Recent advances in large language models (LLM) offered human-like capabilities for comprehending emotion and mental states. Prior studies explored diverse prompt engineering techniques for improving classification performance, but there is a lack of analysis of prompt design space and the impact of each component. To bridge this gap, we conduct a qualitative thematic analysis of existing prompts for emotion and mental health classification tasks to define the key components for prompt design space. We then evaluate the impact of major prompt components, such as persona and task instruction, on classification performance by using four LLM models and five datasets. Modular prompt design offers new insights into examining performance variability as well as promoting transparency and reproducibility in LLM-based tasks within health and well-being intervention systems.
https://dl.acm.org/doi/10.1145/3706598.3713888
Prior work has documented various ways that teens use social media to regulate their emotions. However, little is known about what these processes look like on a moment-by-moment basis. We conducted a diary study to investigate how teens (N=57, Mage = 16.3 years) used Instagram to regulate their emotions. We identified three kinds of emotionally-salient drivers that brought teens to Instagram and two types of behaviors that impacted their emotional experiences on the platform. Teens described going to Instagram to escape, to engage, and to manage the demands of the platform. Once on Instagram, their primary behaviors consisted of mindless diversions and deliberate acts. Although teens reported many positive emotional responses, the variety, unpredictability, and habitual nature of their experiences revealed Instagram to be an unreliable tool for emotion regulation (ER). We present a model of teens’ ER processes on Instagram and offer design considerations for supporting adolescent emotion regulation.
https://dl.acm.org/doi/10.1145/3706598.3713844
Digital food journaling can help support reflection and improvement of wellbeing relating to eating habits. However, it is often viewed as burdensome, and abandoned before gaining benefits. Advances in conversational user interfaces (CUIs) have the potential to support people journaling in a natural and interactive manner, but we lack understanding of how people would ideally prefer to use CUIs when journaling. We conducted 33 co-design sessions with 18 participants to ideate CUI interactions supportive of their health goals and in everyday situations. Our findings reveal that participants expect CUIs to be adaptive by learning goals and personal references, and support depth in detail and goal alignment while respecting situational constraints and intent. While participants expressed concern around navigating long-term data solely through conversations, they envisioned that CUIs could provide empathetic, non-judgmental feedback. We discuss opportunities for CUIs to support empathetic food journaling and accountability while following guardrails for delegated tasks.
https://dl.acm.org/doi/10.1145/3706598.3713875
Sleep diaries are essential self-reporting tools for understanding children's sleep patterns, but maintaining sustained engagement and high-quality self-reporting remains challenging. While voice input has been explored in child-computer interaction research as a method to improve engagement, limited evidence exists regarding its effectiveness in supporting sustained self-reporting over time. To address this gap, we conducted a five-day field study with 20 children aged seven to twelve, using a multimodal sleep diary that integrated both voice and text input modalities. Our findings reveal that voice input significantly supports younger children in maintaining engagement over five days, though their response quality remains lower than that of older children. Two distinct response quality patterns over time also emphasize the importance of accounting for individual differences in task performance. Furthermore, input modality preferences varied by age: older children consistently favored text input, while younger children generally preferred voice input over time. These results highlight the potential of incorporating voice input into text-based sleep diaries to better accommodate the diverse needs of children, enhancing both sustained engagement and response quality. Future studies with longer observation periods are needed to validate and extend these findings.
https://dl.acm.org/doi/10.1145/3706598.3713425
𝜇EMAs allow participants to answer a short survey quickly with a tap on a smartwatch screen or a brief speech input. The short interaction time and low cognitive burden enable researchers to collect self-reports at high frequency (once every 5-15 minutes) while maintaining participant engagement. Systems with single input modality, however, may carry different contextual biases that could affect compliance. We combined two input modalities to create a multimodal-𝜇EMA system, allowing participants to choose between speech or touch input to self-report. To investigate system usability, we conducted a 7-day field study where we asked 20 participants to label their posture and/or physical activity once every five minutes throughout their waking day. Despite the intense prompting interval, participants responded to 72.4% of the prompts. We found participants gravitated towards different modalities based on personal preferences and contextual states, highlighting the need to consider these factors when designing context-aware multimodal 𝜇EMA systems.
https://dl.acm.org/doi/10.1145/3706598.3714086