Skip to main content
. Author manuscript; available in PMC: 2022 Mar 21.
Published in final edited form as: IEEE Rev Biomed Eng. 2017 Oct 24;11:2–20. doi: 10.1109/RBME.2017.2763681

Table V. Methods Used to Assess BR Algorithm Performance.

Category No. publications (%)
Application of BR algorithms
Number of algorithms assessed
1 94 (48.0)
2–5 76 (38.8)
6–10 17 (8.7)
11–15 6(3.1)
≥ 16 3 (1.5)
Input signal(s)
ECG 98 (50.0)
PPG 112(57.1)
Fusion of ECG and PPG 5 (2.6)
Pulse transit time 8 (4.1)
Window duration [s]
< 30 10(5.1)
30–59 46 (23.5)
60–89 50 (25.5)
≥ 90 10(5.1)
Unknown 78 (39.8)
Datasets
Age(s) of subjects [years]
0–0.1: Neonate 5 (2.6)
0.1–17: Pediatric 27 (13.8)
18–39: Young adult 122 (62.2)
40–69: Middle-aged adult 76 (38.8)
≥ 70: Elderly adult 50 (25.5)
Unknown 57 (29.1)
Level(s) of illness
Healthy 127 (64.8)
Sick in community 22(11.2)
Acutely ill 15 (7.7)
Critically ill 52 (26.5)
unknown 9 (4.6)
Type(s) of breathing
Spontaneous 150 (76.5)
Metronome 45 (23.0)
Ventilated 32 (16.3)
Simulated 7 (3.6)
unknown 25 (12.8)
Number of datasets used
1 164 (83.7)
2 30 (15.3)
3 1 (0.5)
4 1 (0.5)
Comparison with reference BRs
Reference BR equipment
Air flow or pressure 45 (23.0)
Impedance pneumography (ImP) 48 (24.5)
Capnography 33 (16.8)
Inductance plethymography (InP) 14(7.1)
Piezoelectric 9 (4.6)
Strain gauge 19 (9.7)
Metronome 9 (4.6)
Other 22(11.2)
None 5 (2.6)
unknown 26 (13.3)
Common statistical measures
Error statistic 127 (64.8)
Breath detection statistic 19 (9.7)
Bias 46 (23.5)
Limits of agreement (LOAs) 46 (23.5)
Correlation 27 (13.8)
Proportion of windows 14(7.1)