Perceived stigma and its role in substance use disorder treatment completion

The American Journal of Drug and Alcohol Abuse · 2025-07-31 · 2 citations

These findings underscore the importance of addressing perceived stigma at treatment intake and its role in predicting treatment retention. Routine screening for stigma and implementing stigma-reduction interventions during care may contribute to better treatment outcomes for individuals with SUDs.

Communities at the Crossroads: Exploring Cultural Variations in Opioid Mortality Across the United States

Drug and Alcohol Dependence · 2025-02-01

Knowing When Not to Answer: Lightweight KB-Aligned OOD Detection for Safe RAG

ArXiv.org · 2025-08-04 · 1 citations

Retrieval-Augmented Generation (RAG) systems are increasingly deployed in high-stakes domains, where safety depends not only on how a system answers, but also on whether a query should be answered given a knowledge base (KB). Out-of-domain (OOD) queries can cause dense retrieval to surface weakly related context and lead the generator to produce fluent but unjustified responses. We study lightweight, KB-aligned OOD detection as an always-on gate for RAG systems. Our approach applies PCA to KB embeddings and scores queries in a compact subspace selected either by explained-variance retention (EVR) or by a separability-driven t-test ranking. We evaluate geometric semantic-search rules and lightweight classifiers across 16 domains, including high-stakes COVID-19 and Substance Use KBs, and stress-test robustness using both LLM-generated attacks and an in-the-wild 4chan attack. We find that low-dimensional detectors achieve competitive OOD performance while being faster, cheaper, and more interpretable than prompted LLM-based judges. Finally, human and LLM-based evaluations show that OOD queries primarily degrade the relevance of RAG outputs, showing the need for efficient external OOD detection to maintain safe, in-scope behavior.

Using LLMs to Aid Annotation and Collection of Clinically-Enriched Data in Bipolar Disorder and Schizophrenia

2025-01-01 · 2 citations

Ankit Aich, Avery Quynh, Pamela Osseyi, Amy Pinkham, Philip Harvey, Brenda Curtis, Colin Depp, Natalie Parde. Proceedings of the 10th Workshop on Computational Linguistics and Clinical Psychology (CLPsych 2025). 2025.

Modeling Human Subjectivity in LLMs Using Explicit and Implicit Human Factors in Personas

arXiv (Cornell University) · 2024-06-20

Large language models (LLMs) are increasingly being used in human-centered social scientific tasks, such as data annotation, synthetic data creation, and engaging in dialog. However, these tasks are highly subjective and dependent on human factors, such as one's environment, attitudes, beliefs, and lived experiences. Thus, it may be the case that employing LLMs (which do not have such human factors) in these tasks results in a lack of variation in data, failing to reflect the diversity of human experiences. In this paper, we examine the role of prompting LLMs with human-like personas and asking the models to answer as if they were a specific human. This is done explicitly, with exact demographics, political beliefs, and lived experiences, or implicitly via names prevalent in specific populations. The LLM personas are then evaluated via (1) subjective annotation task (e.g., detecting toxicity) and (2) a belief generation task, where both tasks are known to vary across human factors. We examine the impact of explicit vs. implicit personas and investigate which human factors LLMs recognize and respond to. Results show that explicit LLM personas show mixed results when reproducing known human biases, but generally fail to demonstrate implicit biases. We conclude that LLMs may capture the statistical patterns of how people speak, but are generally unable to model the complex interactions and subtleties of human perceptions, potentially limiting their effectiveness in social science applications.

