ONLINE SUPPLEMENT A

I.  The Mel-scale spectrum is basically the Discrete Cosine Transform (DCT) of the logarithmic energy spectrum of a signal, where the spectral energies are calculated using logarithmically spaced filters with increasing bandwidth (mel-filters). In more details, the mel- frequency ceptral coefficients, MFCCs, were derived as follows: First, the spectral representation of each 2-sec segment was passed through a bank of band-pass filters{hi}. The filters had a triangular-shape frequency response. Their central frequencies (CF) were linearly spaced in the frequency axis for low frequencies but logarithmically spaced for high frequencies, forming the so-called mel-scale filter bank. The logarithm of the resulting mel-spectrum is calculated and transformed via the DCT, yielding an MFCC sequence for each corresponding CF 1.

II.  The adjusted R-square (coefficient of determination) is defined as Ra2=1-SSResid SSTotaln-1n-d-1, where n is the number of responses and d the degrees of the polynomial. SSTotal measures the total sum of squares: SSTotal=i(yi-y)2, with y the mean of the observed data. SSResid measures the residual sum of squares: SSResid=i(yi- yi)2. Note that SSTotal and SSResid are indicators of the sample variances of the dependent variable and the estimated residuals respectively. Adjusted R-square is a refinement of the R-square statistic accounting for the degrees of freedom, and thus making model fits comparable.

III.  Point-wise (or non-simultaneous) prediction bounds measure the confidence that a new observation will lie within the bound interval, given the predictor values x . These bounds are given by B=y±t s2 + xSxT where s2 is the mean squared error, t is the inverse of Student's-t cumulative distribution with respect to the corresponding confidence level, and S is the covariance matrix, (xTx)-1s2 , of the estimates .

REFERENCES

1.  Davis S, Mermelstein P. Comparison of parametric representations for monosyllabic word recognition in continuously spoken sentences [Internet]. IEEE; 1980. Available from: http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=1163420

ONLINE SUPPLEMENT B

E-Table 1. Linear Regression Line fits of extracted features (rows) with respect to individual subject‘s characteristic Age, Height and Weight (columns). The adjusted R-square is also shown.

AGE / HEIGHT / WEIGHT
Fitted regression line / R2a / Fitted regression line / R2a / Fitted regression line / R2a
RR / 30.449
− 1.953 × age / 0.222 / 44.511
− 0.216 × height / 0.263 / 35.462
− 0.770 × weight / 0.243
HR / 123.525
− 5.076 × age / 0.184 / 155.267
− 0.504 × height / 0.175 / 136.919
− 2.031 × weight / 0.208
MFCC1 / 4.637
+ 0.115 × age / 0.140 / 3.881
+ 0.012 × height / 0.143 / 4.459
+ 0.036 × weight / 0.092
MFCC2 / −0.068
+ 0.088 × age / 0.140 / −0.607
+ 0.009 × height / 0.129 / −0.205
+ 0.027 × weight / 0.093
MFCC3 / 0.991
+ 0.041 × age / 0.081 / 0.719
+ 0.004 × height / 0.085 / 0.936
+ 0.012 × weight / 0.046
PW / 171.631
− 6.142 × age / 0.024 / 213.992
− 0.657 × height / 0.028 / 182.102
− 1.980 × weight / 0.016
SL / −9.505
− 0.293 × age / 0.175 / −7.468
− 0.032 × height / 0.196 / −8.981
− 0.097 × weight / 0.132
PR / 0.990
+ 0.002 × age / 0.098 / 0.981
+ 0.000 × height / 0.082 / 0.987
+ 0.000 × weight / 0.064
PLN / 8586.491
+ 528.046 × age / 0.164 / 4912.753
+ 56.866 × height / 0.185 / 7756.135
+ 164.376 × weight / 0.110

E-Table 2. Linear Regression Line fits of extracted features (rows) with respect to the combined subject‘s characteristic Age, Height and Weight (columns). The adjusted R−square is also shown.

