2025/07/13 by Pablo Tadeo Ríos-Gallardo, Luis Enrique Carranza García, Carlos Balsalobre‐Fernández +1 · 1 voice · 1 citation
Health Professions · Medicine · #Occupational Health and Performance #Sports Performance and Training #Sports injuries and prevention
paper · doi:10.1080/1091367x.2025.2532391
openalex publication_date 2025/07/13 · openalex created_date 2025/10/10 · openalex updated_date 2026/06/19
This study examined the reliability and validity of a 30 fps artificial intelligence (AI)-based smartphone app for estimating countermovement jump (CMJ) height via bounding box displacement. Eighty male participants from four athletic populations completed two test sessions. Jump height was compared to a criterion force platform (impulse-momentum), an infrared timing system (flight time), and manual high-speed video (240 fps; flight time). Between-session reliability was excellent (ICC ≥0.90) for the platform, infrared, and manual methods (ICCs ≥0.96; CVs < 4%), while the AI app showed acceptable reliability (ICC ≈0.90) but greater variability (CV ≈ 7%). Compared to the force platform, AI estimates showed acceptable agreement (ICC ≈0.91), though with systematic underestimation (bias ≈ −3.2 cm), proportional bias, and wider limits of agreement. While highly accessible, the AI app was less precise due to lower frame rate and its alternative measurement approach. Tool selection should consider accuracy, accessibility, and context-specific needs.