Integrating Patient-Generated Digital Data into Mental Health Therapy: A mixed methods analysis of user experience (Preprint)

2024-04-23

<sec> <title>UNSTRUCTURED</title> Therapists and their patients increasingly discuss digital data from social media, smartphone sensors, and other online engagement within the context of psychotherapy. We examined patients’ and mental health therapists’ experiences and perceptions following a randomized controlled trial (RCT) in which they both received regular summaries of patients’ digital data (e.g., dashboard) to review and discuss in session. The dashboard included data which patients consented to share from their social media posts, phone usage and online searches. Following the RCT, patient (n=56) and therapist (n=44) participants completed a debriefing survey after their study completion (from December 2021 - January 2022). Participants were asked about their experience receiving a digital data dashboard in psychotherapy via closed- and open-ended questions. We calculated descriptive statistics for closed-ended questions and conducted qualitative coding via NVivo 10 and natural language processing using the machine learning tool Latent Dirichlet Allocation to analyze open-ended questions. Of 100 participants, nearly half (49%) described their experience with the dashboard as “positive,” while the other half noted a “neutral” experience. Responses to the open-ended questions resulted in three thematic areas (9 sub-categories): (1) dashboard experience (positive; neutral or negative; comfortable), (2) perception of the dashboard’s impact on enhancing therapy (accountability; increased awareness over time; objectivity), and (3) dashboard refinements (additional sources; tailored content; ethics). Patients reported that receiving their digital data helped them stay “accountable,” while therapists indicated that dashboard helped “tailor treatment plans.” Patient and therapist surveys provided important feedback their experience regularly discussing dashboards in psychotherapy. </sec>

Reply to Wang: Clarifying model performance and language markers of depression across races

Proceedings of the National Academy of Sciences · 2024-07-25 · 1 citations

Large volumes of liquid water transiently existed on the surface of Mars more than 3 billion years ago. Much of this water is hypothesized to have been sequestered in the subsurface or lost to space. We use rock physics models and Bayesian inversion to ...

Modeling Human Subjectivity in LLMs Using Explicit and Implicit Human Factors in Personas

2024-01-01 · 3 citations

Salvatore Giorgi, Tingting Liu, Ankit Aich, Kelsey Jane Isman, Garrick Sherman, Zachary Fried, João Sedoc, Lyle Ungar, Brenda Curtis. Findings of the Association for Computational Linguistics: EMNLP 2024. 2024.

Designing clinical trials to address alcohol use and alcohol-associated liver disease: an expert panel Consensus Statement

Nature Reviews Gastroenterology & Hepatology · 2024-06-07 · 44 citations

Integrating Patient-Generated Digital Data Into Mental Health Therapy: Mixed Methods Analysis of User Experience

JMIR Mental Health · 2024-10-04 · 2 citations

Background: Therapists and their patients increasingly discuss digital data from social media, smartphone sensors, and other online engagements within the context of psychotherapy. Objective: We examined patients' and mental health therapists' experiences and perceptions following a randomized controlled trial in which they both received regular summaries of patients' digital data (eg, dashboard) to review and discuss in session. The dashboard included data that patients consented to share from their social media posts, phone usage, and online searches. Methods: Following the randomized controlled trial, patient (n=56) and therapist (n=44) participants completed a debriefing survey after their study completion (from December 2021 to January 2022). Participants were asked about their experience receiving a digital data dashboard in psychotherapy via closed- and open-ended questions. We calculated descriptive statistics for closed-ended questions and conducted qualitative coding via NVivo (version 10; Lumivero) and natural language processing using the machine learning tool latent Dirichlet allocation to analyze open-ended questions. Results: Of 100 participants, nearly half (n=48, 49%) described their experience with the dashboard as "positive," while the other half noted a "neutral" experience. Responses to the open-ended questions resulted in three thematic areas (nine subcategories): (1) dashboard experience (positive, neutral or negative, and comfortable); (2) perception of the dashboard's impact on enhancing therapy (accountability, increased awareness over time, and objectivity); and (3) dashboard refinements (additional sources, tailored content, and privacy). Conclusions: Patients reported that receiving their digital data helped them stay "accountable," while therapists indicated that the dashboard helped "tailor treatment plans." Patient and therapist surveys provided important feedback on their experience regularly discussing dashboards in psychotherapy.