Skip to main content
. 2021 Jun 8;12:650024. doi: 10.3389/fneur.2021.650024

Table 6.

An overview of the AI techniques applicable for gait analysis.

References Input parameters Technique
Lau et al. (163) Kinematic data Support vector machines (SVM), Artificial neural network (ANN), Radial Basis Function network (RBF), and Bayesian Belief Network (BBN).
Lai et al. (56) Spatiotemporal, kinematic, kinetic, and EMG data Signal processing and computational intelligence methods.
Lau et al. (44) Kinematics data SVM, ANN, RBF.
Kaptein et al. (164) kinematic and physiological data Analysis of variance (ANOVA) supplemented by logistic and partial least squares (PLS) regressions.
Laroche et al. (165) Kinematic trajectories SVM.
Karg et al. (166) time series gait data Hidden Markov Model (HMM).
Cippitelli et al. (167) body joint trajectories Algorithm based on anthropometric models.
Joyseeree et al. (168) Spatiotemporal data Random Forest (RF), boosting, Multilayer Perceptron (MLP), and SVM.
LeMoyne et al. (169) Temporal and kinetic data SVM.
Ferber et al. (170) n/a n/a.
Osis et al. (171) Kinematic data and ground reaction forces Principal Component Analysis (PCA).
Zeng et al. (172) Vertical GRF RBF networks.
Hannink et al. (173) Spatiotemporal data Deep convolutional neural networks.
Caldas et al. (103) IMU data artificial intelligence (AI) algorithms [e.g., artificial neural networks (ANN) and hidden Markov models (HMM)].
Park et al. (174) Spatiotemporal and plantar pressure Random forest classification.
Pham and Yan (175) Vertical GRF Tensor decomposition.
Ertelt et al. (176) GRF Bayesian regulated neural networks.
Haji Ghassemi et al. (177) Inertial data Peak detection, two variants of dynamic time warping (DTW) methods [Euclidean DTW (eDTW) and probabilistic DTW (pDTW)], and hierarchical hidden Markov models (hHMM).
Zhan et al. (178) Stride length A rank-based machine-learning algorithm called disease severity score learning (DSSL).
Zhang et al. (179) GRF SVM.
Bastien et al. (180) Ground reaction forces (GRF) A predictive linear model of the fore-aft GRF.
Galbusera et al. (181) review article Machine learning and deep learning.
Jiang et al. (182) Inertial data, GRF Random forest learning.
Nguyen et al. (183) Inertial data PCA, SVM, ANN.
Prado et al. (184) Temporal data Recurrent Neural Network classifier model.
Waugh et al. (185) Accelerometer data Canonical dynamical system (CDS)Fourier series.
Jauhiainen et al. (186) Kinematic data Cluster analysis.