Table 7.
Selection of papers that use machine learning methods in validation. Abbreviations used in the column “Model type”: SVM (support vector machine), GPR (gaussian process regression), NN (neural network), RF (random forest), LSTM (long short time memory), HMM (hidden markov model), kNN (k-nearest neighbors), CNN (convolutional neural network), ROC (receiver operating characteristic), and LDA (linear discriminant analysis). Abbreviations used in the column “Outcome”: r (correlation coefficient), NRMSE (normalized root mean square error), RMSE (root mean square error), AUC (area under curve), sens (sensitivity), spe (specificity), and IQR (interquartile range). Studies that use raw data as input have a number of descriptors that correspond to the number of sensors and/or axes multiplied by the length of the recorded data. This is noted (*n) in the table.
| Author | Task | Model Type | Training Size | # of Descriptors | Outcome |
|---|---|---|---|---|---|
| Dobkin et al. [53] | Speed prediction | Naive Bayes | NA | 24 | r = 0.98 |
| Juen et al. [68] | Healthy/patient | SVM | 10–20 | 8 | accuracy = 89.22–94.13% |
| Juen et al. [69] | Speed prediction Distance prediction |
GPR NN SVM |
24 | 60 | error rate = 2.51% error rate = 10.2% |
| Sprint et al. [95] | FIM motor score prediction | SVM RF |
19 | 18 | NRMSE = 10–30% |
| Raknim et al. [86] | Step length estimation Before/after PD |
SVM | 1 | 2 | accuracy = 98% accuracy = 94% |
| Ilias et al. [63] | Motor function prediction | SVM | 6 | 152 | RMSE = 0.46-0.70 r = 0.78–0.79 |
| Cheng et al. [45] | 3 pulmonary severity stages | SVM | 22–25 | 10 | NA |
| McGinnis et al. [79] | Walking speed | SVM | 16 | 32 | RMSE = 10–20% |
| Lipsmeier et al. [77] | Activities | LSTM | 44 | 6 (*n) | accuracy = 98% |
| Mileti et al. [81] | 4 gait phases | HMM | 1–11 | 3 (*n) | = 0.48–0.98 sens= 80–100% spe = 70–90% |
| goodness Index = 10–40% | |||||
| Aich et al. [35] | Healthy/patient | SVM Decision tree Naive Bayes kNN | 36 | 28 | accuracy=91.42% sens/spe = 90.9%/91.2% |
| Kim et al. [70] | Walking/freezing | CNN | 29 | 8 (*n) | f1-score = 91.8 sen/spe = 93.8%/90.1% |
| Vadnerkar et al. [100] | Gait quality | ROC decision boundary | 8 | 1 | accuracy = 84% sen/spe = 75.9%/95.9% |
| Gadaleta et al. [60] | Right/left foot events | CNN | 138 | 24 (*n) | bias = −0.012–0.000 IQR = 0.004–0.032 |
| Teufl et al. [97] | Healthy/patient | SVM | 40 | 10 | accuracy = 87–97% |
| Antos et al. [38] | With/without assistance | RF SVM Naive Bayes Logistic regression LDA | 1–13 | 56 | accuracy = 90–95% |
| Aich et al. [36] | Healthy/patient | kNN SVM Naive Bayes Decision tree | 62 | 10 | accuracy = 88.5% sens/spe = 92.9%/90.9% |
| Abdollahi et al. [34] | Risk of disability | SVM Perceptron | 93 | 920 | accuracy = 60–75% |
| Meisel et al. [80] | Seizure/healthy | LSTM | 68 | 6 (*n) | accuracy = 43% |