Figure 1. Experimental paradigm of (A) standard TUG task. (B) TUG task with verbal instructions to “stop” and “walk” to facilitate voluntary stopping. TF denotes the transition to FOG, and TS denotes the transition to voluntary stopping.

Figure 2. EEG preprocessing and feature extraction for DL models.

Table 1. Participant-based count of normal walking, transition to FOG, and transition to voluntary stopping.

Table 1. Participant-based count of normal walking, transition to FOG, and transition to voluntary stopping.

Subject No.No. of Normal Walking EpochsNo. of Transition to FOG EpochsNo. of Transition Voluntary Stopping Epochs11183212843110433330552368537716815114988010232301150512302461330314151501533267167161717125

Table 2. Five-fold classification performance for transition to FOG vs. transition to voluntary stopping.

Table 2. Five-fold classification performance for transition to FOG vs. transition to voluntary stopping.

ModelAccuracyF1-ScoreCoh-KappaSensitivitySpecificityEEGNet88.09 ± 4.25%80.09 ± 4.62%68.30 ± 2.50%94.42 ± 4.65%96.21 ± 3.52%Shallow ConvNet89.9 ± 2.31%89.21 ± 3.94%70.11 ± 3.91%96.49 ± 2.97%94.36 ± 3.60%Deep ConvNet92.28 ± 2.70%93.02 ± 2.03%72.94 ± 2.27%96.89 ± 2.04%96.91 ± 2.09%

Table 3. LOSO classification performance for transition to FOG vs. transition to voluntary stopping.

Table 3. LOSO classification performance for transition to FOG vs. transition to voluntary stopping.

ModelAccuracyF1-ScoreCoh-KappaSensitivitySpecificityEEGNet87.28 ± 5.89%87.61 ± 5.53%69.19 ± 4.37%84.89 ± 5.72%84.16 ± 4.71%Shallow ConvNet87.92 ± 4.3%82.16 ± 3.02%71.14 ± 4.84%86.23 ± 3.71%85.55 ± 4.62%Deep ConvNet87.83 ± 5.35%84.81 ± 5.86%70.6  ± 5%86.37 ± 3.31%84.72 ± 2.49%

Table 4. Five-fold classification performance for transition to FOG vs. transition to voluntary stopping vs. normal walking.

Table 4. Five-fold classification performance for transition to FOG vs. transition to voluntary stopping vs. normal walking.

ModelAccuracyF1-ScoreCoh-KappaSensitivitySpecificityEEGNet71.92 ± 5.64%69.49 ± 5.38%52.57 ± 4.63%87.8 ± 5.90%84.02 ± 4.06%Shallow ConvNet73.68 ± 3.87%73.53 ± 3.76%57.14 ± 4.53%89.28 ± 4.59%86.2 ± 3.37%Deep ConvNet75.43 ± 1.48%72.52 ± 1.44%58.11 ± 1.64%92.85 ± 1.70%75.86 ± 1.75%

Table 5. LOSO Classification performance for transition to FOG vs. transition to voluntary stopping vs. normal walking.

Table 5. LOSO Classification performance for transition to FOG vs. transition to voluntary stopping vs. normal walking.

ModelAccuracyF1-ScoreCoh-KappaSensitivitySpecificityEEGNet70.85 ± 3.25%70.79 ± 3.86%52.54 ± 5.89%83.83 ± 5.65%82.80 ± 4.13%Shallow ConvNet73.45 ± 3.69%72.84 ± 3.61%54.43 ± 4.92%88.91 ± 5.08%86.34 ± 5.62%Deep ConvNet74.65 ± 4.19%71.54 ± 4.7%57.52 ± 3.42%91.18 ± 5.04%74.46 ± 4.79%