arxiv: 2604.04915 · v1 · submitted 2026-04-06 · 💻 cs.HC

Recognition: 2 theorem links

· Lean Theorem

Exploring Expert Perspectives on Wearable-Triggered LLM Conversational Support for Daily Stress Management

Poorvesh Dongre , Sameer Neupane , Priyanka Jadhav , Nikitha Donekal Chandrashekar , Christian Webb , Denis Gra\v{c}anin

Authors on Pith no claims yet

Pith reviewed 2026-05-10 19:49 UTC · model grok-4.3

classification 💻 cs.HC

keywords wearable devicesstress detectionLLM conversational supportmental healthdesign probeexpert interviewsdaily stress management

0 comments

The pith

A functional prototype app that triggers LLM chats from wearable stress detection, when shown to experts, surfaces early design tensions for daily mental health tools.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper builds EmBot, a working mobile app that links wearable stress sensing to generative conversations for on-the-spot support. It then deploys this app as a probe during interviews with fifteen mental health experts. The work seeks to reveal concrete design tensions that appear when detection hardware directly initiates AI dialogue. A reader would care because wearable sensors and conversational models are already available, yet their joint use for everyday stress remains uncharted and risks missteps without targeted guidance.

Core claim

By creating EmBot as a fully functional prototype and using it in semi-structured interviews with mental health experts, the authors surface initial design tensions and considerations that emerge specifically from wearable-triggered LLM conversational support, thereby informing how such combined systems should be shaped for daily stress management.

What carries the argument

EmBot, the functional mobile application that pairs wearable-triggered stress detection with LLM-generated conversational support and serves as the design probe to draw out expert perspectives.

If this is right

Systems that connect wearable stress detection to LLM conversations should be shaped around the tensions and considerations identified by mental health experts.
Early use of working prototypes in expert consultations can expose practical issues that abstract descriptions miss.
Future designs for daily mental health support can draw on these surfaced considerations to improve relevance and safety.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same probe approach could be tested directly with people who experience daily stress to see whether user views align with or diverge from expert ones.
The method of linking detection events to generative dialogue might be extended to other momentary states such as anxiety spikes or mood shifts.
Longer deployment studies could check whether addressing the identified tensions actually changes how users engage with or benefit from the support.

Load-bearing premise

That the perspectives gathered from 15 mental health experts using the EmBot design probe are sufficient to identify broadly relevant design tensions and considerations for wearable-triggered LLM systems in daily stress management.

What would settle it

A larger or more diverse follow-up study in which experts or actual users report no meaningful design tensions or entirely different ones from those identified with the fifteen interviewees would undermine the claim that the probe method yields broadly useful guidance.

Figures

Figures reproduced from arXiv: 2604.04915 by Christian Webb, Denis Gra\v{c}anin, Nikitha Donekal Chandrashekar, Poorvesh Dongre, Priyanka Jadhav, Sameer Neupane.

read the original abstract

Wearable devices increasingly support stress detection, while LLMs enable conversational mental health support. However, designing systems that meaningfully connect wearable-triggered stress events with generative dialogue remains underexplored, particularly from a design perspective. We present EmBot, a functional mobile application that combines wearable-triggered stress detection with LLM-based conversational support for daily stress management. We used EmBot as a design probe in semi-structured interviews with 15 mental health experts to examine their perspectives and surface early design tensions and considerations that arise from wearable-triggered conversational support, informing the future design of such systems for daily stress management and mental health support.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper builds a working EmBot prototype that triggers LLM chats from wearable stress data and uses it to surface expert design tensions, which is a concrete step forward for this HCI niche.

