Search

Abstract Although reducing criminal outcomes in individuals with mental illness have long been a priority for governments worldwide, there is still a lack of objective and highly accurate tools that can predict these events at an individual level. Predictive machine learning models may provide a unique opportunity to identify those at the highest risk of criminal activity and facilitate personalized rehabilitation strategies. Therefore, this systematic review and meta-analysis aims to describe the diagnostic accuracy of studies using machine learning techniques to predict criminal and violent outcomes in psychiatry. We performed meta-analyses using the mada, meta, and dmetatools packages in R to predict criminal and violent outcomes in psychiatric patients (n = 2428) (Registration Number: CRD42019127169) by searching PubMed, Scopus, and Web of Science for articles published in any language up to April 2022. Twenty studies were included in the systematic review. Overall, studies used single-nucleotide polymorphisms, text analysis, psychometric scales, hospital records, and resting-state regional cerebral blood flow to build predictive models. Of the studies described in the systematic review, nine were included in the present meta-analysis. The area under the curve (AUC) for predicting violent and criminal outcomes in psychiatry was 0.816 (95% Confidence Interval (CI): 70.57–88.15), with a partial AUC of 0.773, and average sensitivity of 73.33% (95% CI: 64.09–79.63), and average specificity of 72.90% (95% CI: 63.98–79.66), respectively. Furthermore, the pooled accuracy across models was 71.45% (95% CI: 60.88–83.86), with a tau squared (τ2) of 0.0424 (95% CI: 0.0184–0.1553). Based on available evidence, we suggest that prospective models include evidence-based risk factors identified in prior actuarial models. Moreover, there is a need for a greater emphasis on identifying biological features and incorporating novel variables which have not been explored in prior literature. Furthermore, available models remain preliminary, and prospective validation with independent datasets, and across cultures, will be required prior to clinical implementation. Nonetheless, predictive machine learning models hold promise in providing clinicians and researchers with actionable tools to improve how we prevent, detect, or intervene in relevant crime and violent-related outcomes in psychiatry.

Abstract The placebo effect across psychiatric disorders is still not well understood. In the present study, we conducted meta-analyses including meta-regression, and machine learning analyses to investigate whether the power of placebo effect depends on the types of psychiatric disorders. We included 108 clinical trials (32,035 participants) investigating pharmacological intervention effects on major depressive disorder (MDD), bipolar disorder (BD) and schizophrenia (SCZ). We developed measures based on clinical rating scales and Clinical Global Impression scores to compare placebo effects across these disorders. We performed meta-analysis including meta-regression using sample-size weighted bootstrapping techniques, and machine learning analysis to identify the disorder type included in a trial based on the placebo response. Consistently through multiple measures and analyses, we found differential placebo effects across the three disorders, and found lower placebo effect in SCZ compared to mood disorders. The differential placebo effects could also distinguish the condition involved in each trial between SCZ and mood disorders with machine learning. Our study indicates differential placebo effect across MDD, BD, and SCZ, which is important for future neurobiological studies of placebo effects across psychiatric disorders and may lead to potential therapeutic applications of placebo on disorders more responsive to placebo compared to other conditions.

Background: Selected gene mutations are routinely used to guide the selection of cancer drugs for a given patient tumour. Large pharmacogenomic data sets were introduced to discover more of these single-gene markers of drug sensitivity. Very recently, machine learning regression has been used to investigate how well cancer cell line sensitivity to drugs is predicted depending on the type of molecular profile. The latter has revealed that gene expression data is the most predictive profile in the pan-cancer setting. However, no study to date has exploited GDSC data to systematically compare the performance of machine learning models based on multi-gene expression data against that of widely-used single-gene markers based on genomics data. Methods: Here we present this systematic comparison using Random Forest (RF) classifiers exploiting the expression levels of 13,321 genes and an average of 501 tested cell lines per drug. To account for time-dependent batch effects in IC50 measurements, we employ independent test sets generated with more recent GDSC data than that used to train the predictors and show that this is a more realistic validation than K-fold cross-validation. Results and Discussion: Across 127 GDSC drugs, our results show that the single-gene markers unveiled by the MANOVA analysis tend to achieve higher precision than these RF-based multi-gene models, at the cost of generally having a poor recall (i.e. correctly detecting only a small part of the cell lines sensitive to the drug). Regarding overall classification performance, about two thirds of the drugs are better predicted by multi-gene RF classifiers. Among the drugs with the most predictive of these models, we found pyrimethamine, sunitinib and 17-AAG. Conclusions: We now know that this type of models can predict in vitro tumour response to these drugs. These models can thus be further investigated on in vivo tumour models.

