Attempts to use artificial intelligence (AI) in psychiatric disorders show moderate success, highlighting the potential of incorporating information from clinical assessments to improve the models. This study focuses on using large language models (LLMs) to detect suicide risk from medical text in psychiatric care.

To extract information about suicidality status from the admission notes in electronic health records (EHRs) using privacy-sensitive, locally hosted LLMs, specifically evaluating the efficacy of Llama-2 models.

We compared the performance of several variants of the open source LLM Llama-2 in extracting suicidality status from 100 psychiatric reports against a ground truth defined by human experts, assessing accuracy, sensitivity, specificity and F1 score across different prompting strategies.

A German fine-tuned Llama-2 model showed the highest accuracy (87.5%), sensitivity (83.0%) and specificity (91.8%) in identifying suicidality, with significant improvements in sensitivity and specificity across various prompt designs.

The study demonstrates the capability of LLMs, particularly Llama-2, in accurately extracting information on suicidality from psychiatric records while preserving data privacy. This suggests their application in surveillance systems for psychiatric emergencies and improving the clinical management of suicidality by improving systematic quality control and research.

Systemic changes in multiple diseases may influence the onset of dementia. However, the specific temporality between exposure diseases and dementia remains uncertain.

By characterising the full spectrum of temporal disease trajectories before dementia, this study aims to yield a global picture of precursor diseases to dementia and to provide detailed instructions for risk management and primary prevention of dementia.

Using the multicentre, community-based prospective UK Biobank, we constructed disease trajectories before dementia utilising the phenome-wide association analysis, paired directional test and association quantification. Stratified disease trajectories were constructed by dementia subtypes, gender, age of diagnosis and Apolipoprotein E (ApoE) status, respectively.

Our study population comprised 434 266 participants without baseline dementia and 4638 individuals with all-cause dementia. In total, 1253 diseases were extracted as potential components of the disease trajectory before dementia. We identified three clusters of disease trajectories preceding all-cause dementia, initiated by circulatory, metabolic and respiratory diseases occurring approximately 5–15 years before dementia. Cerebral infarction or chronic renal failure following chronic ischaemic heart disease was the specific trajectory before vascular dementia. Apolipoprotein E (ApoE) ε4 non-carriers exhibited more complex trajectories compared with carriers. Lipid metabolism disorders remained in the trajectories regardless of dementia subtypes, gender, age of diagnosis and ApoE status.

This study provides a comprehensive view of the longitudinal disease trajectories before dementia and highlights the potential targets of midlife cardiometabolic dysfunction for dementia screening and prevention.

Despite strong evidence of efficacy of electroconvulsive therapy (ECT) in the treatment of depression, no sensitive and specific predictors of ECT response have been identified. Previous meta-analyses have suggested some pre-treatment associations with response at a population level.

Using 10 years (2009–2018) of routinely collected Scottish data of people with moderate to severe depression (n = 2074) receiving ECT we tested two hypotheses: (a) that there were significant group-level associations between post-ECT clinical outcomes and pre-ECT clinical variables and (b) that it was possible to develop a method for predicting illness remission for individual patients using machine learning.

Data were analysed on a group level using descriptive statistics and association analyses as well as using individual patient prediction with machine learning methodologies, including cross-validation.

ECT is highly effective for moderate to severe depression, with a response rate of 73% and remission rate of 51%. ECT response is associated with older age, psychotic symptoms, necessity for urgent intervention, severe distress, psychomotor retardation, previous good response, lack of medication resistance, and consent status. Remission has the same associations except for necessity for urgent intervention and, in addition, history of recurrent depression and low suicide risk. It is possible to predict remission with ECT with an accuracy of 61%.

Pre-ECT clinical variables are associated with both response and remission and can help predict individual response to ECT. This predictive tool could inform shared decision-making, prevent the unnecessary use of ECT when it is unlikely to be beneficial and ensure prompt use of ECT when it is likely to be effective.

Understanding inequalities in outcomes between demographic groups is a necessary step in addressing them in clinical care. Inequalities in treatment uptake between demographic groups may explain disparities in outcomes in people with first-episode psychosis (FEP).

To investigate disparities between broad demographic groups in symptomatic improvement in patients with FEP and their relationship to treatment uptake.

