Brian Schwartz

Background: The Personalized Advantage Index (PAI) has evolved into a standard framework to identify the individually optimal treatment from available alternatives in mental health and beyond. The traditional PAI compares a patient’s predicted post-treatment symptom scores across treatment alternatives to identify the treatment with the most favorable predicted score. However, clinicians may wish to select the treatment that maximizes that patient’s predicted probability of achieving a clinically defined outcome, such as remission. This study presents an Alternative PAI (AltPAI) optimized for this purpose and evaluates its performance. Methods: We used routine care data (n = 1,264) to implement the traditional PAI and AltPAI using random forests. The indices were compared regarding the benefit for patients in terms of dimensional improvement and remission. Exploratively, the PAI and AltPAI were combined into an ensemble (EnsPAI). Results: The AltPAI outperformed the PAI descriptively but not significantly, yielding higher symptom change (cohens d = 0.09) and remission probability (OR = 1.32) for optimally treated patients with higher index values. Regarding symptom improvement, the PAI performed worse than chance. The EnsPAI was the only algorithm to perform significantly better than chance in terms of symptom remission. Conclusions: The AltPAI and EnsPAI demonstrated performance that was generally descriptively superior to the PAI. The approaches should therefore be investigated in combination with the PAI in future treatment stratification analyses. Additionally, more advanced implementations of the EnsPAI are discussed that would allow the ensemble to be calibrated on the error of the base indices, potentially increasing prediction performance.

Clinicians often struggle to accurately identify patients at risk of treatment failure. Feedback-informed treatment offers real-time outcome monitoring to help detect and address potential treatment failures early. This study examined patterns of change among not-on-track (NOT) patients during feedback-informed treatment, explored how patients' characteristics related to these patterns, and assessed the feasibility of predicting them. Data from 573 NOT patients in a university outpatient psychotherapy clinic were analyzed. Growth mixture modeling identified distinct change trajectories, omnibus tests compared patients' characteristics across patterns, and multinomial logistic regression evaluated the predictive value of these characteristics. Three distinct patterns of change emerged, differing in symptom severity at the point of being flagged as NOT and in subsequent treatment response: C1: severe-modest responders, C2: severe-fast improvers, and C3: moderate-improvers. The patterns were differentiated by key intake variables, including symptom severity, personality disorder traits, interpersonal problems, and emotion regulation difficulties. Two of the patterns were predicted with acceptable accuracy. All patient subgroups demonstrated improvement. Nevertheless, they became NOT for different reasons, and their responses were associated with different clinical characteristics. These findings suggest that a NOT alert should not be interpreted as a homogeneous signal, and patients may require different adaptations. (PsycInfo Database Record (c) 2026 APA, all rights reserved).

Clinical or methodological significance of this article: Premature treatment termination in inpatient psychotherapy can have severe negative consequences for the individual patient and the mental h...

Zusammenfassung: Künstliche Intelligenz (KI) hat in den letzten Jahren große Fortschritte in der Medizin erzielt, insbesondere bei der Analyse medizinischer Bilddaten. Weniger im Fokus steht bislang ihr Potenzial als Analysetool für die Psychologie und Psychotherapie, obwohl aktuelle Entwicklungen zunehmend zeigen, dass KI-basierte Verfahren wertvolle Beiträge zur Diagnostik psychischer Störungen und zur Analyse psychotherapeutischer Gespräche leisten können. Ziel dieses Beitrags ist es, den internationalen Forschungsstand zu KI-basierten Werkzeugen und Techniken für Diagnostik und Gesprächsanalyse darzustellen, deren Potenziale und Grenzen kritisch zu diskutieren und zukünftige Anforderungen für eine verantwortungsvolle klinische Nutzung aufzuzeigen.

Self‐guided digital mental health apps often show high attrition, highlighting the need to understand engagement beyond behavioural metrics. The Homework Rating Scale—Mobile Application (HRS‐MA) was developed to assess cognitive, emotional and attitudinal aspects of engagement. This study evaluated its psychometric properties. The HRS‐MA was integrated into the MoodMission app and administered after 30 days of use, alongside additional self‐report measures. Data from 2100 users were analysed for reliability, construct, convergent, divergent and predictive validity. The HRS‐MA showed excellent internal consistency (Cronbach's α = 0.90; McDonald's ω = 0.94). Factor analyses supported a two‐factor structure—engagement and difficulty and obstacles. The measure demonstrated strong convergent and divergent validity and modest predictive validity. A derived coefficient of psychological effort, representing the interaction between engagement and perceived barriers, significantly predicted symptom reduction, particularly in depression. The HRS‐MA is a psychometrically sound tool for capturing cognitive and attitudinal dimensions of engagement in digital interventions. It complements behavioural indicators by offering a more comprehensive view of user experience and effort, providing researchers and developers with a valuable framework for optimizing mental health app design and evaluation.

