Zachary Daniel Cohen

Evidence-based psychotherapies benefit many patients, but individual outcomes vary and guidance for personalizing treatment remains limited. The verbal theories underlying these treatments describe processes thought to produce improvement but often leave their strength, direction, functional form, and timescales unspecified. This limits theory-informed predictions about how a particular patient would respond to alternative treatments. Researchers have begun formalizing theories of psychopathology in mathematical or computational form to address these limitations. Formal theories make theoretical assumptions explicit, allowing precise and testable predictions to be deduced analytically or through computational models. When these models represent treatment effects and relevant individual differences, researchers can use patient measurements to inform model states and parameters, simulate alternative treatment configurations, and compare predicted outcomes. Empirical evaluation must distinguish the accuracy of outcome predictions, support for the proposed mechanisms, and whether model-guided selection or adaptation improves outcomes relative to other treatment selection strategies. Translating this potential into practice requires aligning the theory with patient assessment, treatment delivery, and evaluation. We propose a roadmap for this process and describe conditions supporting the generation, delivery, and testing of formal theory-driven treatment recommendations. A worked example using a formal theory of panic disorder and a modular digital therapy illustrates the roadmap’s early phases, including provisional model extensions that generate different treatment predictions for two simulated patients. We conclude by discussing challenges in theory development, patient matching, measurement, and empirical evaluation that must be addressed before these predictions can guide clinical care.

The Patient Health Questionnaire-9 (PHQ-9) and its shorter forms are widely used measures of depression symptoms. We evaluated three modular adaptations: (1) separating four double-barreled items into eight single-barreled items (sad mood/hopelessness, insomnia/hypersomnia, poor appetite/overeating, and psychomotor agitation/retardation), (2) adding irritability and low libido items, and (3) revising the instructions and response options to assess symptoms over one week rather than two. We tested these adaptations across six studies (N = 2,467). The single-barreled format showed excellent total score and screening classification agreement with the original format, as well as comparable internal consistency, temporal stability, and convergent validity. It also captured greater symptom-level information: the separated items showed stronger correlations with matched criterion measures, distinct endorsement patterns, and non-redundant inter-correlations. The irritability and libido items correlated moderately to strongly with established measures of these constructs. The one-week versions showed moderate-to-strong internal consistency, convergent validity, and temporal stability. We conclude that these adaptations can be adopted individually or together for symptom-level or past-week assessment.

Purpose: Depression is highly prevalent in middle-aged women (MAW), but head-to-head comparisons of first-line interventions in this population are scarce. This systematic review and individual participant data (IPD) meta-analysis examined the comparative efficacy of antidepressant medication (ADM) and interpersonal psychotherapy (IPT) for MAW (aged 40–64) versus other adults (men and non-MAW) on depression, quality of life, and interpersonal problem measures. Methods: PubMed, PsycINFO, Embase, and the Cochrane Library were searched January 1, 2025 to identify randomized trials comparing ADM and IPT for adults with depression. IPD were requested and analyzed using mixed-effects models. Results: IPD were obtained from 9 of the 15 (60%) identified studies, comprising 1,490 participants (527 MAW, 963 other adults). MAW status did not significantly moderate ADM versus IPT comparative efficacy on depression ( p = .10) and interpersonal problems ( p = .29) outcomes. The two treatments did not significantly differ in treatment effect in the MAW group (depression: b = -0.03, p = .76; interpersonal problems: b = -0.20, p = .31). MAW status significantly moderated comparative efficacy on quality of life measures ( b  = 0.37, p = .02). Whereas IPT was non-significantly more efficacious than ADM for MAW ( b = -0.18, p = .15), ADM were non-significantly more efficacious than IPT for other adults ( b = 0.19, p = .051). Conclusion: This is the first IPD meta-analysis directly comparing depression treatments in MAW. The absence of treatment effect differences indicates that other factors (e.g., patient preference) may guide choice between IPT and ADM in this population.

