Faculty Bibliography

Large language models (LLMs) offer promising applications in mental health care to address gaps in treatment and research. By leveraging clinical notes and transcripts as data, LLMs could improve diagnostics, monitoring, prevention, and treatment of mental health conditions. However, several challenges persist, including technical costs, literacy gaps, risk of biases, and inequalities in data representation. In this Viewpoint, we propose a sociocultural-technical approach to address these challenges. We highlight five key areas for development: (1) building a global clinical repository to support LLMs training and testing, (2) designing ethical usage settings, (3) refining diagnostic categories, (4) integrating cultural considerations during development and deployment, and (5) promoting digital inclusivity to ensure equitable access. We emphasise the need for developing representative datasets, interpretable clinical decision support systems, and new roles such as digital navigators. Only through collaborative efforts across all stakeholders, unified by a sociocultural-technical framework, can we clinically deploy LLMs while ensuring equitable access and mitigating risks.

We present a series of network analyses aiming to uncover the symptom constellations of depression, anxiety and somatization among 2,796 adult primary health care attendees in Goa, India, a low- and middle-income country (LMIC). Depression and anxiety are the leading neuropsychiatric causes of disability. Yet, the diagnostic boundaries and the characteristics of their dynamically intertwined symptom constellations remain obscure, particularly in non-Western settings. Regularized partial correlation networks were estimated and the diagnostic boundaries were explored using community detection analysis. The global and local connectivity of network structures of public versus private healthcare settings and treatment responders versus nonresponders were compared with a permutation test. Overall, depressed mood, panic, fatigue, concentration problems and somatic symptoms were the most central. Leveraging the longitudinal nature of the data, our analyses revealed baseline networks did not differ across treatment responders and nonresponders. The results did not support distinct illness subclusters of the CMDs. For public healthcare settings, panic was the most central symptom, whereas in private, fatigue was the most central. Findings highlight varying mechanism of illness development across socioeconomic backgrounds, with potential implications for case identification and treatment. This is the first study directly comparing the symptom constellations of two socioeconomically different groups in an LMIC.

UNLABELLED: TRIAL REGISTRATION/UNASSIGNED:clinicaltrials.gov Identifier: NCT01179568.

There is an acute need for solutions to treat stress and trauma-related sequelae, and there are well-documented shortages of qualified human professionals. Artificial intelligence (AI) presents an opportunity to create advanced screening, diagnosis, and treatment solutions that relieve the burden on people and can provide just-in-time interventions. Large language models (LLMs), in particular, are promising given the role language plays in understanding and treating traumatic stress and other mental health conditions. In this article, we provide an overview of the state-of-the-art LLMs applications in diagnostic assessments, clinical note generation, and therapeutic support. We discuss the open research direction and challenges that need to be overcome to realize the full potential of deploying language models for use in clinical contexts. We highlight the need for increased representation in AI systems to ensure there are no disparities in access. Public datasets and models will help lead progress toward better models; however, privacy-preserving model training will be necessary for protecting patient data.

The COVID-19 pandemic was an unparalleled stressor that enhanced isolation. Loneliness has been identified as an epidemic by the US Surgeon General. This study aimed to: (1) characterize longitudinal trajectories of loneliness during the acute phase of the COVID-19 pandemic; (2) identify longitudinal mediators of the relationship of loneliness with anxiety and depression; and (3) examine how loneliness naturally clusters and identify factors associated with high loneliness. Two hundred and twenty-nine adults (78% female; mean age = 39.5 ± 13.8) completed an abbreviated version of the UCLA Loneliness Scale, Perceived Stress Scale, Emotion Regulation Questionnaire, State Anxiety Inventory, and Patient Health Questionnaire-8 longitudinally between April 2020 and 2021. Trajectory analyses demonstrated relatively stable loneliness over time, while anxiety and depression symptoms declined. Longitudinal analyses indicated that loneliness effects on anxiety and depression were both partially mediated by perceived stress, while emotion regulation capacity only mediated effects on anxiety. Three stable clusters of loneliness trajectories emerged (high, moderate, and low). The odds of moderate or high loneliness cluster membership were positively associated with higher perceived stress and negatively associated with greater cognitive reappraisal use. Our results demonstrate the important interconnections between loneliness and facets of mental health throughout the early phases of the pandemic and may inform targeted future interventions for loneliness work.

