Validity and Inter-Device Reliability of an Artificial Intelligence App for Real-Time Assessment of 505 Change of Direction Tests.

The present study aimed to explore the validity and inter-device reliability of a novel artificial intelligence app (Asstrapp) for real-time measurement of the traditional (tra505) and modified-505 (mod505) change of direction (COD) tests. Twenty-five male Sports Science students (age, 23.5 ± 3.27 years; body height, 178 ± 9.76 cm; body mass, 79.4 ± 14.7 kg) completed 12 trials each, consisting of six tra505 and six mod505 trials. Completion times were simultaneously recorded via single-beam electronic timing gates (ETG) and two different iPhones (APP1 and APP2). In total 300 trials were collected across the two tests, using all three devices, to establish the reliability and validity of the app. The coefficient of variation indicated a similar level of dispersion between the ETG (≤ 2.73%), APP1 (≤ 2.39%) and APP2 (≤ 2.52%). Intraclass correlation coefficients (ICC) revealed excellent reliability among the three timing devices (ICC ≥ 0.99) and Asstrapp relative reliability was excellent for both APP1 (ICC ≥ 0.91) and APP2 (ICC ≥ 0.91). There was a practically perfect correlation and agreement between ETG and Asstrapp (APP1: r = 0.97; APP2: r = 0.97) for both COD tests. However, small but significant differences were found between smartphones and ETG for tra505 (ES ≤ 0.33; p < 0.05). Collectively, these findings support the use of Asstrapp for real-time assessment of both 505 COD tests.