Accurate motion analysis is essential to sports training. Analysis solutions for classifying motion data encounter limitations. Dynamic Time Warping normalizes temporal discrepancies within time series to identify commonalities, thereby dissolving unique sequencing patterns across motions during comparison. Sports motor skills require precise temporal alignment of body part motions or rhythmic synchronization, which necessitates special consideration to effectively normalize time differences. We present a novel approach for identifying motion differences resulting from skill manipulations used in sports training. The method leverages sensor suit data to visually compare joint positions as skeletons against reference values. By quantifying these positions numerically, it calculates differences using the root-mean-square deviation. After manually aligning the recordings at key points (apex of a motion), the Kabsch algorithm adjusts the orientation and translation of the skeletons to minimize RMSD as a measure of body position differences. Examining minimal RMSD frame by frame reveals the degree of dissimilarity between motions. User studies tested the method’s feasibility for future examinations, specifically on the impact of manipulations on motion execution. Data from a volleyball serve were compared among a player using different ball types. The same applies to a handball standing throw. In both sports, two differently skilled players participated. Findings of subtle within-subject differences demonstrate the method’s feasibility to gain a deeper understanding of manipulations influencing sport-specific motions. The method serves scientific and practical educational purposes and addresses the need for objective and efficient means of motion evaluation. Our understanding of athletes’ motions can be increased, facilitating evidence-based training strategies.
Current ML lifecycle frameworks provide limited support for continuous stakeholder alignment and infrastructure evolution, particularly in sensor-based AI systems. We present CAPTURE, a seven-phase framework (Consult, Articulate, Protocol, Terraform, Utilize, Reify, Evolve) that integrates stakeholder-centered requirements engineering with MLOps practices to address these gaps. The framework was synthesized from four established standards (ISO/IEC 22989, ISO 9241-210, CRISP-ML(Q), SE4ML) and validated through a longitudinal five-year case study of a psychomotor skill learning system alongside semi-structured interviews with ten domain experts. The evaluation demonstrates that CAPTURE supports governance of iterative development and strategic evolution through explicit decision gates. Expert assessments confirm the necessity of the intermediate stakeholder-alignment layer and substantiate the participatory modeling approach. By connecting technical MLOps with human-centered design, CAPTURE reduces the risk that sensor-based AI systems become ungoverned, non-compliant, or misaligned with user needs over time.