arXiv subjects

Saeed Abdullah

Publications and source records attributed to Saeed Abdullah.

12 recordsLinked to original sources

arXiv2026

Creating Multilingual Mental Health Dialogue Datasets: Limits of Persona-Based Localization via Nationality and Language

AI and large language models (LLMs) have emerged as promising tools to address global mental health challenges. Despite the global nature of these challenges, there remains a critical shortage of high-quality datasets for training and evaluating such systems. To mitigate this gap, researchers increasingly generate synthetic clinical personas to simulate user data and test digital mental health support systems. However, most validated personas rely on English-centric contexts. This paper investigates whether similar persona-based methods can be used to generate multilingual mental health datasets. We modified nationality and language parameters in personas to generate clinical dialogues in Mandarin, Bengali, and Hindi. We then examined how different LLMs perform when evaluating the depression severity of these generated multilingual datasets against the baseline in English. Our findings indicate that just adding nationality and language parameters in personas might not be adequate, as it can introduce clinical inconsistency across languages. LLM judge models often exhibit inaccuracies in assessing depression severity in non-English texts, with performance varying across different models. This exposes the systemic limitations of applying English-centric personas to multilingual contexts. Ultimately, our work highlights the urgent need for culturally responsive data generation to ensure equitable mental health systems globally.

cs.CL↗

arXiv2026

AI Safety Training Can be Clinically Harmful

Large language models are being deployed as mental health support agents at scale, yet only 16% of LLM-based chatbot interventions have undergone rigorous clinical efficacy testing, and simulations reveal psychological deterioration in over one-third of cases. We evaluate four generative models on 250 Prolonged Exposure (PE) therapy scenarios and 146 CBT cognitive restructuring exercises (plus 29 severity-escalated variants), scored by a three-judge LLM panel. All models scored near-perfectly on surface acknowledgment (~0.91-1.00) while therapeutic appropriateness collapsed to 0.22-0.33 at the highest severity for three of four models, with protocol fidelity reaching zero for two. Under CBT severity escalation, one model's task completeness dropped from 92% to 71% while the frontier model's safety-interference score fell from 0.99 to 0.61. We identify a systematic, modality-spanning failure: RLHF safety alignment disrupts the therapeutic mechanism of action by grounding patients during imaginal exposure, offering false reassurance, inserting crisis resources into controlled exercises, and refusing to challenge distorted cognitions mentioning self-harm in PE; and through task abandonment or safety-preamble insertion during CBT cognitive restructuring. These findings motivate a five-axis evaluation framework (protocol fidelity, hallucination risk, behavioral consistency, crisis safety, demographic robustness), mapped onto FDA SaMD and EU AI Act requirements. We argue that no AI mental health system should proceed to deployment without passing multi-axis evaluation across all five dimensions.

cs.CL↗

arXiv2025

Fine-Tuning Large Audio-Language Models with LoRA for Precise Temporal Localization of Prolonged Exposure Therapy Elements

Prolonged Exposure (PE) therapy is an effective treatment for post-traumatic stress disorder (PTSD), but evaluating therapist fidelity remains labor-intensive due to the need for manual review of session recordings. We present a method for the automatic temporal localization of key PE fidelity elements, identifying their start and stop times, directly from session audio and transcripts. Our approach fine-tunes a large pre-trained audio-language model, Qwen2-Audio, using Low-Rank Adaptation (LoRA) to process focused 30-second windows of audio-transcript input. Fidelity labels for three core protocol phases, therapist orientation (P1), imaginal exposure (P2), and post-imaginal processing (P3), are generated via LLM-based prompting and verified by trained raters. The model is trained to predict normalized boundary offsets using soft supervision guided by task-specific prompts. On a dataset of 308 real PE sessions, our best configuration (LoRA rank 8, 30s windows) achieves a mean absolute error (MAE) of 5.3s across tasks, within typical rater tolerance for timestamp review, enabling practical fidelity QC. We further analyze the effects of window size and LoRA rank, highlighting the importance of context granularity and model adaptation. This work introduces a privacy-preserving, scalable framework for fidelity tracking in PE therapy, with potential to support clinician training, supervision, and quality assurance.

eess.AS↗

arXiv2025

The Pursuit of Empathy: Evaluating Small Language Models for PTSD Dialogue Support

