A deep learning quantification of patient specificity as a predictor of session attendance and treatment response to internet-enabled cognitive behavioural therapy for common mental health disorders.

BACKGROUND: Increasing an individual's ability to focus on concrete, specific detail, thus reducing the tendency toward overly broad, decontextualised generalisations about the self and world, is a target within cognitive behavioural therapy (CBT). However, empirical investigation of the impact of within-treatment specificity on treatment outcomes is scarce. We evaluated whether the specificity of patient dialogue predicted a) end-of-treatment symptoms and b) session completion for CBT for common mental health issues. METHODS: This preregistered (https://osf.io/agr4t) study trained a deep learning model to score the specificity of patient dialogue in transcripts from 353,614 internet-enabled CBT sessions for common mental health disorders, delivered on behalf of UK NHS services. Data were from obtained from 65,030 participants (n = 47,308 female, n = 241 unstated) aged 18-94 years (M = 34.69, SD = 12.35). Depressive disorders were the most common (39.1 %) primary diagnosis. Primary outcome was end-of-treatment score on the Patient Health Questionnaire-9 (PHQ-9). Secondary outcome was number of sessions attended. RESULTS: Linear mixed-effects models demonstrated that increased patient specificity significantly predicted lower post-treatment symptoms on the PHQ-9, although the size and direction of the effect varied depending on the type of therapeutic activity being completed. Effect sizes were consistently small. Higher patient specificity was associated with completing a greater number of sessions. LIMITATIONS: We are unable to infer causation from our data. CONCLUSIONS: Although effect sizes were small, an effect of specificity was observed across common mental health disorders. Further studies are needed to explore whether encouraging patient specificity during CBT may provide an enhancement of treatment attendance and treatment effects.

Refining our understanding of depressive states and state transitions in response to cognitive behavioural therapy using latent Markov modelling.

BACKGROUND: It is increasingly recognized that existing diagnostic approaches do not capture the underlying heterogeneity and complexity of psychiatric disorders such as depression. This study uses a data-driven approach to define fluid depressive states and explore how patients transition between these states in response to cognitive behavioural therapy (CBT). METHODS: Item-level Patient Health Questionnaire (PHQ-9) data were collected from 9891 patients with a diagnosis of depression, at each CBT treatment session. Latent Markov modelling was used on these data to define depressive states and explore transition probabilities between states. Clinical outcomes and patient demographics were compared between patients starting at different depressive states. RESULTS: A model with seven depressive states emerged as the best compromise between optimal fit and interpretability. States loading preferentially on cognitive/affective v. somatic symptoms of depression were identified. Analysis of transition probabilities revealed that patients in cognitive/affective states do not typically transition towards somatic states and vice-versa. Post-hoc analyses also showed that patients who start in a somatic depressive state are less likely to engage with or improve with therapy. These patients are also more likely to be female, suffer from a comorbid long-term physical condition and be taking psychotropic medication. CONCLUSIONS: This study presents a novel approach for depression sub-typing, defining fluid depressive states and exploring transitions between states in response to CBT. Understanding how different symptom profiles respond to therapy will inform the development and delivery of stratified treatment protocols, improving clinical outcomes and cost-effectiveness of psychological therapies for patients with depression.

Quantifying the Association Between Psychotherapy Content and Clinical Outcomes Using Deep Learning.

Importance: Compared with the treatment of physical conditions, the quality of care of mental health disorders remains poor and the rate of improvement in treatment is slow, a primary reason being the lack of objective and systematic methods for measuring the delivery of psychotherapy. Objective: To use a deep learning model applied to a large-scale clinical data set of cognitive behavioral therapy (CBT) session transcripts to generate a quantifiable measure of treatment delivered and to determine the association between the quantity of each aspect of therapy delivered and clinical outcomes. Design, Setting, and Participants: All data were obtained from patients receiving internet-enabled CBT for the treatment of a mental health disorder between June 2012 and March 2018 in England. Cognitive behavioral therapy was delivered in a secure online therapy room via instant synchronous messaging. The initial sample comprised a total of 17 572 patients (90 934 therapy session transcripts). Patients self-referred or were referred by a primary health care worker directly to the service. Exposures: All patients received National Institute for Heath and Care Excellence-approved disorder-specific CBT treatment protocols delivered by a qualified CBT therapist. Main Outcomes and Measures: Clinical outcomes were measured in terms of reliable improvement in patient symptoms and treatment engagement. Reliable improvement was calculated based on 2 severity measures: Patient Health Questionnaire (PHQ-9) and Generalized Anxiety Disorder 7-item scale (GAD-7), corresponding to depressive and anxiety symptoms respectively, completed by the patient at initial assessment and before every therapy session (see eMethods in the Supplement for details). Results: Treatment sessions from a total of 14 899 patients (10 882 women) aged between 18 and 94 years (median age, 34.8 years) were included in the final analysis. We trained a deep learning model to automatically categorize therapist utterances into 1 or more of 24 feature categories. The trained model was applied to our data set to obtain quantifiable measures of each feature of treatment delivered. A logistic regression revealed that increased quantities of a number of session features, including change methods (cognitive and behavioral techniques used in CBT), were associated with greater odds of reliable improvement in patient symptoms (odds ratio, 1.11; 95% CI, 1.06-1.17) and patient engagement (odds ratio, 1.20, 95% CI, 1.12-1.27). The quantity of nontherapy-related content was associated with reduced odds of symptom improvement (odds ratio, 0.89; 95% CI, 0.85-0.92) and patient engagement (odds ratio, 0.88, 95% CI, 0.84-0.92). Conclusions and Relevance: This work demonstrates an association between clinical outcomes in psychotherapy and the content of therapist utterances. These findings support the principle that CBT change methods help produce improvements in patients' presenting symptoms. The application of deep learning to large clinical data sets can provide valuable insights into psychotherapy, informing the development of new treatments and helping standardize clinical practice.