Purpose: Previous clinical studies have shown that anterior cruciate ligament (ACL) ruptures require reconstructive surgery. The main goal of this study is an objective test definition for unstable knee diagnosis based on real measurements by using infrared cameras and adequate software.
Methods: In the study of gait analysis 35 males with deficient ACL's participated. Pathological parameters for anterior posterior translation (APT) and internal external rotation (IER) and their values of kinematic data were obtained from a gait analysis 3D system. Movement curves were obtained by recording the position of fluorescent markers over time. A machine learning algorithm was developed in order to support decisions on the severity of the ACL injury and its corresponding deficiency. The algorithm was based on logistic regression.
Results: The value of APT, designated as exponentiation of the Ө coefficient (Exp (Ө)) of APT, showed that the likelihood of ACL-deficient knee occurrence due to higher values of APT is 1.1758 (95 % CI) times more frequent than that of the patients with lower values of APT. The value of IER, designated as Exp (Ө) of IER, showed that the patients with higher values of IER present 2.2516 (95 % CI) times higher values of ACL-deficient knee frequency than those with lower values.
Conclusion: This study showed that the creation of ordered pairs of pathological parameters gives a wider picture of ACL deficiency and that such an algorithm may improve both examination and treatment of patients.
Keywords: ACL; AP translation; IE rotation; Knee; Logistic regression.