Recently, Eisele and colleagues (2025) published the ESM-Q, a set of quality criteria for developing and evaluating Experience Sampling Method (ESM) items. While we agree with one of their core recommendations to specify the timeframe an item refers to (e.g., “Right now, I feel happy”), we comment here on how this may be less straightforward than implicitly assumed by ESM researchers. Drawing on our own ESM studies, in which we collected qualitative data from participants, we show that the popular momentary timeframe “right now” is interpreted by participants in different ways, often covering not only the moment just before the assessment but drawing on past, present and anticipated future times. We outline how differences in interpretations can be explained by the linguistic ambiguity of the time adverbs that we use to signify timeframes, as well as by participants’ individual experience of time. We close the commentary with implications on the severity of these differences in interpretation, and provide some concrete recommendations for ESM researchers.

Adding nuance to the between-therapist effect on patient outcomes, research has increasingly demonstrated that a given therapist can differ in their effectiveness depending on who they treat (e.g., patients with different racial/ethnic identities) and/or what they treat (e.g., patients with different presenting problems). This preregistered study examined whether individual therapists are also more or less effective depending on how they treat their patients; that is, delivering one type of therapy versus another. We did so in the context of an individual participant data meta-analysis of clinical trials that compared classes of cognitive-behavioral therapy and psychodynamic therapy for depression. The meta-analytic sample included 30 therapists who were crossed with treatment condition and 492 patients (M = 25.08 patients per therapist; SD = 15.77). Patients completed measures of depression at baseline and posttreatment. Multilevel structural equation models revealed significant variability in the within-therapist treatment condition-outcome association (p < .001), indicating that some therapists were more effective when delivering one treatment over the other. Descriptively, 53% of therapists had similar outcomes across both groups (d < .20), whereas 47% had at least a small-sized treatment-type strength (d ≥ .20; range = 0.21-0.65). Results inform the personalization of treatment usage to the individual provider's effectiveness data. (PsycInfo Database Record (c) 2026 APA, all rights reserved).

Despite growing awareness of and efforts to combat mental illness in recent decades, its global burden continues to rise, and although evidence-based treatments exist, most individuals with mental health symptoms do not receive adequate treatment. Evidence-based digital psychotherapies are highly customizable, can be delivered at low or no cost, and are ideal for addressing issues of access and for developing new and more effective interventions. University of Arizona’s Digital Mental Health Tools (DMHTools; called STAND Digital Therapy at the University of California, Los Angeles [UCLA] where it was originally developed), aims to deliver scalable, effective tools for common mental disorders. DMHTools was developed to address obstacles to accessing high-quality, personalized psychotherapy and to improve our understanding of how and for whom psychological treatments work. Here, we provide a narrative description of the philosophy behind DMHTools and describe its essential components. We then describe its development and note key challenges alongside our decision-making process to highlight looming issues for personalized digital mental health. Finally, we describe preliminary results from ongoing research utilizing DMHTools, and identify future directions for developing, implementing, and disseminating personalized digital therapies at scale.

Introduction: Cognitive behavioural therapy (CBT) and interpersonal psychotherapy (IPT) are both efficacious treatments for depression, but it is less clear how both compare on outcome domains other than depression and in the longer term. Moreover, it is unclear which of these two psychotherapies works better for whom. This article describes the protocol for a systematic review and individual participant data (IPD) meta-analysis that aims to compare the efficacy of CBT and IPT for adults with depression on a range of outcomes in both the short and long term, and to explore moderators of the treatment effect. This study can enhance our understanding of treatments for depression and inform treatment personalisation. Methods and analysis: Systematic literature searches will be conducted in PubMed, PsycINFO, EMBASE and the Cochrane Library from inception to 1 January 2026, to identify randomised clinical trials (RCTs) comparing CBT and IPT for adult depression. Researchers of eligible studies will be invited to contribute their participant-level data. One-stage IPD meta-analyses will be conducted with mixed-effects models to examine (a) treatment efficacy on all outcome measures that are assessed at post-treatment or follow-up in at least two studies, and (b) various baseline participant characteristics as potential moderators of depressive symptom level at treatment completion. Ethics and dissemination: Ethical approval is not required for this study since it will be based on anonymised data from RCTs that have already been completed. The findings of the present study will be disseminated through a peer-reviewed journal or conference presentation.