An Interactive LLM-Based Simulator for Dementia-Related Activities of Daily Living
arXiv:2603.29856v1 Announce Type: new Abstract: Effective dementia caregiving requires training and adaptive communication, but assistive AI and robotics are constrained by a lack of context-rich, privacy-sensitive data on how people living with Alzheimer's disease and related dementias (ADRD) behave during activities of daily living (ADLs). We introduce a web-based simulator that uses a large language model (gpt-5-mini) to generate multi-turn, severity- and care-setting-conditioned patient behaviors during ADL assistance, pairing utterances with lightweight behavioral cues (in parentheses). Users set dementia severity, care setting (and time in setting), and ADL; after each patient turn they rate realism (1-5) with optional critique, then respond as the caregiver via free text or by selec
View PDF HTML (experimental)
Abstract:Effective dementia caregiving requires training and adaptive communication, but assistive AI and robotics are constrained by a lack of context-rich, privacy-sensitive data on how people living with Alzheimer's disease and related dementias (ADRD) behave during activities of daily living (ADLs). We introduce a web-based simulator that uses a large language model (gpt-5-mini) to generate multi-turn, severity- and care-setting-conditioned patient behaviors during ADL assistance, pairing utterances with lightweight behavioral cues (in parentheses). Users set dementia severity, care setting (and time in setting), and ADL; after each patient turn they rate realism (1-5) with optional critique, then respond as the caregiver via free text or by selecting/editing one of four strategy-scaffolded suggestions (Recognition, Negotiation, Facilitation, Validation). We ran an online formative expert-in-the-loop study (14 dementia-care experts, 18 sessions, 112 rated turns). Simulated behavior was judged moderately to highly plausible, with a typical session length of six turns. Experts wrote custom replies for 54.5 percent of turns; Recognition and Facilitation were the most-used suggested strategies. Thematic analysis of critiques produced a six-category failure-mode taxonomy, revealing recurring breakdowns in ADL grounding and care-setting consistency and guiding prompt/workflow refinements. The simulator and logged interactions enable an evidence-driven refinement loop toward validated patient-caregiver co-simulation and support data collection, caregiver training, and assistive AI and robot policy development.
Subjects:
Human-Computer Interaction (cs.HC); Graphics (cs.GR); Robotics (cs.RO)
Cite as: arXiv:2603.29856 [cs.HC]
(or arXiv:2603.29856v1 [cs.HC] for this version)
https://doi.org/10.48550/arXiv.2603.29856
arXiv-issued DOI via DataCite (pending registration)
Submission history
From: Kruthika Gangaraju [view email] [v1] Fri, 6 Mar 2026 03:15:17 UTC (15,675 KB)
Sign in to highlight and annotate this article
Conversation starters
Daily AI Digest
Get the top 5 AI stories delivered to your inbox every morning.
More about
modellanguage modeltraining
Stop Putting Everything in .env — Runtime Config for Django
<p>Hello all! 👋</p> <p>TL;DR I created a runtime configuration Django app called <a href="https://github.com/krishnamodepalli/django-sysconfig" rel="noopener noreferrer"><code>django-sysconfig</code></a>.</p> <p><a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fgithub.com%2Fkrishnamodepalli%2Fdjango-sysconfig%2Freleases%2Fdownload%2Fv0.3.0%2Fdjango-sysconfig.gif" class="article-body-image-wrapper"><img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fgithub.com%2Fkrishnamodepalli%2Fdjango-sysconfig%2Freleases%2Fdownload%2Fv0.3.0%2Fdjango-sysconfig.gif" alt="Django Sysconfig app usage" width="760" height="381"></a></p> <p>Here's the problem that
Why Your Frontend Is Actually a State Machine (And AI Makes It More Complicated)
<p>When most developers think about frontend development, they imagine components, UI elements, and responsive layouts. </p> <p>What we rarely acknowledge is that <strong>every modern frontend is fundamentally a state machine</strong> — a system where the state drives the UI, and events drive state changes. </p> <p>Add AI-driven features, predictive models, or automated agents, and your “simple” frontend suddenly becomes a complex web of interacting states, transitions, and events.</p> <h2> Frontends Are State Machines </h2> <p>Consider what a state machine is:</p> <ul> <li> <strong>States</strong> represent the current status of your system.</li> <li> <strong>Transitions</strong> are triggered by events (user clicks, API responses, timers, etc.).</li> <li> <strong>Actions</strong> happen
Say Goodbye to JavaScript's Date Object: The Temporal API is Here to Save Your Sanity
<h2> Finally, a date-time API that actually makes sense and won't make you want to pull your hair out </h2> <p>Ever found yourself staring at JavaScript's Date object thinking, <em>"There has to be a better way"</em>? You're not alone. How many times have you written <code>new Date().getMonth() + 1</code> just to get the actual month number, or struggled with timezone hell that made your perfectly working local code break in production?</p> <p><strong>Quick question for you:</strong> What's the most frustrating Date-related bug you've encountered? Drop it in the comments—I bet others have been there too! </p> <p>If you've ever dealt with JavaScript dates, you know the pain. But here's some great news: <strong>the Temporal API is coming to replace the Date object</strong>, and it's going to
Knowledge Map
Connected Articles — Knowledge Graph
This article is connected to other articles through shared AI topics and tags.
