Towards Understanding Emotions for Engaged Mental Health Conversations
Kellie Yu Hui Sim, Kohleen Tijing Fortuno, Kenny Tsu Wei Choo
TL;DR
The paper proposes an unobtrusive emotion-detection system for text-based mental health conversations by fusing keystroke dynamics (KD) with sentiment analysis. Using a within-subjects study (n=31), it evaluates dimensional and categorical emotion representations with manual annotations and GPT-4 Turbo labeling, plus KD and content features fed into various classifiers, including a fusion of KD and text features. Results show that while categorical emotion detection generally outperforms dimensional, fusion improves detection for anger, surprise, and fear, with varying gains across other emotions; real-time KD-based inference demonstrates feasibility. The work discusses implications for emotion-aware crisis-care platforms, highlights privacy and human-in-the-loop considerations, and outlines future directions to enhance predictive performance and integrate additional physiological signals.
Abstract
Providing timely support and intervention is crucial in mental health settings. As the need to engage youth comfortable with texting increases, mental health providers are exploring and adopting text-based media such as chatbots, community-based forums, online therapies with licensed professionals, and helplines operated by trained responders. To support these text-based media for mental health--particularly for crisis care--we are developing a system to perform passive emotion-sensing using a combination of keystroke dynamics and sentiment analysis. Our early studies of this system posit that the analysis of short text messages and keyboard typing patterns can provide emotion information that may be used to support both clients and responders. We use our preliminary findings to discuss the way forward for applying AI to support mental health providers in providing better care.