An auditable framework for ranking movement tasks in skeleton-derived autism intervention data
Year: 2026
Type: article
Abstract: We conducted a secondary computational analysis of the public MMASD autism intervention dataset to examine how movement tasks are ranked using skeleton-derived evidence. The analysis comprised 1,693 traceable derived logical segments from 32 autistic children across 11 activities; paired native-24/native-71 representations were collapsed. The 1,315 clips reported in the source study were retained as a historical reference, not equated with the derived segments. A seven-dimension Deployment Suitability Score (DSS) was examined using single-criterion rankings, TOPSIS, weight and proxy-score perturbations, participant-balanced aggregation, and approximate degradation tests. Under the primary configuration, Arm Swing, Chest Expansion and Body Swing ranked first to third (DSS 0.892, 0.839 and 0.829); Squat ranked fourth (0.759). Participant balancing moved Squat to second and Chest Expansion to fourth, changing top-three membership. Severe derived-table degradation preserved the top-three set but not exact full ranking. These findings demonstrate that apparent ranking stability depends on which uncertainty and representation choices are tested. The study evaluates task-level measurement and ranking behaviour, not clinical relevance, smartphone recording feasibility or a service workflow. No diagnosis, screening, child-level prediction or clinical benefit was assessed.
Source: Scientific Reports
Author: Runchen Cai
Institution: Quanzhou Normal University
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: gold
Estimated APC paid: $2,690
OpenAlex ID: W7235116349
