Although interventions exist to reduce violent crime, optimal implementation requires accurate targeting. We report the results of an attempt to develop an actuarial model using machine learning methods to predict future violent crimes among US Army soldiers.

A consolidated administrative database for all 975 057 soldiers in the US Army in 2004–2009 was created in the Army Study to Assess Risk and Resilience in Servicemembers (Army STARRS). Of these soldiers, 5771 committed a first founded major physical violent crime (murder-manslaughter, kidnapping, aggravated arson, aggravated assault, robbery) over that time period. Temporally prior administrative records measuring socio-demographic, Army career, criminal justice, medical/pharmacy, and contextual variables were used to build an actuarial model for these crimes separately among men and women using machine learning methods (cross-validated stepwise regression, random forests, penalized regressions). The model was then validated in an independent 2011–2013 sample.

Key predictors were indicators of disadvantaged social/socioeconomic status, early career stage, prior crime, and mental disorder treatment. Area under the receiver-operating characteristic curve was 0.80–0.82 in 2004–2009 and 0.77 in the 2011–2013 validation sample. Of all administratively recorded crimes, 36.2–33.1% (male-female) were committed by the 5% of soldiers having the highest predicted risk in 2004–2009 and an even higher proportion (50.5%) in the 2011–2013 validation sample.

Although these results suggest that the models could be used to target soldiers at high risk of violent crime perpetration for preventive interventions, final implementation decisions would require further validation and weighing of predicted effectiveness against intervention costs and competing risks.

Domestic and sexual violence are significant public health problems but little is known about the extent to which men and women with severe mental illness (SMI) are at risk compared with the general population. We aimed to compare the prevalence and impact of violence against SMI patients and the general population.

Three hundred and three randomly recruited psychiatric patients, in contact with community services for ⩾1 year, were interviewed using the British Crime Survey domestic/sexual violence questionnaire. Prevalence and correlates of violence in this sample were compared with those from 22 606 general population controls participating in the contemporaneous 2011/12 national crime survey.

Past-year domestic violence was reported by 27% v. 9% of SMI and control women, respectively [odds ratio (OR) adjusted for socio-demographics, aOR 2.7, 95% confidence interval (CI) 1.7–4.0], and by 13% v. 5% of SMI and control men, respectively (aOR 1.6, 95% CI 1.0–2.8). Past-year sexual violence was reported by 10% v. 2.0% of SMI and control women respectively (aOR 2.9, 95% CI 1.4–5.8). Family (non-partner) violence comprised a greater proportion of overall domestic violence among SMI than control victims (63% v. 35%, p < 0.01). Adulthood serious sexual assault led to attempted suicide more often among SMI than control female victims (53% v. 3.4%, p < 0.001).

Compared to the general population, patients with SMI are at substantially increased risk of domestic and sexual violence, with a relative excess of family violence and adverse health impact following victimization. Psychiatric services, and public health and criminal justice policies, need to address domestic and sexual violence in this at-risk group.