read the letter

This is an exploratory HCI study that puts together a functional mobile app linking wearable stress detection to LLM conversations for daily use. They treat the app as a design probe and run semi-structured interviews with 15 mental health experts to pull out early tensions around timing, privacy, response style, and integration into routines. The prototype itself is the clearest new piece; having something that actually runs gives the experts a shared reference point instead of vague hypotheticals, and the resulting tensions read as practical rather than abstract. That approach fits the scope they set for themselves—early considerations, not validated guidelines or efficacy claims. The abstract keeps the claims modest, which helps. The main soft spot is the thin methods description. Recruitment details, interview guide, and how they moved from transcripts to the listed tensions are not visible here, so it is hard to judge how systematically the data were handled or whether the sample of experts covers enough variation. If the full paper supplies a standard thematic analysis with clear steps, that gap closes. The work stays within qualitative design-probe norms and does not overclaim. It is aimed at HCI researchers and digital-health designers who are already thinking about wearable-AI combinations for mental health. A reader looking for concrete examples of expert pushback on real-time LLM support would find usable material. It deserves a serious referee because the prototype is implemented and the interview data exist; any review would mainly tighten the methods and perhaps ask for more on limitations. I would send it to review at a conference like CHI or DIS with the expectation of moderate revisions rather than desk rejection.

Referee Report

1 major / 1 minor

Summary. The paper presents EmBot, a functional mobile application integrating wearable-triggered stress detection with LLM-based conversational support for daily stress management. It deploys EmBot as a design probe in semi-structured interviews with 15 mental health experts to elicit their perspectives and identify early design tensions and considerations for such systems, with the goal of informing future design of wearable-LLM tools for mental health support.

Significance. As an exploratory HCI design-probe study, the work offers timely initial insights into the underexplored intersection of wearable sensing, generative AI dialogue, and daily stress management. The use of a concrete, functional prototype strengthens the elicitation of grounded expert feedback compared to purely hypothetical scenarios. If the derived tensions prove robust, the paper can usefully guide ethical and practical considerations in an emerging application area.

major comments (1)

The manuscript supplies no details on recruitment (selection criteria, expert backgrounds, or sampling strategy for the 15 participants), the semi-structured interview protocol, the qualitative analysis method (e.g., thematic analysis steps or coding process), or the procedure used to derive and validate the reported design tensions from the interview data. These omissions make it impossible to assess the rigor or traceability of the central claims about surfaced tensions. (See Study Procedure and Analysis sections.)

minor comments (1)

The abstract could explicitly note the exploratory scope and the number of expert participants to better set reader expectations.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their thoughtful review and positive assessment of the timeliness and value of our exploratory design-probe study. We agree that greater methodological transparency is essential for assessing the rigor of our qualitative findings and will revise the manuscript accordingly.

read point-by-point responses

Referee: The manuscript supplies no details on recruitment (selection criteria, expert backgrounds, or sampling strategy for the 15 participants), the semi-structured interview protocol, the qualitative analysis method (e.g., thematic analysis steps or coding process), or the procedure used to derive and validate the reported design tensions from the interview data. These omissions make it impossible to assess the rigor or traceability of the central claims about surfaced tensions. (See Study Procedure and Analysis sections.)

Authors: We acknowledge this is a valid and important point; the current version of the manuscript does not provide sufficient detail on these elements, which limits traceability. In the revised manuscript we will expand the Study Procedure and Analysis sections to include: (1) recruitment details specifying selection criteria (e.g., minimum 3 years of clinical experience in mental health), participant backgrounds (e.g., mix of psychologists, counselors, and psychiatrists with years of experience and primary focus areas), and sampling strategy (purposive sampling via professional networks and snowball referrals); (2) the complete semi-structured interview protocol, including the core question guide and example probes; (3) a step-by-step account of the thematic analysis process (familiarization, open coding, theme generation, review, and refinement); and (4) the explicit procedure used to surface and validate design tensions (iterative clustering by the research team, cross-checking against raw transcripts, and resolution of disagreements through discussion). These additions will directly address the referee's concern and strengthen the paper. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper is a qualitative HCI design-probe study that builds a functional prototype (EmBot) and uses it to conduct semi-structured interviews with 15 mental health experts, followed by thematic analysis to surface design tensions. There are no equations, fitted parameters, predictions, or mathematical derivations of any kind. Claims rest directly on the collected interview data and standard qualitative practices rather than any self-referential definitions, self-citation chains, or reductions of outputs to inputs by construction. The central contribution is exploratory and bounded to the probe-elicited perspectives, with no load-bearing steps that collapse into prior inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is a qualitative design-probe study in HCI with no mathematical models, fitted parameters, or postulated entities; it relies on standard interview methods and prototype development.

