Search

Various methods have been developed to combine inference across multiple sets of results for unsupervised clustering, within the ensemble clustering literature. The approach of reporting results from one ‘best’ model out of several candidate clustering models generally ignores the uncertainty that arises from model selection, and results in inferences that are sensitive to the particular model and parameters chosen. Bayesian model averaging (BMA) is a popular approach for combining results across multiple models that offers some attractive benefits in this setting, including probabilistic interpretation of the combined cluster structure and quantification of model-based uncertainty. In this work we introduce clusterBMA, a method that enables weighted model averaging across results from multiple unsupervised clustering algorithms. We use clustering internal validation criteria to develop an approximation of the posterior model probability, used for weighting the results from each model. From a combined posterior similarity matrix representing a weighted average of the clustering solutions across models, we apply symmetric simplex matrix factorisation to calculate final probabilistic cluster allocations. In addition to outperforming other ensemble clustering methods on simulated data, clusterBMA offers unique features including probabilistic allocation to averaged clusters, combining allocation probabilities from ‘hard’ and ‘soft’ clustering algorithms, and measuring model-based uncertainty in averaged cluster allocation. This method is implemented in an accompanying R package of the same name. We use simulated datasets to explore the ability of the proposed technique to identify robust integrated clusters with varying levels of separation between subgroups, and with varying numbers of clusters between models. Benchmarking accuracy against four other ensemble methods previously demonstrated to be highly effective in the literature, clusterBMA matches or exceeds the performance of competing approaches under various conditions of dimensionality and cluster separation. clusterBMA substantially outperformed other ensemble methods for high dimensional simulated data with low cluster separation, with 1.16 to 7.12 times better performance as measured by the Adjusted Rand Index. We also explore the performance of this approach through a case study that aims to identify probabilistic clusters of individuals based on electroencephalography (EEG) data. In applied settings for clustering individuals based on health data, the features of probabilistic allocation and measurement of model-based uncertainty in averaged clusters are useful for clinical relevance and statistical communication.

BackgroundA priority for health services is to reduce self-harm in young people. Predicting self-harm is challenging due to their rarity and complexity, however this does not preclude the utility of prediction models to improve decision-making regarding a service response in terms of more detailed assessments and/or intervention. The aim of this study was to predict self-harm within six-months after initial presentation.MethodThe study included 1962 young people (12-30 years) presenting to youth mental health services in Australia. Six machine learning algorithms were trained and tested with ten repeats of ten-fold cross-validation. The net benefit of these models were evaluated using decision curve analysis.ResultsOut of 1962 young people, 320 (16%) engaged in self-harm in the six months after first assessment and 1642 (84%) did not. The top 25% of young people as ranked by mean predicted probability accounted for 51.6% - 56.2% of all who engaged in self-harm. By the top 50%, this increased to 82.1%-84.4%. Models demonstrated fair overall prediction (AUROCs; 0.744-0.755) and calibration which indicates that predicted probabilities were close to the true probabilities (brier scores; 0.185-0.196). The net benefit of these models were positive and superior to the 'treat everyone' strategy. The strongest predictors were (in ranked order); a history of self-harm, age, social and occupational functioning, sex, bipolar disorder, psychosis-like experiences, treatment with antipsychotics, and a history of suicide ideation.ConclusionPrediction models for self-harm may have utility to identify a large sub population who would benefit from further assessment and targeted (low intensity) interventions. Such models could enhance health service approaches to identify and reduce self-harm, a considerable source of distress, morbidity, ongoing health care utilisation and mortality.

