Prediction and interpretation of distributed neural activity with sparse models.

We explore to what extent the combination of predictive and interpretable modeling can provide new insights for functional brain imaging. For this, we apply a recently introduced regularized regression technique, the Elastic Net, to the analysis of the PBAIC 2007 competition data. Elastic Net regression controls via one parameter the number of voxels in the resulting model, and via another the degree to which correlated voxels are included. We find that this method produces highly predictive models of fMRI data that provide evidence for the distributed nature of neural function. We also use the flexibility of Elastic Net to demonstrate that model robustness can be improved without compromising predictability, in turn revealing the importance of localized clusters of activity. Our findings highlight the functional significance of patterns of distributed clusters of localized activity, and underscore the importance of models that are both predictive and interpretable.

Predictors of early recovery from major depression among persons admitted to community-based clinics: an observational study.

BACKGROUND: Twenty years have elapsed since the National Institute of Mental Health Collaborative Depression Study reported on the early course and treatment of major depression within the mental health sector. Using similar methods, an observational study was conducted to assess relationships between initial depression severity, personality dysfunction and other baseline characteristics, subsequent treatment, and 3-month outcomes among persons admitted to public and voluntary sector outpatient clinics, including 1 academic program. METHODS: A 2-stage sampling technique was used to recruit subjects (N = 165) diagnosed by the Structured Clinical Interview for DSM-IV, Patient Version, as having a major depression episode. Sociodemographic and clinical characteristics were assessed at admission. Data on treatment and outcome were obtained at 3 months using structured instruments from the Longitudinal Interview Follow-up Evaluation. Logistic regression was used to assess hypothesized predictors of early recovery. Analyses were carried out in the total sample and after dichotomizing subjects by baseline depression severity. RESULTS: Fifty (30.3%) of the 165 subjects met recovery criteria. Less than half of the subjects (45%) met criteria for adequate pharmacotherapy. Less severe depression, having received adequate antidepressant treatment, female sex, and being married independently predicted early recovery. In the more depressed subgroup, early recovery was associated with female sex. Among less severely depressed subjects, high personality dysfunction scores and being married were significant predictors. CONCLUSIONS: Initial depression severity and receiving adequate pharmacotherapy predict early recovery in individuals with major depression seeking outpatient treatment. A minority of persons receive intensive antidepressant treatment. Less severe personality dysfunction and being married predicts early recovery among persons with less severe depression.