Scientific foundations for digital mental health. Our work in multimodal sensing, clinical psychology, and adaptive assessment powers MindCanvas products.
MindCanvas research integrates cutting-edge AI with clinical psychology to deliver safe, effective mental health interventions at scale. From UESTC MediaLab's decade of work in education and youth development, we have built a rigorous technical foundation validated across millions of users.
Our research focuses on three pillars: multimodal emotion understanding, clinically constrained dialogue systems, and adaptive psychological assessment. These enable the full-scenario ecosystem—from smart screening hubs to embodied robots and digital companions.
We analyze facial expressions, acoustic features, and semantic content in sync to capture emotional cues that go beyond words. Traditional questionnaires can be easily faked or invalid; multimodal sensing detects subtle fluctuations in voice, expression, and physiology—providing a more accurate basis for screening and intervention.
All dialogue and intervention flows are bound by clinical psychology principles. Our systems are designed to be safe, controllable, and traceable—reducing risk while maintaining human warmth. Every interaction can be logged and reviewed by qualified professionals.
We move beyond rigid pencil-and-paper questionnaires. AI-guided interactive assessment adapts questions in real time based on responses, improving validity and engagement. This enables more precise screening and longitudinal tracking.
Our research contributes to academic and industry practice in digital mental health, embodied AI, and educational technology.
Selected publications and technical reports will be listed here.
Our research is conducted in collaboration with UESTC (University of Electronic Science and Technology of China), drawing on expertise in AI, embodied robotics, and psychology.