This paper investigates the capacity of small language models (0.5B-5B parameters) to generate empathetic responses for individuals with PTSD. We introduce Trauma-Informed Dialogue for Empathy (TIDE), a novel dataset comprising 10,000 two-turn conversations across 500 diverse, clinically-grounded PTSD personas (https://huggingface.co/datasets/yenopoya/TIDE). Using frontier model outputs as ground truth, we evaluate eight small LLMs in zero-shot settings and after fine-tuning. Fine-tuning enhances empathetic capabilities, improving cosine similarity and perceived empathy, although gains vary across emotional scenarios and smaller models exhibit a "knowledge transfer ceiling." As expected, Claude Sonnet 3.5 consistently outperforms all models, but surprisingly, the smaller models often approach human-rated empathy levels. Demographic analyses showed that older adults favored responses that validated distress before offering support (p = .004), while graduate-educated users preferred emotionally layered replies in specific scenarios. Gender-based differences were minimal (p > 0.15), suggesting the feasibility of broadly empathetic model designs. This work offers insights into building resource-efficient, emotionally intelligent systems for mental health support.

cs.CL↗

arXiv2025

How Real Are Synthetic Therapy Conversations? Evaluating Fidelity in Prolonged Exposure Dialogues

Synthetic data adoption in healthcare is driven by privacy concerns, data access limitations, and high annotation costs. We explore synthetic Prolonged Exposure (PE) therapy conversations for PTSD as a scalable alternative for training clinical models. We systematically compare real and synthetic dialogues using linguistic, structural, and protocol-specific metrics like turn-taking and treatment fidelity. We introduce and evaluate PE-specific metrics, offering a novel framework for assessing clinical fidelity beyond surface fluency. Our findings show that while synthetic data successfully mitigates data scarcity and protects privacy, capturing the most subtle therapeutic dynamics remains a complex challenge. Synthetic dialogues successfully replicate key linguistic features of real conversations, for instance, achieving a similar Readability Score (89.2 vs. 88.1), while showing differences in some key fidelity markers like distress monitoring. This comparison highlights the need for fidelity-aware metrics that go beyond surface fluency to identify clinically significant nuances. Our model-agnostic framework is a critical tool for developers and clinicians to benchmark generative model fidelity before deployment in sensitive applications. Our findings help clarify where synthetic data can effectively complement real-world datasets, while also identifying areas for future refinement.

cs.CL↗

arXiv2025

Thousand Voices of Trauma: A Large-Scale Synthetic Dataset for Modeling Prolonged Exposure Therapy Conversations

The advancement of AI systems for mental health support is hindered by limited access to therapeutic conversation data, particularly for trauma treatment. We present Thousand Voices of Trauma, a synthetic benchmark dataset of 3,000 therapy conversations based on Prolonged Exposure therapy protocols for Post-traumatic Stress Disorder (PTSD). The dataset comprises 500 unique cases, each explored through six conversational perspectives that mirror the progression of therapy from initial anxiety to peak distress to emotional processing. We incorporated diverse demographic profiles (ages 18-80, M=49.3, 49.4% male, 44.4% female, 6.2% non-binary), 20 trauma types, and 10 trauma-related behaviors using deterministic and probabilistic generation methods. Analysis reveals realistic distributions of trauma types (witnessing violence 10.6%, bullying 10.2%) and symptoms (nightmares 23.4%, substance abuse 20.8%). Clinical experts validated the dataset's therapeutic fidelity, highlighting its emotional depth while suggesting refinements for greater authenticity. We also developed an emotional trajectory benchmark with standardized metrics for evaluating model responses. This privacy-preserving dataset addresses critical gaps in trauma-focused mental health data, offering a valuable resource for advancing both patient-facing applications and clinician training tools.

cs.CY↗

arXiv2025

Cognitive Performance Measurements and the Impact of Sleep Quality Using Wearable and Mobile Sensors

Human cognitive performance is an underlying factor in most of our daily lives, and numerous factors influence cognitive performance. In this work, we investigate how changes in sleep quality influence cognitive performance, measured from a dataset collected during a 2-month field study. We collected cognitive performance data (alertness) with the Psychomotor Vigilance Task (PVT), mobile keyboard typing metrics from participants' smartphones, and sleep quality metrics through a wearable sleep tracking ring. Our findings highlight that specific sleep metrics like night-time heart rate, sleep latency, sleep timing, sleep restfulness, and overall sleep quantity significantly influence cognitive performance. To strengthen the current research on cognitive measurements, we introduce smartphone typing metrics as a proxy or a complementary method for continuous passive measurement of cognitive performance. Together, our findings contribute to ubiquitous computing via a longitudinal case study with a novel wearable device, the resulting findings on the association between sleep and cognitive function, and the introduction of smartphone keyboard typing as a proxy of cognitive function.