Abstract Background Post-viral symptoms have long been known in the medical community but have received more public attention during the COVID-19 pandemic. Many post-viral symptoms were reported as particularly frequent after SARS-CoV-2 infection. However, there is still a lack of evidence regarding the specificity, frequency and persistence of these symptoms in comparison to other viral infectious diseases such as influenza. Methods We investigated a large population-based cohort based on German routine healthcare data. We matched 573,791 individuals with a PCR-test confirmed SARS-CoV-2 infection from the year 2020 to contemporary controls without SARS-CoV-2 infection and controls from the last influenza outbreak in 2018 and followed them up to 18 months. Results We found that post-viral symptoms as defined for COVID-19 by the WHO as well as tissue damage were more frequent among the COVID-19 cohort than the influenza or contemporary control cohort. The persistence of post-viral symptoms was similar between COVID-19 and influenza. Conclusion Post-viral symptoms following SARS-CoV-2 infection constitute a substantial disease burden as they are frequent and often persist for many months. As COVID-19 is becoming endemic, the disease must not be trivialized. Research should focus on the development of effective treatments for post-viral symptoms.

BackgroundLong-term health sequelae of the Coronavirus Disease 2019 (COVID-19) are a major public health concern. However, evidence on post-acute COVID-19 syndrome (post-COVID-19) is still limited, particularly for children and adolescents. Utilizing comprehensive healthcare data on approximately 46% of the German population, we investigated post-COVID-19-associated morbidity in children/adolescents and adults.Methods and findingsWe used routine data from German statutory health insurance organizations covering the period between January 1, 2019 and December 31, 2020. The base population included all individuals insured for at least 1 day in 2020. Based on documented diagnoses, we identified individuals with polymerase chain reaction (PCR)-confirmed COVID-19 through June 30, 2020. A control cohort was assigned using 1:5 exact matching on age and sex, and propensity score matching on preexisting medical conditions. The date of COVID-19 diagnosis was used as index date for both cohorts, which were followed for incident morbidity outcomes documented in the second quarter after index date or later.Overall, 96 prespecified outcomes were aggregated into 13 diagnosis/symptom complexes and 3 domains (physical health, mental health, and physical/mental overlap domain). We used Poisson regression to estimate incidence rate ratios (IRRs) with 95% confidence intervals (95% CIs). The study population included 11,950 children/adolescents (48.1% female, 67.2% aged between 0 and 11 years) and 145,184 adults (60.2% female, 51.1% aged between 18 and 49 years). The mean follow-up time was 236 days (standard deviation (SD) = 44 days, range = 121 to 339 days) in children/adolescents and 254 days (SD = 36 days, range = 93 to 340 days) in adults. COVID-19 and control cohort were well balanced regarding covariates. The specific outcomes with the highest IRR and an incidence rate (IR) of at least 1/100 person-years in the COVID-19 cohort in children and adolescents were malaise/fatigue/exhaustion (IRR: 2.28, 95% CI: 1.71 to 3.06, p < 0.01, IR COVID-19: 12.58, IR Control: 5.51), cough (IRR: 1.74, 95% CI: 1.48 to 2.04, p < 0.01, IR COVID-19: 36.56, IR Control: 21.06), and throat/chest pain (IRR: 1.72, 95% CI: 1.39 to 2.12, p < 0.01, IR COVID-19: 20.01, IR Control: 11.66). In adults, these included disturbances of smell and taste (IRR: 6.69, 95% CI: 5.88 to 7.60, p < 0.01, IR COVID-19: 12.42, IR Control: 1.86), fever (IRR: 3.33, 95% CI: 3.01 to 3.68, p < 0.01, IR COVID-19: 11.53, IR Control: 3.46), and dyspnea (IRR: 2.88, 95% CI: 2.74 to 3.02, p < 0.01, IR COVID-19: 43.91, IR Control: 15.27). For all health outcomes combined, IRs per 1,000 person-years in the COVID-19 cohort were significantly higher than those in the control cohort in both children/adolescents (IRR: 1.30, 95% CI: 1.25 to 1.35, p < 0.01, IR COVID-19: 436.91, IR Control: 335.98) and adults (IRR: 1.33, 95% CI: 1.31 to 1.34, p < 0.01, IR COVID-19: 615.82, IR Control: 464.15). The relative magnitude of increased documented morbidity was similar for the physical, mental, and physical/mental overlap domain. In the COVID-19 cohort, IRs were significantly higher in all 13 diagnosis/symptom complexes in adults and in 10 diagnosis/symptom complexes in children/adolescents. IRR estimates were similar for age groups 0 to 11 and 12 to 17. IRs in children/adolescents were consistently lower than those in adults. Limitations of our study include potentially unmeasured confounding and detection bias.ConclusionsIn this retrospective matched cohort study, we observed significant new onset morbidity in children, adolescents, and adults across 13 prespecified diagnosis/symptom complexes, following COVID-19 infection. These findings expand the existing available evidence on post-COVID-19 conditions in younger age groups and confirm previous findings in adults.Trial registrationClinicalTrials.gov https://clinicaltrials.gov/ct2/show/NCT05074953.