Comparing A Novel Smartphone Application To The Kinect V2 For Assessing ACL Injury Risk

Screening for athletes at high risk for anterior cruciate ligament (ACL) tear who may optimally benefit from risk reduction programs remains a challenge. Initial coronal (IC), peak coronal (PC), and peak sagittal (PS) angles of a drop vertical jump (DVJ) measured by the Kinect v2 device (Microsoft) have been associated with ACL tear risk in collegiate varsity athletes. Development of a similar modality utilizing a smartphone application would improve simplicity, cost, and accessibility, allowing for widespread screening. PURPOSE: To compare the DVJ tracking performance of a novel smartphone application to the previously validated Kinect v2 device. METHODS: Two hundred fifty-two collegiate varsity athletes performed three DVJs that were recorded simultaneously with a novel smartphone application developed using the Google Mediapipe framework version 0.7 (Pixel 6; Google) and the Kinect v2. Agreement on IC, PC, and PS angles between the two devices was compared using intraclass correlation coefficient (ICC) tests (two-way mixed-effects model, consistency, single measures). High-risk athletes were identified based on published angle cutoffs for Kinect v2 measurements. Receiver operating characteristic (ROC) analysis was used to assess the accuracy of the smartphone application in identifying the same high-risk athletes. RESULTS: ICC values for IC, PC, and PS angles were 0.696 (0.627 - 0.755, P < 0.01), 0.664 (0.589 - 0.728, P < 0.01), and 0.756 (0.697 - 0.804, P < 0.01), respectively, representing good agreement for IC and PC angles and excellent agreement for PS angles. Ninety-six (38%), 23 (9%), and 90 (36%) high-risk athletes were identified based on IC, PC, and PS angle cutoffs for the Kinect v2. ROC analysis demonstrated an area under the curve of 0.89 (0.85 - 0.93), 0.85 (0.77 - 0.93), and 0.89 (0.85 - 0.93) for identifying these high-risk athletes using the smartphone application, indicating excellent accuracy for all parameters. Smartphone application-derived IC, PC, and PS angle cutoffs of 0.74°, 2.20°, and 73.50° identified 83%, 78%, and 84% of high-risk athletes, respectively. CONCLUSIONS: The novel smartphone application identifies athletes marked as high-risk for ACL tear from Kinect v2 cutoffs with excellent accuracy and shows potential for use in widespread screening. This work was supported by MEDTEQ+, Emovi Inc., and Semperform Inc.