cs.HC↗

arXiv2023

Designing Voice Interfaces to Support Mindfulness-Based Pain Management

Objective: Chronic pain is a critical public health issue affecting approximately 20% of the adult population in the United States. Given the opioid crisis, there has been an urgent focus on non-addictive pain management methods including Mindfulness-Based Stress Reduction (MBSR). Prior work has successfully used MBSR for pain management. However, ensuring longitudinal engagement to MBSR practices remains a serious challenge. In this work, we explore the utility of a voice interface to support MBSR home practice. Methods: We interviewed ten mindfulness program facilitators to understand how such a technology might fit in the context of the MBSR class and identify potential usability issues with our prototype. We then used directed content analysis to identify key themes and sub-themes within the interview data. Results: Our findings show that facilitators supported the use of the voice interface for MBSR, particularly for individuals with limited motor function. Facilitators also highlighted unique affordances of voice interfaces, including perceived social presence, to support sustained engagement. Conclusion: We demonstrate the acceptability of a voice interface to support home practice for MBSR participants among trained mindfulness facilitators. Based on our findings, we outline design recommendations for technologies aiming to provide longitudinal support for mindfulness-based interventions. Future work should further these efforts towards making non-addictive pain management interventions accessible and efficacious for a wide audience of users.

cs.HC↗

arXiv2023

Supportive Fintech for Individuals with Bipolar Disorder: Financial Data Sharing Preferences to Support Longitudinal Care Management

Financial stability is a key challenge for individuals living with bipolar disorder (BD). Symptomatic periods in BD are associated with poor financial decision-making, contributing to a negative cycle of worsening symptoms and an increased risk of bankruptcy. There has been an increased focus on designing supportive financial technologies (fintech) to address varying and intermittent needs across different stages of BD. However, little is known about this population's expectations and privacy preferences related to financial data sharing for longitudinal care management. To address this knowledge gap, we have deployed a factorial vignette survey using the Contextual Integrity framework. Our data from individuals with BD (N=480) shows that they are open to sharing financial data for long term care management. We have also identified significant differences in sharing preferences across age, gender, and diagnostic subtype. We discuss the implications of these findings in designing equitable fintech to support this marginalized community.

cs.HC↗

arXiv2022

Privacy Sensitive Speech Analysis Using Federated Learning to Assess Depression

Recent studies have used speech signals to assess depression. However, speech features can lead to serious privacy concerns. To address these concerns, prior work has used privacy-preserving speech features. However, using a subset of features can lead to information loss and, consequently, non-optimal model performance. Furthermore, prior work relies on a centralized approach to support continuous model updates, posing privacy risks. This paper proposes to use Federated Learning (FL) to enable decentralized, privacy-preserving speech analysis to assess depression. Using an existing dataset (DAIC-WOZ), we show that FL models enable a robust assessment of depression with only 4--6% accuracy loss compared to a centralized approach. These models also outperform prior work using the same dataset. Furthermore, the FL models have short inference latency and small memory footprints while being energy-efficient. These models, thus, can be deployed on mobile devices for real-time, continuous, and privacy-preserving depression assessment at scale.

cs.HC↗

arXiv2022

Alexa as an Active Listener: How Backchanneling Can Elicit Self-Disclosure and Promote User Experience

Active listening is a well-known skill applied in human communication to build intimacy and elicit self-disclosure to support a wide variety of cooperative tasks. When applied to conversational UIs, active listening from machines can also elicit greater self-disclosure by signaling to the users that they are being heard, which can have positive outcomes. However, it takes considerable engineering effort and training to embed active listening skills in machines at scale, given the need to personalize active-listening cues to individual users and their specific utterances. A more generic solution is needed given the increasing use of conversational agents, especially by the growing number of socially isolated individuals. With this in mind, we developed an Amazon Alexa skill that provides privacy-preserving and pseudo-random backchanneling to indicate active listening. User study (N = 40) data show that backchanneling improves perceived degree of active listening by smart speakers. It also results in more emotional disclosure, with participants using more positive words. Perception of smart speakers as active listeners is positively associated with perceived emotional support. Interview data corroborate the feasibility of using smart speakers to provide emotional support. These findings have important implications for smart speaker interaction design in several domains of cooperative work and social computing.

cs.HC↗

arXiv2022

Financial technologies (FinTech) for mental health: The potential of objective financial data to better understand the relationships between financial behavior and mental health

In this paper, we present novel research methods for collecting and analyzing personal financial data alongside mental health factors, illustrated through a N=1 case study using data from one individual with bipolar disorder. While we have not found statistically significant trends nor our findings are generalizable beyond this case, our approach provides an insight into the challenges of accessing objective financial data. We outline what data is currently available, what can be done with it, and what factors to consider when working with financial data. More specifically, using these methods researchers might be able to identify symptomatic traces of mental ill health in personal financial data such as identifying early warning signs and thereby enable preemptive care for individuals with serious mental illnesses. Based on this work, we have also explored future directions for developing interventions to support financial wellbeing. Furthermore, we have described the technical, ethical, and equity challenges for financial data-driven assessments and intervention methods, as well as provided a broad research agenda to address these challenges. By leveraging objective, personalized financial data in a privacy-preserving and ethical manner help lead to a shift in mental health care.

cs.HC↗