Search

Abstract. Modern prediction methods from machine learning (ML) and artificial intelligence (AI) are becoming increasingly popular, also in the field of psychological assessment. These methods provide unprecedented flexibility for modeling large numbers of predictor variables and non-linear associations between predictors and responses. In this paper, we aim to look at what these methods may contribute to the assessment of criterion validity and their possible drawbacks. We apply a range of modern statistical prediction methods to a dataset for predicting the university major completed, based on the subscales and items of a scale for vocational preferences. The results indicate that logistic regression combined with regularization performs strikingly well already in terms of predictive accuracy. More sophisticated techniques for incorporating non-linearities can further contribute to predictive accuracy and validity, but often marginally.

OBJECTIVE: Many individuals with an eating disorder do not receive appropriate care. Low-threshold interventions could help bridge this treatment gap. The study aim was to evaluate the effectiveness of Featback, a fully automated online self-help intervention, online expert-patient support and their combination. METHOD: A randomized controlled trial with a 12-month follow-up period was conducted. Participants aged 16 or older with at least mild eating disorder symptoms were randomized to four conditions: (1) Featback, a fully automated online self-help intervention, (2) chat or email support from a recovered expert patient, (3) Featback with expert-patient support and (4) a waiting list control condition. The intervention period was 8 weeks and there was a total of six online assessments. The main outcome constituted reduction of eating disorder symptoms over time. RESULTS: Three hundred fifty five participants, of whom 43% had never received eating disorder treatment, were randomized. The three active interventions were superior to a waitlist in reducing eating disorder symptoms (d = -0.38), with no significant difference in effectiveness between the three interventions. Participants in conditions with expert-patient support were more satisfied with the intervention. DISCUSSION: Internet-based self-help, expert-patient support and their combination were effective in reducing eating disorder symptoms compared to a waiting list control condition. Guidance improved satisfaction with the internet intervention but not its effectiveness. Low-threshold interventions such as Featback and expert-patient support can reduce eating disorder symptoms and reach the large group of underserved individuals, complementing existing forms of eating disorder treatment. PUBLIC SIGNIFICANCE STATEMENT: Individuals with eating-related problems who received (1) a fully automated internet-based intervention, (2) chat and e-mail support by a recovered individual or (3) their combination, experienced stronger reductions in eating disorder symptoms than those who received (4) usual care. Such brief and easy-access interventions play an important role in reaching individuals who are currently not reached by other forms of treatment. ispartof: INTERNATIONAL JOURNAL OF EATING DISORDERS vol:55 issue:10 pages:1361-1373 ispartof: location:United States status: published

Background Adding short-term psychodynamic psychotherapy (STPP) to antidepressants increases treatment efficacy, but it is unclear which patients benefit specifically. This study examined efficacy moderators of combined treatment (STPP + antidepressants) v. antidepressants for adults with depression. Methods For this systematic review and meta-analysis (PROSPERO registration number: CRD42017056029), we searched PubMed, PsycINFO, Embase.com, and the Cochrane Library from inception to 1 January 2022. We included randomized clinical trials comparing combined treatment (antidepressants + individual outpatient STPP) v. antidepressants in the acute-phase treatment of depression in adults. Individual participant data were requested and analyzed combinedly using mixed-effects models (adding Cochrane risk of bias items as covariates) and an exploratory machine learning technique. The primary outcome was post-treatment depression symptom level. Results Data were obtained for all seven trials identified (100%, n = 482, combined: n = 238, antidepressants: n = 244). Adding STPP to antidepressants was more efficacious for patients with high rather than low baseline depression levels [B = −0.49, 95% confidence interval (CI) −0.61 to −0.37, p < 0.0001] and for patients with a depressive episode duration of >2 years rather than B = −0.68, 95% CI −1.31 to −0.05, p = 0.03) and than 1–2 years (B = −0.86, 95% CI −1.66 to −0.06, p = 0.04). Heterogeneity was low. Effects were replicated in analyses controlling for risk of bias. Conclusions To our knowledge, this is the first study that examines moderators across trials assessing the addition of STPP to antidepressants. These findings need validation but suggest that depression severity and episode duration are factors to consider when adding STPP to antidepressants and might contribute to personalizing treatment selection for depression.