Table 3.

Techniques for fusion of RR estimates.

Abbr.	Technique
	Modulation, ${\displaystyle F_{\mathrm{M}1,\dots ,4}}$
FM1	Smart fusion (Karlen et al 2013): RRs estimated from BW, AM and FM respiratory signals (${\displaystyle X_{\mathrm{B}1,2,3}}$) are quality assessed. If their standard deviation is ${\displaystyle ⩽4}$ bpm then RR is estimated as the mean, otherwise no RR is output.
FM2	Spectral peak-conditioned averaging (Lázaro Plaza 2015): Frequency spectra calculated from BW, AM and FM respiratory signals (${\displaystyle X_{\mathrm{B}1,2,3}}$) using the Welch periodogram (FT7) are fused to give a mean spectrum. Only those spectra for which a certain proportion of spectral power is contained within a frequency range centred on the frequency corresponding to the maximum spectral power are included (a modification of the reported method). RR is estimated as the frequency corresponding to the maximum power in the mean spectrum.
FM3	Pole magnitude criterion (Orphanidou et al 2013): The respiratory pole is chosen as the highest magnitude pole obtained from auto-regressive spectral analysis of BW, AM and FM respiratory signals (${\displaystyle X_{\mathrm{B}1,2,3}}$).
FM4	Pole ranking criterion (Orphanidou et al 2009a): The pair of highest magnitude poles obtained from auto-regressive spectral analysis of BW, AM and FM respiratory signals (${\displaystyle X_{\mathrm{B}1,2,3}}$) with the greatest pole ranking criterion (PRC) is selected. ${\displaystyle \mathrm{P}\mathrm{R}\mathrm{C}=\overline{m_{i,j}}/{d{\theta }_{i,j}}^2}$, where ${\displaystyle \overline{m_{i,j}}=(m_i+m_j)/2}$ and ${\displaystyle d{\theta }_{i,j}=|{\theta }_i-{\theta }_j|}$, for ${\displaystyle i,j=1,\dots ,N}$, where N is the number of poles calculated. θ and m are the pole angles and magnitudes respectively. RR is estimated from the mean frequency corresponding to the selected pair.
	Temporal, FT1
FT1	Temporal smoothing (Lázaro et al 2013): estimated RRs, ${\displaystyle {\mathrm{R}\mathrm{R}}_{\mathrm{e}\mathrm{s}\mathrm{t}}}$, are smoothed to give the final RR, ${\displaystyle {\mathrm{R}\mathrm{R}}_i}$, using ${\displaystyle {\mathrm{R}\mathrm{R}}_i=0.2{\mathrm{R}\mathrm{R}}_{\mathrm{e}\mathrm{s}\mathrm{t}}+0.8{\mathrm{R}\mathrm{R}}_{i-1}}$.