Yoga-Based Motor Assessment Tool (YOGA-MAT)

This dataset accompanies the article titled “Development, Validation and Feasibility of a Yoga-Based Motor Assessment Tool (Y-MAT) for Children with ADHD: A Cross-Sectional Pilot Study”. It contains de-identified pilot-phase data generated during the development and validation of the Y-MAT, a structured performance-based instrument designed to assess motor and self-regulatory domains in children aged 8 to 15 years, with a specific focus on Attention-Deficit/Hyperactivity Disorder (ADHD). Y-MAT integrates culturally grounded yoga tasks with domain-specific motor scoring to capture both motor execution and regulatory control. The tool assesses seven domains: static and dynamic balance, motor sequencing and rhythm, bilateral and cross-limb coordination, impulse control and movement regulation, body awareness and spatial orientation, fine motor dexterity, and breath control. The dataset includes demographic variables, domain-wise expert ratings using a standardized 0 to 3 scoring framework, AI-derived kinematic metrics, and face-blurred task performance videos with pose-estimation overlays and synchronized scoring annotations This dataset supports reproducibility, secondary analyses of motor profiling in ADHD, and further development of AI-based movement quantification approaches in pediatric neurodevelopmental research. The Y-MAT was developed to address limitations in conventional motor assessment tools that emphasize motor execution but do not adequately capture regulatory aspects such as impulse control, breath awareness, and sequencing deficits. The tool was piloted in a tertiary neuropsychiatric setting in India. Children completed structured yoga tasks under standardized conditions. Each task was scored using a unified 0 to 3 scale based on accuracy, stability, sequencing fidelity, and regulatory control. Methods Summary Population: Children aged 8 to 15 years (ADHD and control participants in pilot phase) Setting: Quiet, distraction-minimized assessment room with standardized camera setup Administration: Guided demonstration followed by performance-based observation Scoring: Domain-wise 0 to 3 rating scale with explicit behavioral anchors AI Metrics: Pose-estimation–derived kinematic features extracted from video recordings for selected tasks Files Included Demographic Data The demographic dataset contains de-identified participant-level information including age, sex, group status (ADHD or control), and related study characteristics. These variables enable group-level comparisons and covariate-adjusted analyses in motor and regulatory performance. Human Rating Dataset The human rating dataset includes domain-wise Y-MAT scores assigned by trained expert raters using the standardized 0 to 3 scoring framework described in the manual . Where applicable, inter-rater comparison variables are included to facilitate reliability analyses and agreement statistics. AI-Derived Task Metrics Separate CSV files contain pose-estimation–derived kinematic features extracted from recorded task performances. These metrics provide objective quantification of movement quality, sequencing fidelity, bilateral coordination, balance stability, and fine motor control. Task-specific datasets are provided for Dynamic Trikonasana, Vrikshasana, Surya Namaskara, and Alternating Hand Mudras. These features are intended to support validation of automated scoring approaches and computational modeling of motor regulation. Combined Human and AI Rating File An integrated dataset links participant identifiers to both expert domain scores and AI-derived kinematic variables. This combined file enables direct comparison between clinical ratings and computational metrics for validation and exploratory modeling. Administration Manual (PDF) The administration manual provides the standardized protocol for task delivery, environmental setup, ADHD-friendly instructions, demonstration procedures, and scoring workflow . Expert Rating Manual (PDF) The expert rating manual details domain definitions, scoring anchors, task-to-domain mapping, and the unified 0 to 3 scoring framework applied across tasks . STROBE Checklist (PDF) The repository includes the completed Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist corresponding to the associated cross-sectional pilot study. The checklist maps reporting items to relevant sections of the manuscript and is provided to enhance methodological transparency, reporting completeness, and reproducibility. Face-Blurred Video Dataset The repository includes de-identified MP4 video files corresponding to each task. All faces have been digitally blurred prior to release. Each video contains a frame-wise pose-estimation skeletal overlay to allow independent verification of extracted kinematic features. Task-specific movement sequences are preserved in full, and scoring timestamps or annotations are embedded within the video or provided as linked metadata files. These videos enable reproducible scoring, external validation of AI-derived metrics, and development of alternative computer vision pipelines while maintaining participant confidentiality. Ethical Considerations All data are de-identified. No personally identifiable information is included. The study received institutional ethics approval prior to data collection.