MindTS-MMD: A Chinese Multimodal Dataset for Emotion Expression Analysis in Children with Tourette Syndrome

Multimodal resources for emotion expression analysis in pediatric clinical populations remain limited, particularly for children with Tourette syndrome (TS). MindTS-MMD was developed as a Chinese multimodal dataset to support computational research on emotional expression and tic-related behavior in this population. The dataset contains 10,034 instance-level samples from 60 children with TS aged 6–12 years, collected through semi-structured emotion-elicitation tasks. Each sample includes available symbolic representations from three modalities—visual, acoustic, and semantic—and is linked to a corresponding annotation record. The annotations cover seven categories: anxious, calm, focused, irritable, relaxed, shy, and tense, together with child-reported and experimenter-observed valence–arousal ratings and tic occurrence, anatomical location, and frequency when observable. Annotation reliability, signal quality, audiovisual synchronization, facial tracking, modality completeness, and tic–emotion co-occurrence were evaluated. Unimodal and multimodal baselines are provided to illustrate the computational usability of the released representations.