Table 2.
Performance of several gait phase detection systems reported in the literature. The number of gait phases considered in these papers is four. The relative temporal error is calculated by dividing the absolute temporal error to the stride duration.
| Paper | Sensors | Healthy Subject # | Sample Accuracy or (TPR, TNR) | Phase accuracy (Deletions, Insertions) | Absolute Temporal Error (ms) | Relative Temporal Error (%) (Stride Duration, Speed) |
|---|---|---|---|---|---|---|
| Rueterbories et al. [33] | Acceler-ometers | 10 | (>96%, >91%) | (15.8%, N/A) | 50.2 | 3.2% (1.55 s, 70 steps/min) |
| Mannini et al. [34] | Gyro | 4 | (95%, 95%) | (0%, 12%) | 35 | 3.0% (1.15 s, 3–6 km/h) |
| Taborri et al. 2014 [35] | IMU | 10 | (>95%, >95%) | N/A | N/A | N/A (1.60 s, 1.8–5.4 km/h) |
| Mannini et al. [10] | Gyro | 6 | (95%, 98%) | (0%, 4.3%) | < 20 | N/A (N/A, 3–7 km/h) |
| Mannini et al. [36] | Gyro | 9 | N/A | (0%, 0.45%) | 45–35 | 3.7% (1.09 s, 4.8 km/h) |
| Pappas et al. [9] | Gyro, insole FSRs | 10 | N/A | (0%, 0%) | 53.8 | 4.5% (1.2 s, 3 km/h) |
| The present paper | FMG on ankle | 9 | 91.2% | (0.03%, 0%) | 55.2 | 2.1% (2.60 s, 1–2 km/h) |