Predicting Lifetime Suicide Attempts in a Community Sample of Adolescents Using Machine Learning Algorithms.

Suicide is a major global health concern and a prominent cause of death in adolescents. Previous research on suicide prediction has mainly focused on clinical or adult samples. To prevent suicides at an early stage, however, it is important to screen for risk factors in a community sample of adolescents. We compared the accuracy of logistic regressions, elastic net regressions, and gradient boosting machines in predicting suicide attempts by 17-year-olds in the Millennium Cohort Study (N = 7,347), combining a large set of self- and other-reported variables from different categories. Both machine learning algorithms outperformed logistic regressions and achieved similar balanced accuracies (.76 when using data 3 years before the self-reported lifetime suicide attempts and .85 when using data from the same measurement wave). We identified essential variables that should be considered when screening for suicidal behavior. Finally, we discuss the usefulness of complex machine learning models in suicide prediction.

Predicting treatment response using machine learning: A registered report.

OBJECTIVE: Previous research on psychotherapy treatment response has mainly focused on outpatients or clinical trial data which may have low ecological validity regarding naturalistic inpatient samples. To reduce treatment failures by proactively screening for patients at risk of low treatment response, gain more knowledge about risk factors and to evaluate treatments, accurate insights about predictors of treatment response in naturalistic inpatient samples are needed. METHODS: We compared the performance of different machine learning algorithms in predicting treatment response, operationalized as a substantial reduction in symptom severity as expressed in the Patient Health Questionnaire Anxiety and Depression Scale. To achieve this goal, we used different sets of variables-(a) demographics, (b) physical indicators, (c) psychological indicators and (d) treatment-related variables-in a naturalistic inpatient sample (N = 723) to specify their joint and unique contribution to treatment success. RESULTS: There was a strong link between symptom severity at baseline and post-treatment (R2 = .32). When using all available variables, both machine learning algorithms outperformed the linear regressions and led to an increment in predictive performance of R2 = .12. Treatment-related variables were the most predictive, followed psychological indicators. Physical indicators and demographics were negligible. CONCLUSIONS: Treatment response in naturalistic inpatient settings can be predicted to a considerable degree by using baseline indicators. Regularization via machine learning algorithms leads to higher predictive performances as opposed to including nonlinear and interaction effects. Heterogenous aspects of mental health have incremental predictive value and should be considered as prognostic markers when modelling treatment processes.

Mapping established psychopathology scales onto the Hierarchical Taxonomy of Psychopathology (HiTOP).

The Hierarchical Taxonomy of Psychopathology (HiTOP) organizes phenotypes of mental disorder based on empirical covariation, offering a comprehensive organizational framework from narrow symptoms to broader patterns of psychopathology. We argue that established self-report measures of psychopathology from the pre-HiTOP era should be systematically integrated into HiTOP to foster cumulative research and further the understanding of psychopathology structure. Hence, in this study, we mapped 92 established psychopathology (sub)scales onto the current HiTOP working model using data from an extensive battery of self-report assessments that was completed by community participants and outpatients (N = 909). Content validity ratings of the item pool were used to select indicators for a bifactor-(S-1) model of the p factor and five HiTOP spectra (i.e., internalizing, thought disorder, detachment, disinhibited externalizing, and antagonistic externalizing). The content-based HiTOP scales were validated against personality disorder diagnoses as assessed by standardized interviews. We then located established scales within the taxonomy by estimating the extent to which scales reflected higher-level HiTOP dimensions. The analyses shed light on the location of established psychopathology scales in HiTOP, identifying pure markers and blends of HiTOP spectra, as well as pure markers of the p factor (i.e., scales assessing mentalizing impairment and suspiciousness/epistemic mistrust).

The Road to Hell Is Paved With Good Intentions: How Common Practices in Scale Construction Hurt Validity.

Sound scale construction is pivotal to the measurement of psychological constructs. Common item sampling procedures emphasize aspects of reliability to the disadvantage of aspects of validity, which are less tangible. We use a health knowledge test as an example to demonstrate how item sampling strategies that focus on either factor saturation or construct coverage influence scale composition and demonstrate how to find a trade-off between these two opposing needs. More specifically, we compile three 75-item health knowledge scales using Ant Colony Optimization, a metaheuristic algorithm that is inspired by the foraging behavior of ants, to optimize factor saturation, construct coverage, or a compromise of both. We demonstrate that our approach is well suited to balance out construct coverage and factor saturation when constructing a health knowledge test. Finally, we discuss conceptual problems with the modeling of declarative knowledge and provide recommendations for the assessment of health knowledge.

The HEX-ACO-18: Developing an Age-Invariant HEXACO Short Scale Using Ant Colony Optimization.

In this study, we developed an age-invariant 18-item short form of the HEXACO Personality Inventory for use in developmental personality research. We combined the item selection procedure ant colony optimization (ACO) and the model estimation approach local structural equation modeling (LSEM). ACO is a metaheuristic algorithm that evaluates items based on the quality of the resulting short scale, thus directly optimizing criteria that can only be estimated with combinations of items, such as model fit and measurement invariance. LSEM allows for model estimation and measurement invariance testing across a continuous age variable by weighting participants, rather than splitting the sample into artificial age groups. Using a HEXACO-100 dataset of N = 6,419 participants ranging from 16 to 90 years of age, we selected a short form optimized for model fit, measurement invariance, facet coverage, and balance of item keying. To achieve scalar measurement invariance and brevity, but maintain construct coverage, we selected 18 items to represent three out of four facets from each HEXACO trait domain. The resulting HEX-ACO-18 short scale showed adequate model fit and scalar measurement invariance across age. Furthermore, the usefulness and versatility of the item and person sampling procedures ACO and LSEM is demonstrated.