pith-pipeline@v0.9.0 · 5422 in / 1095 out tokens · 48038 ms · 2026-05-10T19:49:53.164605+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We present EmBot, a functional mobile application that combines wearable-triggered stress detection with LLM-based conversational support... semi-structured interviews with 15 mental health experts... surface early design tensions
IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Findings on detection transparency, notification calibration, structured conversational support, safety/privacy

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

34 extracted references · 12 canonical work pages

[1]

Alaa Abd-Alrazaq, Rawan AlSaad, Sarah Aziz, Arfan Ahmed, Kerstin Denecke, Mowafa Househ, Faisal Farooq, and Javaid Sheikh. 2023. Wearable artificial intelligence for anxiety and depression: scoping review.Journal of Medical Internet Research25 (2023), e42672

2023
[2]

Samuel L Battalio, David E Conroy, Walter Dempsey, Peng Liao, Marianne Menictas, Susan Murphy, Inbal Nahum-Shani, Tianchen Qian, Santosh Kumar, and Bonnie Spring. 2021. Sense2Stop: a micro-randomized trial using wearable sensors to optimize a just-in-time-adaptive stress management intervention for smoking relapse prevention.Contemporary Clinical Trials10...

2021
[3]

Virginia Braun and Victoria Clarke. 2006. Using thematic analysis in psychology.Qualitative research in psychology3, 2 (2006), 77–101

2006
[4]

Francesca Calabrese, Raffaella Molteni, Giorgio Racagni, and Marco A Riva. 2009. Neuronal plasticity: a link between stress and mood disorders. Psychoneuroendocrinology34 (2009), S208–S216

2009
[5]

Varun Chandra, Ankit Priyarup, and Divyashikha Sethia. 2021. Comparative study of physiological signals from Empatica E4 wristband for stress classification. InAdvances in Computing and Data Sciences: 5th International Conference, ICACDS 2021, Nashik, India, April 23–24, 2021, Revised Selected Papers, Part II 5. Springer, India, 218–229

2021
[6]

Kemeng Chen, Wolfgang Fink, Janet Roveda, Richard D Lane, John Allen, and Johnny Vanuk. 2015. Wearable sensor based stress management using integrated respiratory and ECG waveforms. In2015 IEEE 12th International Conference on Wearable and Implantable Body Sensor Networks (BSN). IEEE, Cambridge, MA, USA, 1–6. doi:10.1109/BSN.2015.7299369

work page doi:10.1109/bsn.2015.7299369 2015
[7]

Akshat Choube, Ha Le, Jiachen Li, Kaixin Ji, Vedant Das Swain, and Varun Mishra. 2025. GLOSS: Group of LLMs for Open-ended Sensemaking of Passive Sensing Data for Health and Wellbeing.Proc. ACM Interact. Mob. Wearable Ubiquitous Technol.9, 3, Article 76 (Sept. 2025), 32 pages. doi:10.1145/3749474

work page doi:10.1145/3749474 2025
[8]

Poorvesh Dongre. 2024. Physiology-Driven Empathic Large Language Models (EmLLMs) for Mental Health Support. InExtended Abstracts of the CHI Conference on Human Factors in Computing Systems(Honolulu, HI, USA)(CHI EA ’24). Association for Computing Machinery, New York, NY, USA, Article 452, 5 pages. doi:10.1145/3613905.3651132

work page doi:10.1145/3613905.3651132 2024
[9]

Poorvesh Dongre, Majid Behravan, and Denis Gračanin. 2025. Empathic Extended Reality in the Era of Generative AI.Empathic Computing1, 2 (2025), 202509–202509

2025
[10]

Angela Liegey Dougall and Andrew Baum. 2001. Stress, health, and illness.Handbook of health psychology2 (2001), 53–78

2001
[11]

Steven R Erickson, Sally Guthrie, Michelle VanEtten-Lee, Joseph Himle, Jody Hoffman, Susana F Santos, Amy S Janeck, Kara Zivin, and James L Abelson. 2009. Severity of anxiety and work-related outcomes of patients with anxiety disorders.Depression and Anxiety26, 12 (2009), 1165–1171

2009
[12]

