Project Summary. Youths receiving mental healthcare often have multiple disorders, distinctive treatment- relevant personal and family characteristics, and problems that may shift during treatment. For these youths, modular psychotherapies, in which providers select from a menu of therapeutic elements to build personalized treatments, may be especially appropriate. One well-studied modular youth psychotherapy is Modular Approach to Therapy for Children with Anxiety, Depression, Trauma, or Conduct Problems (MATCH). In some trials with intensive, one-on-one expert support for element selection, MATCH significantly outperformed usual care and evidence-based standardized psychotherapies; but in trials with less intensive (more practically feasible) support, MATCH did not outperform usual care. This discrepancy highlights a key challenge of MATCH and other modular psychotherapies: selection of treatment elements. Because MATCH allows any number of sessions implemented in any order, a virtually unlimited array of treatment element sequences could be selected. This flexibility supports personalization but may also impact the effectiveness of MATCH and other modular psychotherapies, because it can be unclear which modules should be used when. As in other modular therapies, clinicians using MATCH are asked to use clinical judgment and are provided with decision- making guidance based on past literature and client weekly assessments. But research has shown that statistical decision-making models using archival data very often outperform clinician judgment. Currently, no modular youth psychotherapies employ data-driven statistical models to inform decision-making. This is understandable: no such models are currently available. The proposed study will be an initial step toward filling this gap; it will use statistical models of archival treatment data to provide decision-making guidance in modular youth psychotherapy, aimed at enhancing its efficiency and effectiveness. Per NIMH Priority 3.2, the proposed project will inform “tailoring existing interventions to optimize outcomes” by “reanalyzing… aggregated clinical trials” using “computational approaches… to facilitate clinical decision-making”. Specifically, the project will aggregate data from six MATCH trials (N=602, ages 6-15), to: model short-term between-person and within- person associations between use of each treatment element and subsequent symptom change at different stages of treatment (Aim 1); use statistical learning to identify moderators of these associations between elements and subsequent symptoms (Aim 2); and, in a holdout sample, test whether agreement between the model recommendations and youths’ treatment course predicts long-term treatment outcomes (Aim 3). To inform future development of clinical decision guidance tools (after the F31), clinicians will be interviewed about how model-based findings might be used in clinical practice (Exploratory Aim 4). Ultimately, consistent with NIM...