) prior convictions, B) kind of index offences committed, C) age at index offence, D) diversity of criminal history, and E) substance abuse (according to WHO ICD-10). The variables are ordered by importance. The average partial ensemble mortality is predicted by the Random Survival Forest. All effects are adjusted for one another. dngr: danger to public safety, drugs: violation of narcotic laws, freedm: illegal restraints, life.limb: Offence against life and limb other than homicide, misc: miscellaneous offences, harm: assault, kill: homicide, prpty: property crimes, robb: robbery, sex: sex offence.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4384374&req=5

Fig2: Average partial OOB ensemble mortality (OOBEM, i.e. a hazard-rate) as function of A) prior convictions, B) kind of index offences committed, C) age at index offence, D) diversity of criminal history, and E) substance abuse (according to WHO ICD-10). The variables are ordered by importance. The average partial ensemble mortality is predicted by the Random Survival Forest. All effects are adjusted for one another. dngr: danger to public safety, drugs: violation of narcotic laws, freedm: illegal restraints, life.limb: Offence against life and limb other than homicide, misc: miscellaneous offences, harm: assault, kill: homicide, prpty: property crimes, robb: robbery, sex: sex offence.
Mentions: The number of prior criminal offences was dropped from the model since it was highly correlated with the number of prior convictions as indicated by the cross-validation procedure. The resultant five most important predictors contributed differentially and in a non-linear way to the average partial OOB ensemble mortality (OOBEM, i.e. hazard-rate) (see Figure 2). The number of prior convictions contributed most to OOBEM. In fact, OOBEM increased as a square root function of the number of prior convictions by a slope of 19.4% and an intercept of 23.6% (F1,14 = 135.6, p < .001). Basically, offenders committing rather less serious index offences, such as violation of narcotics law (59%), property crimes (54%), assault (53%), and robbery (49%), were tightly associated with an elevated risk for reoffending (remaining index offences entailed an OOBEM between 30% and 40%). Other predictors such as age below 30 years (<30ys OOBEM ~ 54%, > 30ys OOBEM ~ 30%; F2,22 = 150.5, p < .001), substance abuse (increased OOBEM by 4%), and the diversity of criminal history (cyclic deflection of ±2.9% from baseline at 44.2% OOBEM; F2,3 = 9.1; p = .053) contributed, likewise, significantly. According to time at risk, a 25%, 50%, 75%, and 100% recidivism (hazard) rate was observed after approximately 7, 10, 12, and 14.5 year’s observation period.Figure 2

Bottom Line: Psychiatric expert opinions are supposed to assess the accused individual's risk of reoffending based on a valid scientific foundation.With our statistical approach we were able to correctly identify 58-95% of all reoffenders and 65-97% of all committed offences (AUC = .90).This approach might serve not only for expert opinions in court, but also for risk management strategies and therapeutic interventions.

View Article: PubMed Central - PubMed

Affiliation: Department of Forensic Psychiatry, University Psychiatric Clinics, Wilhelm Klein-Str. 27, CH-4012, Basel, Switzerland. marlon.pflueger@upkbs.ch.

ABSTRACT

Background: Psychiatric expert opinions are supposed to assess the accused individual's risk of reoffending based on a valid scientific foundation. In contrast to specific recidivism, general recidivism has only been poorly considered in Continental Europe; we therefore aimed to develop a valid instrument for assessing the risk of general criminal recidivism of mentally ill offenders.

Method: Data of 259 mentally ill offenders with a median time at risk of 107 months were analyzed and combined with the individuals' criminal records. We derived risk factors for general criminal recidivism and classified re-offences by using a random forest approach.

Results: In our sample of mentally ill offenders, 51% were reconvicted. The most important predictive factors for general criminal recidivism were: number of prior convictions, age, type of index offence, diversity of criminal history, and substance abuse. With our statistical approach we were able to correctly identify 58-95% of all reoffenders and 65-97% of all committed offences (AUC = .90).

Conclusions: Our study presents a new statistical approach to forensic-psychiatric risk-assessment, allowing experts to evaluate general risk of reoffending in mentally disordered individuals, with a special focus on high-risk groups. This approach might serve not only for expert opinions in court, but also for risk management strategies and therapeutic interventions.

Show MeSH
Related in: MedlinePlus