Neuropsychiatric disorders (including substance misuse) are associated with the greatest burden of functional disability in young people, and contributory factors remain poorly understood. Early-onset substance use is one candidate risk factor which may inform functional prognosis and facilitate direction of interventions aiming to curtail impairment. Accordingly, we modelled associations between early-onset use of alcohol, tobacco, cannabis and amphetamine-type stimulants (ATSs) and longitudinal socio-occupational functioning (indexed by the Social and Occupational Functioning Assessment Scale) in an observational cohort presenting to early intervention mental health services. A clinical proforma collated demographic, clinical, and socio-occupational information for up to 60-months from presentation to services in young people aged 17-30. Of the wider cohort (n = 2398), 446 participants were selected with complete alcohol and substance use data. Latent class analysis was used to derive an 'early-onset' (n = 243) and 'later-onset' class (n = 203) based on age of first use of alcohol, tobacco, cannabis and ATSs. Maximum-likelihood multilevel analyses modelled functioning over time in care and tested associations with substance use latent class, age, gender and diagnosis. Membership in the 'early-onset' class (B = -1.64, p = 0.05), male gender (B = -3.27, p

Although early-stage affective disorders are associated with both cognitive dysfunction and sleep-wake disruptions, relationships between these factors have not been specifically examined in young adults. Sleep and circadian rhythm disturbances in those with affective disorders are considerably heterogeneous, and may not relate to cognitive dysfunction in a simple linear fashion. This study aimed to characterise profiles of sleep and circadian disturbance in young people with affective disorders and examine associations between these profiles and cognitive performance. Actigraphy monitoring was completed in 152 young people (16-30 years; 66% female) with primary diagnoses of affective disorders, and 69 healthy controls (18-30 years; 57% female). Patients also underwent detailed neuropsychological assessment. Actigraphy data were processed to estimate both sleep and circadian parameters. Overall neuropsychological performance in patients was poor on tasks relating to mental flexibility and visual memory. Two hierarchical cluster analyses identified three distinct patient groups based on sleep variables and three based on circadian variables. Sleep clusters included a 'long sleep' cluster, a 'disrupted sleep' cluster, and a 'delayed and disrupted sleep' cluster. Circadian clusters included a 'strong circadian' cluster, a 'weak circadian' cluster, and a 'delayed circadian' cluster. Medication use differed between clusters. The 'long sleep' cluster displayed significantly worse visual memory performance compared to the 'disrupted sleep' cluster. No other cognitive functions differed between clusters. These results highlight the heterogeneity of sleep and circadian profiles in young people with affective disorders, and provide preliminary evidence in support of a relationship between sleep and visual memory, which may be mediated by use of antipsychotic medication. These findings have implications for the personalisation of treatments and improvement of functioning in young adults early in the course of affective illness.

Evidence shows that there are reductions in gray matter volume (GMV) and changes in long association white matter fibres within the left insula-temporoparietal junction (TPJ) during the early stages of psychotic disorders but less is known about short association fibres (sAFs). In this study we sought to characterise the changes in sAFs and associated volumetric changes of the left insula-TPJ during the early stages of psychosis. Magnetic resonance imaging was obtained from a sample of young people with psychosis (n = 42) and healthy controls (n = 45), and cortical parcellations of the left insula-TPJ were used as seeding masks to reconstruct 13 sAFs. Compared to healthy counterparts, the psychosis group showed significant reductions in fractional anisotropy (FA) in the sAFs connecting the superior (STG) and middle temporal gyri (MTG) and as well as reduced GMV within the inferior temporal gyrus and increased white matter volume (WMV) within Heschl's gyrus (HG). Furthermore, adolescent-onset psychosis subjects (onset 18 year or earlier) showed FA reductions in the STG-HG sAF when compared to adult-onset subjects, but this was not associated with changes in GMV nor WMV of the STG or HG. These findings suggest that during the early stages of psychosis, changes in sAFs and associated cortical GMV and WMV appear to occur independently, however age of onset of a psychotic syndrome/disorder influences the pattern of neuroanatomical abnormalities.