Cathy Mengying Fang, Valdemar Danry, Nathan Whitmore, Andria Bao, Andrew Hutchison, Cayden Pierce, and Pattie Maes. 2024. Physiollm: Supporting personalized health insights with wearables and large language models. In2024 IEEE EMBS International Conference on Biomedical and Health Informatics (BHI). IEEE, New York, 1–8

2024
[13]

Martin Gjoreski, Mitja Luštrek, Matjaž Gams, and Hristijan Gjoreski. 2017. Monitoring stress with a wrist device using context.Journal of biomedical informatics73 (2017), 159–170

2017
[14]

Constance Hammen. 2005. Stress and depression.Annu. Rev. Clin. Psychol.1, 1 (2005), 293–319

2005
[15]

Michael V Heinz, Daniel M Mackin, Brianna M Trudeau, Sukanya Bhattacharya, Yinzhou Wang, Haley A Banta, Abi D Jewett, Abigail J Salzhauer, Tess Z Griffin, and Nicholas C Jacobson. 2025. Randomized trial of a generative AI chatbot for mental health treatment.Nejm Ai2, 4 (2025), AIoa2400802

2025
[16]

Rafal Kocielnik, Natalia Sidorova, Fabrizio Maria Maggi, Martin Ouwerkerk, and Joyce HDM Westerink. 2013. Smart technologies for long-term stress monitoring at work. Inproceedings of the 26th IEEE international symposium on computer-based medical systems. IEEE, Porto, Portugal, 53–58

2013
[17]

Tatiana Kuzmowycz. 2023. Introducing stress monitor: A new way to monitor and manage stress. https://www.whoop.com/thelocker/introducing- stress-monitor-a-new-way-to-monitor-manage-stress/

2023
[18]

Psy-llm: Scaling up global mental health psychological services with ai-based large language models, September 2023

Tin Lai, Yukun Shi, Zicong Du, Jiajie Wu, Ken Fu, Yichao Dou, and Ziqi Wang. 2023.Psy-LLM: Scaling up global mental health psychological services with ai-based large language models. arXiv 2307.11991 [cs.CL]. arXiv

work page arXiv 2023
[19]

Chatcounselor: A large language models for mental health support.arXiv preprint arXiv:2309.15461, 2023

June M. Liu, Donghao Li, He Cao, Tianhe Ren, Zeyi Liao, and Jiamin Wu. 2023.ChatCounselor: A Large Language Models for Mental Health Support. arXiv 2309.15461 [cs.CL]. arXiv

work page arXiv 2023
[20]

Subigya Nepal, Arvind Pillai, William Campbell, Talie Massachi, Michael V Heinz, Ashmita Kunwar, Eunsol Soul Choi, Xuhai Xu, Joanna Kuc, Jeremy F Huckins, et al. 2024. MindScape Study: Integrating LLM and Behavioral Sensing for Personalized AI-Driven Journaling Experiences. Proceedings of the ACM on interactive, mobile, wearable and ubiquitous technologie...

2024
[21]

Sameer Neupane, Poorvesh Dongre, Denis Gracanin, and Santosh Kumar. 2025. Wearable Meets LLM for Stress Management: A Duoethnographic Study Integrating Wearable-Triggered Stressors and LLM Chatbots for Personalized Interventions. InProceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ’25). Association f...

work page doi:10.1145/3706599.3720197 2025
[22]

Almeida, and Santosh Kumar

Sameer Neupane, Mithun Saha, Nasir Ali, Timothy Hnat, Shahin Alan Samiei, Anandatirtha Nandugudi, David M. Almeida, and Santosh Kumar
[23]

InProceedings of the 2024 CHI Conference on Human Factors in Computing Systems(Honolulu, HI, USA)(CHI ’24)

Momentary Stressor Logging and Reflective Visualizations: Implications for Stress Management with Wearables. InProceedings of the 2024 CHI Conference on Human Factors in Computing Systems(Honolulu, HI, USA)(CHI ’24). Association for Computing Machinery, New York, NY, USA, Article 809, 19 pages. doi:10.1145/3613904.3642662 Manuscript submitted to ACM 8 Don...

work page doi:10.1145/3613904.3642662 2024
[24]

