Comprehensive psychiatry · 2026
Forensic outcome in schizophrenia: It's the system, not the symptoms.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Machetanz Lena, Hofmann Andreas B, Dörner Marc, Seifritz Erich, Homan Philipp, Kirchebner Johannes
Аннотация
BACKGROUND: Schizophrenia spectrum disorders (SSD) are associated with increased risks of criminal behavior, especially if mediated by factors such as substance use. However, it remains unclear whether these factors retain their predictive relevance once broader social and systemic conditions are considered. To date, comparisons between forensic (FPP) and general psychiatric (GPP) SSD patients that systematically test the weight of these factors in complex contexts are scarce. We sought to evaluate whether established clinical risk factors for criminal behavior hold explanatory power once embedded in wider contextual frameworks. Group membership was used as a proxy for such forensic trajectories.
METHODS: A retrospective study was conducted using data from 740 patients (370 FPP, 370 GPP) diagnosed with SSD receiving treatment at one institution in Switzerland. Several machine learning algorithms were tested. Gradient Boosting emerged as the most suitable model. Performance metrics such as balanced accuracy, area under the curve (AUC), sensitivity, and specificity were used for model evaluation. Key predictive variables were ranked based on their influence.
RESULTS: Gradient Boosting achieved a balanced accuracy of 77.5% and an AUC of 0.85 (analysis excluding item 'olanzapine-equivalent dose at discharge', which was identified as a potential downstream marker of institutional placement; primary model with all items: 81.6% and 0.88, respectively), outperforming other algorithms in discriminating between the groups. Notable predictors of a forensic-psychiatric course included social isolation across life span and limited mental health care system integration, while psychopathology did not emerge as a relevant predictor.
CONCLUSION: When comprehensively comparing forensic and general psychiatric SSD patients, social isolation, antipsychotic dosage, and mental health system integration emerge as the primary discriminators, overshadowing well established risk factors. Preventing forensic pathways in SSD requires strengthening social networks and system integration, marking a paradigm shift away from symptom-centered risk models.
METHODS: A retrospective study was conducted using data from 740 patients (370 FPP, 370 GPP) diagnosed with SSD receiving treatment at one institution in Switzerland. Several machine learning algorithms were tested. Gradient Boosting emerged as the most suitable model. Performance metrics such as balanced accuracy, area under the curve (AUC), sensitivity, and specificity were used for model evaluation. Key predictive variables were ranked based on their influence.
RESULTS: Gradient Boosting achieved a balanced accuracy of 77.5% and an AUC of 0.85 (analysis excluding item 'olanzapine-equivalent dose at discharge', which was identified as a potential downstream marker of institutional placement; primary model with all items: 81.6% and 0.88, respectively), outperforming other algorithms in discriminating between the groups. Notable predictors of a forensic-psychiatric course included social isolation across life span and limited mental health care system integration, while psychopathology did not emerge as a relevant predictor.
CONCLUSION: When comprehensively comparing forensic and general psychiatric SSD patients, social isolation, antipsychotic dosage, and mental health system integration emerge as the primary discriminators, overshadowing well established risk factors. Preventing forensic pathways in SSD requires strengthening social networks and system integration, marking a paradigm shift away from symptom-centered risk models.