arXiv subjects

Junhui Qian

Publications and source records attributed to Junhui Qian.

3 recordsLinked to original sources

arXiv2025

ChatCLIDS: Simulating Persuasive AI Dialogues to Promote Closed-Loop Insulin Adoption in Type 1 Diabetes Care

Real-world adoption of closed-loop insulin delivery systems (CLIDS) in type 1 diabetes remains low, driven not by technical failure, but by diverse behavioral, psychosocial, and social barriers. We introduce ChatCLIDS, the first benchmark to rigorously evaluate LLM-driven persuasive dialogue for health behavior change. Our framework features a library of expert-validated virtual patients, each with clinically grounded, heterogeneous profiles and realistic adoption barriers, and simulates multi-turn interactions with nurse agents equipped with a diverse set of evidence-based persuasive strategies. ChatCLIDS uniquely supports longitudinal counseling and adversarial social influence scenarios, enabling robust, multi-dimensional evaluation. Our findings reveal that while larger and more reflective LLMs adapt strategies over time, all models struggle to overcome resistance, especially under realistic social pressure. These results highlight critical limitations of current LLMs for behavior change, and offer a high-fidelity, scalable testbed for advancing trustworthy persuasive AI in healthcare and beyond.

cs.AI↗

arXiv2025

ChatThero: An LLM-Supported Chatbot for Behavior Change and Therapeutic Support in Addiction Recovery

Substance use disorders (SUDs) affect millions of people, and relapses are common, requiring multi-session treatments. Access to care is limited, which contributes to the challenge of recovery support. We present \textbf{ChatThero}, an innovative low-cost, multi-session, stressor-aware, and memory-persistent autonomous \emph{language agent} designed to facilitate long-term behavior change and therapeutic support in addiction recovery. Unlike existing work that mostly finetuned large language models (LLMs) on patient-therapist conversation data, ChatThero was trained in a multi-agent simulated environment that mirrors real therapy. We created anonymized patient profiles from recovery communities (e.g., Reddit). We classify patients as \texttt{easy}, \texttt{medium}, and \texttt{difficult}, three scales representing their resistance to recovery. We created an external environment by introducing stressors (e.g., social determinants of health) to simulate real-world situations. We dynamically inject clinically-grounded therapeutic strategies (motivational interview and cognitive behavioral therapy). Our evaluation, conducted by both human (blinded clinicians) and LLM-as-Judge, shows that ChatThero is superior in empathy and clinical relevance. We show that stressor simulation improves robustness of ChatThero. Explicit stressors increase relapse-like setbacks, matching real-world patterns. We evaluate ChatThero with behavioral change metrics. On a 1--5 scale, ChatThero raises \texttt{motivation} by $+1.71$ points (from $2.39$ to $4.10$) and \texttt{confidence} by $+1.67$ points (from $1.52$ to $3.19$), substantially outperforming GPT-5. On \texttt{difficult} patients, ChatThero reaches the success milestone with $26\%$ fewer turns than GPT-5.

cs.AI↗

arXiv2025

Analysis of Distributional Dynamics for Repeated Cross-Sectional and Intra-Period Observations

This paper introduces a novel approach to investigate the dynamics of state distributions, which accommodate both cross-sectional distributions of repeated panels and intra-period distributions of a time series observed at high frequency. In our approach, densities of the state distributions are regarded as functional elements in a Hilbert space, and are assumed to follow a functional autoregressive model. We propose an estimator for the autoregressive operator, establish its consistency, and provide tools and asymptotics to analyze the forecast of state density and the moment dynamics of state distributions. We apply our methodology to study the time series of distributions of the GBP/USD exchange rate intra-month returns and the time series of cross-sectional distributions of the NYSE stocks monthly returns. Finally, we conduct simulations to evaluate the density forecasts based on our model.

econ.EM↗