<![CDATA[Background: Planned change-of-direction (COD) performance is important in martial arts, yet field-based assessment remains vulnerable to manual timing and scoring variability. Sensor-based systems may improve measurement standardisation, but their correspondence with conventional field methods and the cross-sectional relationships among kinesthetic perception, body mass index (BMI), sprint performance, and planned COD performance require careful evaluation. Aims: This study evaluated sensor-based versus conventional measurements of planned COD performance and kinesthetic perception and estimated the cross-sectional indirect associations of kinesthetic perception and BMI with planned COD performance through 30-m sprint time. Methods: A cross-sectional sample of 123 junior martial arts athletes aged 14-17 years completed the Smart Star Agility Test (SSAT), SmartPlus Autopad assessment, 30-m sprint test, Modified Martial Arts Dynamic Balance Test, and anthropometric measurements. Method association was described with Pearson correlation; the prespecified two-way mixed-effects, absolute-agreement, average-measures intraclass correlation coefficient (ICC) was retained as a descriptive agreement coefficient. Bland–Altman mean bias and approximate 95% limits of agreement (LoA) were calculated in a common sensor-minus-comparator direction. Cross-sectional indirect effects were estimated with PROCESS Model 4 using 5,000 bootstrap samples. Results: SSAT and stopwatch measurements were strongly associated (r = 0.996, p < 0.001; average-measures ICC = 0.998). The SSAT-minus-stopwatch mean bias was -0.05 s, with approximate 95% LoA from -0.55 to 0.45 s. SmartPlus and manual kinesthetic scores were also strongly associated (r = 0.964, p < 0.001; average-measures ICC = 0.981); the SmartPlus-minus-manual mean bias was 0.12 points, with approximate LoA from -3.06 to 3.30 points. The bootstrap indirect association through 30-m sprint time was non-zero for kinesthetic perception (indirect effect = -0.0224; 95% BootCI [-0.0558, -0.0006]) and BMI (indirect effect = 0.0570; 95% BootCI [0.0121, 0.1191]). Conclusion: The sensor-based measures showed very high correspondence and a small average bias relative to their field comparators under the present protocol. However, no a priori acceptable LoA were defined, and the comparator methods were not criterion standards; therefore, practical interchangeability is not established. The indirect-effect results describe cross-sectional statistical associations rather than causal mediation.]]>