We used data from 6813 patients from the 2021–2022 National Clinical Audit of Psychosis data-set. Data were grouped by category type to obtain mean outcomes before adjustment to see whether disparities in outcomes remained after differences in treatment uptake had been accounted for. After matching, the average effect of each demographic variable in terms of outcome change was calculated. Moderator effects on specific treatments were investigated using interaction terms in a regression model.

Observational results showed that patients aged 18–24 years were less likely to improve in outcome, unless adjusted for intervention uptake. Patients classified as Black and Black British were less likely to improve in outcome (moderation effect 0.04, 95% CI 0–0.07) after adjusting for treatment take-up and demographic factors. Regression analysis showed the general positive effect of supported employment interventions in improving outcomes (coefficient −0.13, 95% CI −0.07 to −0.18, P < 0.001), and moderator analysis suggested targeting particular groups for interventions.

Inequalities in treatment uptake and psychotic symptom outcome of FEP by social and demographic factors require monitoring over time. Our analysis provides a framework for monitoring health inequalities across national clinical audits in the UK.

Understanding how childhood psychosocial adjustment (CPA) influences later life health outcomes is crucial for developing interventions to mitigate the long-term risk of cardiometabolic diseases (CMDs).

To investigate the association between CPA and incident CMDs in mid-life, and the mediating roles of educational attainment, smoking habits and depression during young adulthood.

A prospective cohort study utilised data from the 1958 National Child Development Study (NCDS; 1958–2013) and the 1970 British Cohort Study (BCS70; 1970–2018), encompassing 22 012 participants assessed for CPA in childhood, who were subsequently evaluated for educational attainment, smoking habits and depression in young adulthood, followed by assessments for CMDs in mid-life. CPA was assessed using the Bristol Social Adjustment Guides in the NCDS and the Rutter Child Behaviour Scale in the BCS70, with higher scores indicating poorer psychosocial adjustment. The primary outcomes were the mid-life incidences of hypertension, diabetes and obesity.

Compared with children in the lowest tertile for CPA scores, those in the middle tertile had an adjusted odds ratio for hypertension of 0.98 (95% CI 0.90–1.06), whereas those in the highest tertile had an odds ratio of 1.17 (95% CI 1.08–1.26). For diabetes, the corresponding odds ratios (95% CI) were 1.15 (0.98–1.35) and 1.39 (1.19–1.62). For obesity, the corresponding odds ratios (95% CI) were 1.08 (1.00–1.16) and 1.18 (1.09–1.27). These associations were partially mediated by educational attainment (2.4–13.9%) and depression during young adulthood (2.5–14.9%).

Poorer CPA is correlated with the development of hypertension, diabetes and obesity in mid-life. Interventions aimed at improving CPA may help in reducing the burden of these diseases in later life.

Clozapine is the most effective medication for treatment-resistant psychoses, but the balance of benefits and risks is understudied in real-world settings.

To examine the relative re-hospitalisation rates for mental health relapse and adverse events associated with clozapine and other antipsychotics in adult and child/youth cohorts.

Data were obtained from the Canadian Institute of Health Information for adults (n = 45 616) and children/youth (n = 1476) initially hospitalised for mental health conditions in British Columbia, Manitoba and Saskatchewan from 2008 to 2018. Patient demographics and hospitalisations were linked with antipsychotic prescriptions dispensed following the initial visit. Recurrent events survival analysis for relapse and adverse events were created and compared between clozapine and other antipsychotics.

In adults, clozapine was associated with a 14% lower relapse rate versus other drugs (adjusted hazard ratio: 0.86, 95% CI: 0.83–0.90) over the 10-year follow-up. In the first 21 months, the relapse rate was higher for clozapine but then reversed. Over 1000 person-months, clozapine-treated adults could be expected to have 38 relapse hospitalisations compared with 45 for other drugs. In children/youth, clozapine had a 38% lower relapse rate compared with other antipsychotic medications (adjusted hazard ratio: 0.62, 95% CI: 0.49–0.78) over the follow-up period. This equates to 29 hospitalisations for clozapine and 48 for other drugs over 1000 person-months. In adults, clozapine had a higher risk for adverse events (hazard ratio: 1.34, 95% CI: 1.18–1.54) over the entire follow-up compared with other antipsychotics. This equates to 1.77 and 1.30 hospitalisations over 1000 person-months for clozapine and other drugs, respectively.

Clozapine was associated with lower relapse overall, but this was accompanied by higher adverse events for adults. For children/youth, clozapine was associated with lower relapse all throughout and had no difference in adverse events compared with other antipsychotics.