Mobile health applications show promise for scalable physical activity promotion but are often insufficiently personalized. In contrast, health coaching offers highly personalized support but can be prohibitively expensive and inaccessible. This study draws inspiration from health coaching to explore how large language models (LLMs) might address personalization challenges in mobile health. We conduct formative interviews with 12 health professionals and 10 potential coaching recipients to develop design principles for an LLM-based health coach. We then built GPTCoach, a chatbot that implements the onboarding conversation from an evidence-based coaching program, uses conversational strategies from motivational interviewing, and incorporates wearable data to create personalized physical activity plans. In a lab study with 16 participants using three months of historical data, we find promising evidence that GPTCoach gathers rich qualitative information to offer personalized support, with users feeling comfortable sharing concerns. We conclude with implications for future research on LLM-based physical activity support.
https://dl.acm.org/doi/10.1145/3706598.3713819
LLM-based agents improve upon standalone LLMs, which are optimized for immediate intent-satisfaction, by allowing the pursuit of more extended objectives, such as helping users over the long term. To do so, LLM-based agents need to reason before responding. For complex tasks like personalized coaching, this reasoning can be informed by adding relevant information at key moments, shifting it in the desired direction. However, the pursuit of objectives beyond interaction quality may compromise this very quality. Moreover, as the depth and informativeness of reasoning increase, so do the number of tokens required, leading to higher latency and cost. This study investigates how an LLM-based coaching agent can adjust its reasoning depth using a discrepancy mechanism that signals how much reasoning effort to allocate based on how well the objective is being met. Our discrepancy-based mechanism constrains reasoning to better align with alternative objectives, reducing cost roughly tenfold while minimally impacting interaction quality.
https://dl.acm.org/doi/10.1145/3706598.3713606
As esports grows into a multi-million dollar industry of professional players and competitions, so too grows the interest in and need for professional coaching. Accordingly, there are increased demands and attempts to support and improve coaching for esports. A more comprehensive, granular understanding of the esports coaching process would provide a valuable foundation to inform opportunities to advance the domain via HCI theories and practices. However, in-depth studies of coaching practice, from the lens of HCI, are far less common in existing literature. In this paper, we take the first steps to provide such a foundation through an observation study conducted at an elite, award-winning League of Legends training academy. By analyzing 112 hours of dialogue and footage from coaching sessions, we identify pertinent activities and events that occur within the coaching process, which enable us to consider how esports coaching can be improved via theory, practice, and technology from HCI.
https://dl.acm.org/doi/10.1145/3706598.3713141
Co-viewing, traditionally defined as watching content together in the same physical space, enhances emotional connections through shared experiences. With the rise of remote viewing during the COVID-19 pandemic, existing solutions, such as second-screen platforms and rule-based AI companions, struggle to facilitate meaningful social interactions. This study explores the potential of Large Language Models, which offer human-like interactions and personalization. Our formative study with ten participants revealed the importance of managing arousal levels, highlighting the need to balance between high- and low-arousal levels across different viewing contexts. Based on these insights, we developed `BleacherBot', a sports co-viewing agent with distinct interaction styles that vary in arousal levels. Our main study with 27 participants demonstrated that matching users' preferred arousal levels with the agent's interaction style significantly enhanced their engagement and overall enjoyment. We propose design guidelines for AI co-viewing agents that consider their role as complements to human social interactions.
https://dl.acm.org/doi/10.1145/3706598.3714178
We present a methodology for designing an AI feedback system aimed at assisting basketball beginners in refining their shooting techniques during independent practice sessions. Mastering shooting mechanics requires consistent, precise repetition, which traditionally depends on coaching feedback and the breakdown of movements into steps during the early stages. However, due to limited coaching resources, this guidance is often unavailable, leading to ineffective and even detrimental motor learning. To bridge this gap, we propose a Standard Operating Procedure (SOP) framework grounded in expert human knowledge, or knowledge-based SOP, which allows our AI-driven system to verify and guide players' movements in real-time. Through a formative study involving interviews with 13 coaches and players, we identified key challenges faced by beginners, such as uncertainty in movement correctness and lack of guidance during unsupervised practice. Our AI system addresses these issues by providing immediate, actionable feedback using SOP tailored to individual players. In a study with 28 participants, we confirmed that our system improves shooting form, increases confidence in adjustments, and enhances self-awareness during practice. This work highlights the potential of integrating coaching expertise with AI to empower athletes with more effective tools for self-directed practice.
https://dl.acm.org/doi/10.1145/3706598.3713324
A rapidly emerging research community at the intersection of sport and human-computer interaction (SportsHCI) explores how technology can support physically active humans, such as athletes. At highly competitive levels, coaching staff play a central role in the athlete experience by using data to enhance performance, reduce injuries, and foster team success. However, little is known about the practices and needs of these coaching staff. We conducted five focus groups with 17 collegiate coaching staff across three women’s teams and two men’s teams at an elite U.S. university. Our findings show that coaching staff selectively use data with the goal of balancing performance goals, athlete emotional well-being, and privacy. This paper contributes design recommendations to support coaching staff in operating across the data life cycle through gathering, sharing, deciding, acting, and assessing data as they aim to support team success and foster the well-being of student-athletes.
https://dl.acm.org/doi/10.1145/3706598.3714026