BACKGROUND:Improving safe and effective access to ketamine therapy is of high priority given the growing burden of mental illness. Telehealth-supported administration of sublingual ketamine is being explored toward this goal. METHODS:In this longitudinal study, moderately-to-severely depressed patients received four doses of ketamine at home over four weeks within a supportive digital health context. Treatment was structured to resemble methods of therapeutic psychedelic trials. Patients receiving a second course of treatment were also examined. Symptoms were assessed using the Patient Health Questionnaire (PHQ-9) for depression. We conducted preregistered machine learning and symptom network analyses to investigate outcomes (osf.io/v2rpx). RESULTS:A sample of 11,441 patients was analyzed, demonstrating a modal antidepressant response from both non-severe (n = 6384, 55.8 %) and severe (n = 2070, 18.1 %) baseline depression levels. Adverse events were detected in 3.0-4.8 % of participants and predominantly neurologic or psychiatric in nature. A second course of treatment helped extend improvements in patients who responded favorably to initial treatment. Improvement was most strongly predicted by lower depression scores and age at baseline. Symptoms of Depressed mood and Anhedonia sustained depression despite ongoing treatment. LIMITATIONS/CONCLUSIONS:This study was limited by the absence of comparison or control groups and lack of a fixed-dose procedure for ketamine administration. CONCLUSIONS:At-home, telehealth-supported ketamine administration was largely safe, well-tolerated, and associated with improvement in patients with depression. Strategies for combining psychedelic-oriented therapies with rigorous telehealth models, as explored here, may uniquely address barriers to mental health treatment.

BACKGROUND:Although network analysis studies of psychiatric syndromes have increased in recent years, most have emphasized centrality symptoms and robust edges. Broadening the focus to include bridge symptoms within a systematic review could help to elucidate symptoms having the strongest links in network models of psychiatric syndromes. We conducted this systematic review and statistical evaluation of network analyses on depressive and anxiety symptoms to identify the most central symptoms and bridge symptoms, as well as the most robust edge indices of networks. METHODS:A systematic literature search was performed in PubMed, PsycINFO, Web of Science, and EMBASE databases from their inception to May 25, 2022. To determine the most influential symptoms and connections, we analyzed centrality and bridge centrality rankings and aggregated the most robust symptom connections into a summary network. After determining the most central symptoms and bridge symptoms across network models, heterogeneity across studies was examined using linear logistic regression. RESULTS:Thirty-three studies with 78,721 participants were included in this systematic review. Seventeen studies with 23 cross-sectional networks based on the Patient Health Questionnaire (PHQ) and Generalized Anxiety Disorder (GAD-7) assessments of clinical and community samples were examined using centrality scores. Twelve cross-sectional networks based on the PHQ and GAD-7 assessments were examined using bridge centrality scores. We found substantial variability between study samples and network features. 'Sad mood', 'Uncontrollable worry', and 'Worrying too much' were the most central symptoms, while 'Sad mood', 'Restlessness', and 'Motor disturbance' were the most frequent bridge centrality symptoms. In addition, the connection between 'Sleep' and 'Fatigue' was the most frequent edge for the depressive and anxiety symptoms network model. CONCLUSION/CONCLUSIONS:Central symptoms, bridge symptoms and robust edges identified in this systematic review can be viewed as potential intervention targets. We also identified gaps in the literature and future directions for network analysis of comorbid depression and anxiety.

BACKGROUND:Somatic Symptom and Related Disorders (SSRD), including chronic pain, result in frequent primary care visits, depression and anxiety symptoms, and diminished quality of life. Treatment access remains limited due to structural barriers and functional impairment. Digital delivery offers to improve access and enables transcript analysis via Natural Language Processing (NLP) to inform treatment. Therefore, we investigated asynchronous message-delivered SSRD treatment, and used NLP methods to identify symptom reduction markers from emotional valence. METHODS:173 individuals diagnosed with SSRD received interventions from licensed therapists via messaging 5 days/week for 8 weeks. Depression and anxiety symptoms were measured with the PHQ-9 and GAD-7 from baseline every three weeks. Symptoms trajectories were identified using unsupervised random forest clustering. Emotional valence expressed and use of emotional words were extracted from patients' de-identified transcripts, respectively using VADER and NCR Lexicon. Valence differences were examined using logistic regression. RESULTS:Two subpopulations were identified showing symptoms Improvement (n = 72; 41.62 %) and non-response (n = 101; 58.38 %). Improvement patients expressed more positive valence in the first week of treatment (OR = 1.84, CI: 1.12-3.02; p = .015) and were less likely to express negative valence by the end of treatment (OR = 0.05; CI: 0.30-0.83; p = .008). Non-response patients used more negative valence words, including pain. LIMITATIONS/CONCLUSIONS:Findings were derived from observational data obtained during an ecological intervention, without the inclusion of a control group. CONCLUSIONS:NLP identified linguistic markers distinguishing changes in anxiety and depression symptoms over treatment. Digital interventions offer new forms of delivery and provide the opportunity to automatically collect data for linguistic analysis.

