คลายใจ : แพลตฟอร์มผู้ช่วยรับฟังความคิดและคลายความเครียดผ่านการสะท้อนแบบไม่รุกล้ำ

Abstract

Thailand is currently facing critically elevated levels of occupational stress and mental health challenges among its workforce, driven by risk factors such as prolonged working hours and digital fatigue, which directly impair work performance while access to support systems and emotional outlet remain limited. This engineering project therefore designed and develop an AI Chatbot platform, “CalmJai”, integrating large language model technology with a psychologically structured conversational flamework based on the GROW Model – a format emphasizing constructive listening and reflective thinking. Physiological data were collected via vascular pulse wave signals using a Xiaomi Band 6 smartwatch and direct electrocardiographic measurement through contact electrodes using a Polar H10 chest strap, alongside self-assessed stress and burnout scales. Two system variants were evaluated with separate participant groups: a fully structured format (Version 1, n = 15) and a flexible conversational format (Version 2, n = 12). Result indicated that Version 1 demonstrated a significant trend toward parasympathetic nervous system activation and relaxation response in the small sample, with notable increases in heart rate variability indices – both RMSSD and SDNN. Regarding subjective perception, both groups showed a tendency toward reduced stress scores following system use; Version 2 exhibited a large effect size in stress score reduction (r = 0.533, p = 0.069) and a greater proportional decrease in stress scores compared to Version 1 (-0.069 vs -0.038), as well as a stronger tendency to attenuate the accumulation of burnout symptoms (adjusted mean difference = −0.008). Furthermore, the majority of users reported high satisfaction with Version 1, particularly on the dimensions of empathy and credibility as assessed by the PETS Framework. Expert evaluation by three specialists found that the overall mean scores across all dimensions of the system's counseling performance, based on the GROW Model principles, were rated at an excellent level. This engineering project presents a preliminary approach to the practical application of artificial intelligence technology in supporting and alleviating emotional distress in a concrete and measurable manner