Incidence and course of adolescent deliberate self-harm in Victoria, Australia, and Washington State

PURPOSE: There have been few longitudinal studies of deliberate self-harm (DSH) in adolescents. This cross-national longitudinal study outlines risk and protective factors for DSH incidence and persistence. METHODS: Seventh and ninth grade students (average ages 13 and 15 years) were recruited as state-representative cohorts, surveyed, and then followed up 12 months later (N = 3,876), using the same methods in Washington State and Victoria, Australia. The retention rate was 99% in both states at follow-up. A range of risk and protective factors for DSH were examined using multivariate analyses. RESULTS: The prevalence of DSH in the past year was 1.53% in Grade 7 and .91% in Grade 9 for males and 4.12% and 1.34% for Grade 7 and Grade 9 females, respectively, with similar rates across states. In multivariate analyses, incident DSH was lower in Washington State (odds ratio [OR] = .67; 95% confidence interval [CI] = .45-1.00) relative to Victoria 12 months later. Risk factors for incident DSH included being female (OR = 1.93; CI = 1.35-2.76), high depressive symptoms (OR = 3.52; CI = 2.37-5.21), antisocial behavior (OR = 2.42; CI = 1.46-4.00), and lifetime (OR = 1.85; CI = 1.11-3.08) and past month (OR = 2.70; CI = 1.57-4.64) alcohol use relative to never using alcohol. CONCLUSIONS: Much self-harm in adolescents resolves over the course of 12 months. Young people who self-harm have high rates of other health risk behaviors associated with family and peer risks that may all be targets for preventive intervention.

Learning vector representation of medical objects via EMR-driven nonnegative restricted Boltzmann machines (eNRBM)

Electronic medical record (EMR) offers promises for novel analytics. However, manual feature engineering from EMR is labor intensive because EMR is complex - it contains temporal, mixed-type and multimodal data packed in irregular episodes. We present a computational framework to harness EMR with minimal human supervision via restricted Boltzmann machine (RBM). The framework derives a new representation of medical objects by embedding them in a low-dimensional vector space. This new representation facilitates algebraic and statistical manipulations such as projection onto 2D plane (thereby offering intuitive visualization), object grouping (hence enabling automated phenotyping), and risk stratification. To enhance model interpretability, we introduced two constraints into model parameters: (a) nonnegative coefficients, and (b) structural smoothness. These result in a novel model called eNRBM (EMR-driven nonnegative RBM). We demonstrate the capability of the eNRBM on a cohort of 7578 mental health patients under suicide risk assessment. The derived representation not only shows clinically meaningful feature grouping but also facilitates short-term risk stratification. The F-scores, 0.21 for moderate-risk and 0.36 for high-risk, are significantly higher than those obtained by clinicians and competitive with the results obtained by support vector machines.

Differentially private multi-task learning

Privacy restrictions of sensitive data repositories imply that the data analysis is performed in isolation at each data source. A prime example is the isolated nature of building prognosis models from hospital data and the associated challenge of dealing with small number of samples in risk classes (e.g. suicide) while doing so. Pooling knowledge from other hospitals, through multi-task learning, can alleviate this problem. However, if knowledge is to be shared unrestricted, privacy is breached. Addressing this, we propose a novel multi-task learning method that preserves privacy of data under the strong guarantees of differential privacy. Further, we develop a novel attribute-wise noise addition scheme that significantly lifts the utility of the proposed method. We demonstrate the effectiveness of our method with a synthetic and two real datasets.

Differentiating sub-groups of online depression-related communities using textual cues

Depression is a highly prevalent mental illness and is a comorbidity of other mental and behavioural disorders. The Internet allows individuals who are depressed or caring for those who are depressed, to connect with others via online communities; however, the characteristics of these online conversations and the language styles of those interested in depression have not yet been fully explored. This work aims to explore the textual cues of online communities interested in depression. A random sample of 5,000 blog posts was crawled. Five groupings were identified: depression, bipolar, self-harm, grief, and suicide. Independent variables included psycholinguistic processes and content topics extracted from the posts. Machine learning techniques were used to discriminate messages posted in the depression sub-group from the others.Good predictive validity in depression classification using topics and psycholinguistic clues as features was found. Clear discrimination between writing styles and content, with good predictive power is an important step in understanding social media and its use in mental health.