Table 2.
Occurrence of respiration, non-speech sounds and classification accuracy of the proposed algorithm across individual groups.
| HC | PN | EP | HD | MSA | PSP | All | |
|---|---|---|---|---|---|---|---|
| Occurrence of respirations | |||||||
| Mean/SD (%) (range) | 2.50/4.27 (0–20) | 3.06/3.99 (0–12) | 3.92/5.50 (0–24) | 6.20/6.57 (0–25) | 4.18/6.69 (0–24) | 3.07/4.40 (0–13) | 3.77/5.34 (0–25) |
| Occurrence of non-speech sounds | |||||||
| Mean/SD (%) (range) | 4.90/3.98 (0–13) | 5.64/4.59 (0–17) | 10.90/11.57 (0–45) | 8.60/7.88 (0–36) | 7.21/7.31 (0–22) | 6.62/4.09 (0–13) | 7.43/7.78 (0–45) |
| Algorithm accuracy | |||||||
| False positives FPa | |||||||
| Mean/SD (%) (range) | 0/0 (0–0) | 0/0 (0–0) | 0.39/1.53 (0–19) | 0.49/2.15 (0–12) | 0/0 (0–0) | 0/0 (0–0) | 0.18/1.18 (0–19) |
| False negatives FNb | |||||||
| Mean/SD (%) (range) | 0/0 (0–0) | 0.11/0.72 (0–5) | 0.68/2.75 (0–14) | 0.11/0.66 (0–4) | 0/0 (0–0) | 0.69/2.00 (0–9) | 0.27/1.55 (0–14) |
| Overall ACC | |||||||
| Mean/SD (%) (range) | 100/0 (100–100) | 99.89/0.72 (95–100) | 98.92/3.06 (86–100) | 99.40/2.22 (88–100) | 100/0 (100–100) | 99.30/2.00 (92–100) | 99.56/1.93 (86–100) |
aFalse positives represent percentage of respirations and non-speech sounds identified by algorithm as syllables.
bFalse negatives represent percentage of syllables that were not identified by algorithm.