BACKGROUND:Stressors for health care workers (HCWs) during the COVID-19 pandemic have been manifold, with high levels of depression and anxiety alongside gaps in care. Identifying the factors most tied to HCWs' psychological challenges is crucial to addressing HCWs' mental health needs effectively, now and for future large-scale events. OBJECTIVE:In this study, we used natural language processing methods to examine deidentified psychotherapy transcripts from telemedicine treatment during the initial wave of COVID-19 in the United States. Psychotherapy was delivered by licensed therapists while HCWs were managing increased clinical demands and elevated hospitalization rates, in addition to population-level social distancing measures and infection risks. Our goal was to identify specific concerns emerging in treatment for HCWs and to compare differences with matched non-HCW patients from the general population. METHODS:We conducted a case-control study with a sample of 820 HCWs and 820 non-HCW matched controls who received digitally delivered psychotherapy in 49 US states in the spring of 2020 during the first US wave of the COVID-19 pandemic. Depression was measured during the initial assessment using the Patient Health Questionnaire-9, and anxiety was measured using the General Anxiety Disorder-7 questionnaire. Structural topic models (STMs) were used to determine treatment topics from deidentified transcripts from the first 3 weeks of treatment. STM effect estimators were also used to examine topic prevalence in patients with moderate to severe anxiety and depression. RESULTS:The median treatment enrollment date was April 15, 2020 (IQR March 31 to April 27, 2020) for HCWs and April 19, 2020 (IQR April 5 to April 27, 2020) for matched controls. STM analysis of deidentified transcripts identified 4 treatment topics centered on health care and 5 on mental health for HCWs. For controls, 3 STM topics on pandemic-related disruptions and 5 on mental health were identified. Several STM treatment topics were significantly associated with moderate to severe anxiety and depression, including working on the hospital unit (topic prevalence 0.035, 95% CI 0.022-0.048; P<.001), mood disturbances (prevalence 0.014, 95% CI 0.002-0.026; P=.03), and sleep disturbances (prevalence 0.016, 95% CI 0.002-0.030; P=.02). No significant associations emerged between pandemic-related topics and moderate to severe anxiety and depression for non-HCW controls. CONCLUSIONS:The study provides large-scale quantitative evidence that during the initial wave of the COVID-19 pandemic, HCWs faced unique work-related challenges and stressors associated with anxiety and depression, which required dedicated treatment efforts. The study further demonstrates how natural language processing methods have the potential to surface clinically relevant markers of distress while preserving patient privacy.

Neuropsychiatric disorders pose a high societal cost, but their treatment is hindered by lack of objective outcomes and fidelity metrics. AI technologies and specifically Natural Language Processing (NLP) have emerged as tools to study mental health interventions (MHI) at the level of their constituent conversations. However, NLP's potential to address clinical and research challenges remains unclear. We therefore conducted a pre-registered systematic review of NLP-MHI studies using PRISMA guidelines (osf.io/s52jh) to evaluate their models, clinical applications, and to identify biases and gaps. Candidate studies (n = 19,756), including peer-reviewed AI conference manuscripts, were collected up to January 2023 through PubMed, PsycINFO, Scopus, Google Scholar, and ArXiv. A total of 102 articles were included to investigate their computational characteristics (NLP algorithms, audio features, machine learning pipelines, outcome metrics), clinical characteristics (clinical ground truths, study samples, clinical focus), and limitations. Results indicate a rapid growth of NLP MHI studies since 2019, characterized by increased sample sizes and use of large language models. Digital health platforms were the largest providers of MHI data. Ground truth for supervised learning models was based on clinician ratings (n = 31), patient self-report (n = 29) and annotations by raters (n = 26). Text-based features contributed more to model accuracy than audio markers. Patients' clinical presentation (n = 34), response to intervention (n = 11), intervention monitoring (n = 20), providers' characteristics (n = 12), relational dynamics (n = 14), and data preparation (n = 4) were commonly investigated clinical categories. Limitations of reviewed studies included lack of linguistic diversity, limited reproducibility, and population bias. A research framework is developed and validated (NLPxMHI) to assist computational and clinical researchers in addressing the remaining gaps in applying NLP to MHI, with the goal of improving clinical utility, data access, and fairness.