Therapists vary widely in their use of feedback, which may reflect differences in how they incorporate feedback into their clinical judgment-especially when their evaluations diverge from their patients'. This process, often referred to as belief updating, involves adapting one's perspective based on new information and is thought to be a key mechanism underlying feedback effect on therapy outcomes. This study investigates whether therapists who align their ratings more closely with patient feedback-reflecting stronger integration and potential belief updating-benefit more from feedback. Cross-lagged effects of patient (PR) and therapist (TR) symptom severity and well-being ratings were analyzed from 1,019 therapist-patient dyads (729 received feedback) using dynamic structural equation modeling. Feedback integration was operationalized as the cross-lagged relationship between PR and subsequent TR under feedback conditions compared to no feedback. PR were significantly associated with TR in the subsequent session. Feedback as a condition was not found to predict this cross-lagged association. Therapists showed significant variability (3.64%) regarding the association between PR and their own TR. Therapists' degree of feedback integration moderated the effect of feedback on patient outcomes under feedback conditions. Therapists who incorporated patient feedback benefited more from feedback interventions. This integration aligns with the belief updating process, hypothesized to be central to feedback effect. By adjusting their clinical judgments to reflect patient perspectives, therapists may improve treatment effectiveness. These findings emphasize the importance of therapist-specific factors and highlight how feedback and belief updating interact to shape therapeutic outcomes. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Background: Although mental disorders are highly prevalent among older adults, evidence suggests that they underutilize psychotherapy. However, formal estimates of their actual representation in routine clinical settings are scarce. This study applied a representation quotient approach to identify and quantify age-related disparities in outpatient psychotherapy utilization in Germany. Methods: We analyzed data from 13,635 adult patients who initiated cognitive-behavioral therapy between 2018 and 2023 at 29 university outpatient clinics. Using a representation quotient approach, we compared the observed proportions of young-old adults (65–74 years) and old-old adults (≥75 years) with prevalence-stratified population age distributions. The robustness of the results was investigated by extensive sensitivity analyses, including alternative prevalence assumptions and adjustments for long-term care needs. Results: Even under conservative assumptions, young-old adults only accounted for about 25% and old-old adults for about 13% of their expected proportion. In contrast, young working-age adults (18–34 years) were particularly overrepresented in the sample. Underrepresentation of older adults was most pronounced among patients with post-traumatic stress disorder, eating disorders, and dysthymia, less pronounced among patients with generalized anxiety disorder, and not significant among patients with pain disorders. Conclusion: This study demonstrates the utility of representation quotients for systematically identifying and quantifying age-related disparities in psychotherapy utilization based on routine care data. Our analyses of large-scale data from university outpatient clinics in Germany revealed a marked underrepresentation of older adults in this setting. These findings highlight the need to improve access to, and utilization of, psychotherapeutic services for older adults.

Rating scales have shaped psychological research, but are resource-intensive and can burden participants. Large Language Models (LLMs) offer a tool to assess latent constructs in text. This study introduces LLM rating scales, which use LLM responses instead of human ratings. We demonstrate this approach with an LLM rating scale measuring patient engagement in therapy transcripts. Automatically transcribed videos of 1,131 sessions from 155 patients were analyzed using DISCOVER, a software framework for local multimodal human behavior analysis. Llama 3.1 8B LLM rated 120 engagement items, averaging the top eight into a total score. Psychometric evaluation showed a normal distribution, strong reliability (ω = 0.953), and acceptable fit (CFI = 0.968, SRMR = 0.022), except RMSEA = 0.108. Validity was supported by significant correlations with engagement determinants (e.g., motivation, r = .413), processes (e.g., between-session efforts, r = .390), and outcomes (e.g., symptoms, r = − .304). Results remained robust across bootstrap resampling and cross-validation, accounting for nested data. The LLM rating scale exhibited strong psychometric properties, demonstrating the potential of the approach as an assessment tool. Importantly, this automated approach uses interpretable items, ensuring clear understanding of measured constructs, while supporting local implementation and protecting confidential data.

Introduction: The mechanisms of change in psychological therapies have not yet been sufficiently empirically validated. In particular, the time course of their effects on patients’ symptom distress is unknown. Therefore, the aims of the present study were (1) to examine the effects of 4 patient experiences, namely, interpersonal experience (IE), coping experience (CE), affective experience (AE), and resource-related experience (RE), on symptom distress within and between patients; (2) to consider the temporal development of the within-patient effects; and (3) to examine their associations with the final treatment outcome. Methods: Within- and between-patient effects of IE, CE, AE, and RE on session-level symptom distress were calculated for n = 2,208 outpatients treated by n = 253 therapists. Patient experiences were measured after each session using session reports, and outcome was assessed with the HSCL-11 at the beginning of each session and with the OQ-30 after treatment. Linear mixed models (LMMs) were used to examine the within- and between-patient associations. Temporal changes in the within-patient effects were modeled using rolling windows and three-level LMMs (sessions nested in windows and patients). Results: Within-patient components of all 4 patient experiences were significantly associated with symptom distress (b = −0.006 to b = −0.057), while none of the between-patient components were. The effects of all 4 patient experiences diminished over time (b = 0.000 to b = 0.001). The effects of IE, CE, and RE remained negative and significant, but the effect of AE crossed zero and became significantly positive with an increasing window number. A greater decrease in the within-patient effects of IE (b = −0.001) and CE (b = −0.001) on HSCL-11 was associated with better final treatment outcomes. Conclusion: The results suggest that all four change factors are drivers of successful therapy, particularly in the early sessions. Over the course of treatment, their impact on symptom distress diminishes, which appears to be beneficial for the final treatment outcome. In the future, differences between patients in the temporal dynamics of change factors should be investigated to inform transtheoretical and process-based treatments that support therapists in personalizing their treatments to individual patients.