The impact of feature extraction and selection for the classification of gait patterns between ACL deficient and intact knees based on different classification models

Abstract The anterior cruciate ligament (ACL) plays an important role in stabilizing translation and rotation of the tibia relative to femur. Individuals with ACL deficiency usually demonstrate alterations gait characteristics. Evidence indicates that walking speed, kinetics kinematics on deficient limb, inter-limb asymmetries between intact knees may contribute poor long-term outcomes following deficiency. They corrode function knee joint put it at higher risk degeneration. For purpose developing automatic highly accurate system for detection deficiency, this study investigated classification capability different dynamical features extracted from kinematic kinetic signals when evaluating their impact models. A general feature extraction framework was proposed various features, such as recurrence rate, determinism entropy quantification analysis, fuzzy entropy, Teager-Kaiser energy statistical were included. Different models, including support vector machine (SVM), K-nearest neighbor (KNN), naive Bayes (NB) classifier, decision tree (DT) classifier ensemble learning based Adaboost (ELA) derived discriminant analysis multiple evaluated a comparative study. effectiveness strategy verified using dataset knee, hip ankle waveforms 43 patients unilateral When 2-fold, 10-fold leave-one-out cross-validation styles, highest accuracy discriminating groups contralateral reported be 91.22 $$\%$$ % , 95.12 96.34 respectively,by SVM optimal set. other four classifiers, KNN achieved 78.05 85.37 87.80 respectively. NB 57.56 60.98 61.22 DT 77.56 80.49 83.66 ELA 73.66 79.27 Compared state-of-the-art methods, results superior performance validity method.