Considering the marked changes in sleep and circadian rhythms across the lifespan, age may contribute to the heterogeneity in sleep-wake profiles linked to mood disorders. This study aimed to investigate the contributions of age and depression severity to sleep-wake disturbances. The Hamilton Depression Rating Scale (HDRS) was administered to assess current symptoms severity in 238 persons with a history of a mood disorder between 12 and 90 years of age (y.o.). Actigraphy was recorded over five to 22 days. Regression analyses and analyses of variance [age (12-19 y.o., 20-39 y.o., 40-59 y.o., and ≥ 60 y.o.) by depression severity (HDRS< and ≥ 8)] were conducted. The 12-19 y.o. and 20-39 y.o. groups had a delayed sleep schedule and acrophase compared to all other groups. The ≥ 60 y.o. group had a lower rhythmicity and amplitude (p ≤ .006) than the 12-19 y.o. group (p ≤ .046). Participants with a HDRS ≥ 8 spent longer time in bed, had later sleep offset times and had lower circadian rhythmicity than those with a HDRS

To date, most studies of white matter changes in Bipolar Disorder (BD) have been conducted in older subjects and with well-established disorders. Studies of young people who are closer to their illness onset may help to identify core neurobiological characteristics and separate these from consequences of repeated illness episodes or prolonged treatment. Diffusion tensor imaging (DTI) was used to examine white matter microstructural changes in 58 young patients with BD (mean age 23 years; range 16-30 years) and 40 controls. Whole brain voxelwise measures of fractional anisotropy (FA), parallel diffusivity (λ//) and radial diffusivity (λ⊥) were calculated for all subjects. White matter microstructure differences (decreased FA corrected p

BACKGROUND: Clinical symptoms and neuropsychological deficits are longitudinally associated with functional outcome in chronic psychiatric cohorts. The current study extended these findings to young and early-course psychiatric outpatients, with the aim of identifying cognitive markers that predict later socio-occupational functioning. METHODS: At baseline, 183 young psychiatric outpatients were assessed. Ninety-three returned for follow-up (M = 21.6 years old; SD = 4.5) with an average re-assessment interval of 21.6 months (SD = 7.0), and primary diagnoses of major depressive disorder (n = 34), bipolar disorder (n = 29), or psychosis (n = 30). The primary outcome measure was cross-validated with various other functional measures and structural equation modelling was used to map out the interrelationships between predictors and later functional outcome. RESULTS: Good socio-occupational functioning at follow-up was associated with better quality of life, less disability, current employment and being in a romantic relationship. The final structural equation model explained 47.5% of the variability in functional outcome at follow-up, with baseline neuropsychological functioning (a composite of memory, working memory and attentional switching) the best independent predictor of later functional outcome. Notably, depressive and negative symptoms were only associated with functioning cross-sectionally. Diagnosis at follow-up was not associated with functional outcome. CONCLUSIONS: Neuropsychological functioning was the single best predictor of later socio-occupational outcome among young psychiatric outpatients. Therefore, framing psychiatric disorders along a neuropsychological continuum is likely to be more useful in predicting functional trajectory than traditional symptom-based classification systems. The current findings also have implications for early intervention utilising cognitive remediation approaches.

Antisocial behaviours and psychopathic traits place an individual at risk for criminality, mental illness, substance dependence, and psychosocial dysfunction. Social cognition deficits appear to be associated with psychopathic traits and are believed to contribute to interpersonal dysfunction. Most research investigating the relationship of these traits with social cognition has been conducted either in children or adult forensic settings. We investigated whether psychopathic traits were associated with social cognition in 91 young people presenting for mental healthcare (aged between 15 and 25 years). Participants completed symptom severity measures, neuropsychological tests, the Reading the Mind in the Eyes Test of social cognition (RMET), and the Antisocial Process Screening Device (APSD) to assess psychopathic personality traits. Correlation analyses showed poorer social cognition was associated with greater psychopathic traits (r = -.36, p = .01). Interestingly, social cognition performance predicted unique variance in concurrent psychopathic personality traits above gender, IQ sustained attention, and working memory performance. These findings suggest that social cognitive impairments are associated with psychopathic tendencies in young people presenting for community mental healthcare. Research is needed to establish the directionality of this relationship and to determine whether social cognition training is an effective treatment amongst young people with psychopathic tendencies.