Simon Ollander, Christelle Godin, Aurélie Campagne, and Sylvie Charbonnier. 2016. A comparison of wearable and stationary sensors for stress detection. In2016 IEEE International Conference on Systems, Man, and Cybernetics (SMC). IEEE, Budapest, Hungary, 004362–004366

2016
[25]

Pedro Sanches, Kristina Höök, Elsa Vaara, Claus Weymann, Markus Bylund, Pedro Ferreira, Nathalie Peira, and Marie Sjölinder. 2010. Mind the body! designing a mobile stress management application encouraging personal reflection. InProceedings of the 8th ACM Conference on Designing Interactive Systems(Aarhus, Denmark)(DIS ’10). Association for Computing Mac...

work page doi:10.1145/1858171.1858182 2010
[26]

Epstein, Kenzie L

Hillol Sarker, Matthew Tyburski, Md Mahbubur Rahman, Karen Hovsepian, Moushumi Sharmin, David H. Epstein, Kenzie L. Preston, C. Debra Furr-Holden, Adam Milam, Inbal Nahum-Shani, Mustafa al’Absi, and Santosh Kumar. 2016. Finding Significant Stress Episodes in a Discontinuous Time Series of Rapidly Varying Mobile Sensor Data. InProceedings of the 2016 CHI C...

work page doi:10.1145/2858036.2858218 2016
[27]

Rutvik V Shah, Gillian Grennan, Mariam Zafar-Khan, Fahad Alim, Sujit Dey, Dhakshin Ramanathan, and Jyoti Mishra. 2021. Personalized machine learning of depressed mood using wearables.Translational psychiatry11, 1 (2021), 338

2021
[28]

Gayle Beck, Sudip Vhaduri, Kenzie Preston, and Santosh Kumar

Moushumi Sharmin, Andrew Raij, David Epstien, Inbal Nahum-Shani, J. Gayle Beck, Sudip Vhaduri, Kenzie Preston, and Santosh Kumar. 2015. Visualization of time-series sensor data to inform the design of just-in-time adaptive stress interventions. InProceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing(Osaka, Japan)...

work page doi:10.1145/2750858.2807537 2015
[29]

Pragya Singh, Ankush Gupta, Mohan Kumar, and Pushpendra Singh. 2025. AnnoSense: A Framework for Physiological Emotion Data Collection in Everyday Settings for AI.Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies9, 3 (2025), 1–47

2025
[30]

Eric N Smith, Erik Santoro, Neema Moraveji, Michael Susi, and Alia J Crum. 2020. Integrating wearables in stress management interventions: Promising evidence from a randomized trial.International Journal of Stress Management27, 2 (2020), 172–182

2020
[31]

Nicholas A Troop, Alison Holbrey, and Janet L Treasure. 1998. Stress, coping, and crisis support in eating disorders.International Journal of eating disorders24, 2 (1998), 157–166

1998
[32]

Rui Wang, Weichen Wang, Alex DaSilva, Jeremy F Huckins, William M Kelley, Todd F Heatherton, and Andrew T Campbell. 2018. Tracking depression dynamics in college students using mobile phone and wearable sensing.Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies2, 1 (2018), 1–26

2018
[34]

Adler, Joey Castillo, Tanzeem Choudhury, and Fei Wang

Xingbo Wang, Janessa Griffith, Daniel A. Adler, Joey Castillo, Tanzeem Choudhury, and Fei Wang. 2025. Exploring Personalized Health Support through Data-Driven, Theory-Guided LLMs: A Case Study in Sleep Health. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems (CHI ’25). Association for Computing Machinery, New York, NY, USA, ...

work page doi:10.1145/3706598.3713852 2025
[35]

Johnny Chun Yiu Wong, Jun Wang, Eugene Yujun Fu, Hong Va Leong, and Grace Ngai. 2020. Activity Recognition and Stress Detection via Wristband. InProceedings of the 17th International Conference on Advances in Mobile Computing & Multimedia(Munich, Germany)(MoMM2019). Association for Computing Machinery, New York, NY, USA, 102–106. doi:10.1145/3365921.33659...

work page doi:10.1145/3365921.3365950 2020