More in Models
5 Things I Believed About SEO That AI Search Proved Completely Wrong
<p>`<br> I have been writing technical content for about three years. In that time I developed a fairly confident set of beliefs about what makes content rank well. Keyword research, internal linking, page speed, backlink acquisition, content length. I followed the playbook. My traffic grew. I assumed I understood what I was doing.</p> <p>Then, about six weeks ago, I started seriously looking at whether my content was being cited by ChatGPT, Perplexity AI, and Google Gemini. Not just vaguely appearing somewhere, but actually being used as a source when someone asked a question my posts directly answer.</p> <p>The results were uncomfortable. And almost everything I had been confident about turned out to be either wrong or only partially true when applied to how AI search actually works.</p>
I asked ChatGPT if AI can be empathetic: The answer surprised me - PharmaLive
<a href="https://news.google.com/rss/articles/CBMilgFBVV95cUxOUXFNejdyOHdTbm5SdzBNbTVGcTg1ZURkUENfcGtqTVA4dTJaemJQZGV3M1hGR1ZFUV9tMmJjbGZVaVB2czVwaTRrZWNFQkNrbGhiQkNRWjEwY2F4LUtLSi12Y1lMbDlEWERNY2RkakVaQ3NKMVZhWS1CZjN1UTRzQVpmR0FWNFB4YzlwTURkUGwtSF92MkE?oc=5" target="_blank">I asked ChatGPT if AI can be empathetic: The answer surprised me</a> <font color="#6f6f6f">PharmaLive</font>
Silicon Valley Has Stopped Talking Politics—Except for This Google Executive - WSJ
<a href="https://news.google.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?oc=5" target="_blank">Silicon Valley Has Stopped Talking Politics—Except for This Google Executive</a> <font color="#6f6f6f">WSJ</font>
Shutterstock Launches Licensed Content App In Chatgpt, Bringing Commercial-Ready Assets Into Ai-Native Workflows - TradingView
<a href="https://news.google.com/rss/articles/CBMijgJBVV95cUxOc0ZvNHktZE5sRkFoY3Q0eWlVOGFiT05jbHdOZTFjUWI5Q0NWTjNCMFhfSXItWHpNOUoxQnlENnkxSTlrRGdTUDljS05FRk4yT0JXUExUYXExZ2duRXh2TmV3SzFSZXhDXzVtOVdzaEZNZ1MxaXozM1ZaNFJwa2RRdWhJRE5pS3dwN2k1dUsxanZMQmZBdFVWRGRPRm5RSEVLQVBwQXJab2M5eEg5QUZLUmxpeDdNZVdvejVaWDRGRU1nMHVJVG1Qc0RDY2hWRHNIbkpuN3ZEb0NSTkVqZFdPTlpRejNSYTZXUWQtaUcxSG4xeGV4bVZDME9UOHA0RTRPTF8tZXpEMlZ6XzExcmc?oc=5" target="_blank">Shutterstock Launches Licensed Content App In Chatgpt, Bringing Commercial-Ready Assets Into Ai-Native Workflows</a> <font color="#6f6f6f">TradingView</font>
Discussion
Sign in to join the discussion
No comments yet — be the first to share your thoughts!