AGE / HEIGHT / WEIGHT
Fitted regression line / R2a
RR / 44.974
+ 0.456 × age − 0.196 × height − 0.260 × weight / 0.259
HR / 118.632
− 3.453 × age + 0.277 × height − 1.842 × weight / 0.208
MFCC1 / 3.973
+ 0.050 × age + 0.013 × height − 0.024 × weight / 0.142
MFCC2 / −0.244
+ 0.074 × age + 0.004 × height − 0.009 × weight / 0.131
MFCC3 / 0.694
+ 0.015 × age + 0.006 × height − 0.013 × weight / 0.085
PW / 220.637
− 0.670 × age − 0.892 × height + 1.226 × weight / 0.016
SL / −7.109
− 0.013 × age − 0.043 × height + 0.050 × weight / 0.192
PR / 0.991
+ 0.002 × age − 0.000023 × height − 0.000041 × weight / 0.086
PLN / 3844.605
+ 30.715 × age + 89.111 × height − 144.062 × weight / 0.190

ONLINE SUPPLEMENT C

E-Figure 1: Linear fit (solid line) for each feature (rows, y axis) with respect to patient characteristics (columns, × axis). Point-wise prediction bounds with 95% confidence level are also shown with dashed lines. Inset: R2a, the adjusted coefficient of determination of the quadratic fit; r, the linear correlation coefficient only if significant correlation was achieved.

E-Table 3 Linear Regression Line fits of extracted features (rows) with respect to individual subject’s z-scores Weight-for-Height WHZ, Weight-for-Age WAZ, Height-for-Age HAZ and Body mass index-for-Age BAZ. The adjusted R-square is also shown.

WHZ / HAZ / WAZ / BAZ
Fitted regression line / R2a / Fitted regression line / R2a / Fitted regression line / R2a / Fitted regression line / R2a
RR / 26.362
- 0.571 × WHZ / 0.015 / 26.266
+ 0.123 × HAZ / −0.006 / 26.203
− 0.592 × WAZ / 0.009 / 26.387
− 0.617 × BAZ / 0.020
HR / 113.080
− 2.213 × WHZ / 0.033 / 112.794
+ 0.717 × HAZ / −0.004 / 112.472
− 2.051 × WAZ / 0.016 / 113.164
− 2.349 × BAZ / 0.041
MFCC1 / 4.894
− 0.033 × WHZ / 0.007 / 4.884
− 0.006 × HAZ / −0.006 / 4.884
− 0.042 × WAZ / 0.007 / 4.894
− 0.032 × BAZ / 0.006
MFCC2 / 0.127
− 0.017 × WHZ / −0.001 / 0.117
− 0.016 × HAZ / −0.003 / 0.122
− 0.029 × WAZ / 0.005 / 0.127
− 0.016 × BAZ / −0.001
MFCC3 / 1.083
− 0.017 × WHZ / 0.011 / 1.078
− 0.002 × HAZ / −0.006 / 1.078
− 0.021 × WAZ / 0.010 / 1.083
− 0.017 × BAZ / 0.011
PW / 157.639
+ 2.831 × WHZ / 0.001 / 158.105
− 0.641 × HAZ / −0.006 / 158.423
+ 2.818 × WAZ / −0.002 / 157.529
+ 3.015 × BAZ / 0.002
SL / −10.156
+ 0.066 × WHZ / 0.004 / −10.138
+ 0.003 × HAZ / −0.007 / −10.137
+ 0.078 × WAZ / 0.003 / −10.155
+ 0.060 × BAZ / 0.002
PR / 0.993
− 0.000 × WHZ / −0.006 / 0.993
− 0.001 × HAZ / 0.008 / 0.993
− 0.000 × WAZ / 0.001 / 0.993
− 0.000 × BAZ / −0.007
PLN / 9769.056
− 160.492 × WHZ / 0.011 / 9732.245
+ 7.307 × HAZ / −0.007 / 9724.069
− 177.698 × WAZ / 0.008 / 9769.243
− 148.156 × BAZ / 0.009

E-Table 4. Linear Regression Line fits of extracted features (rows) with respect to the combined subject’s z-scores Weight-for-Height (WHZ), Weight-for-Age (WAZ), Height-for-Age (HAZ) and Body mass index-for-Age (BAZ). The adjusted R-square is also shown.

Fitted regression line / R2a
RR / 26.236
+ 9.971 × WHZ − 0.188 × HAZ − 1.125 × WAZ / 0.039
HR / 112.863
+ 27.695 × WHZ − 5.253 × HAZ + 5.220 × WAZ / 0.061
MFCC1 / 4.888
− 0.003 × WHZ + 0.353 × HAZ − 0.609 × WAZ / 0.002
MFCC2 / 0.118
+ 0.073 × WHZ + 0.134 × HAZ − 0.266 × WAZ / −0.011
MFCC3 / 1.080
+ 0.076 × WHZ + 0.036 × HAZ − 0.081 × WAZ / −0.005
PW / 158.082
− 40.475 × WHZ + 7.350 × HAZ − 6.515 × WAZ / −0.009
SL / −10.151
+ 0.590 × WHZ − 1.333 × HAZ + 2.163 × WAZ / 0.024
PR / 0.993
− 0.001 × WHZ + 0.001 × HAZ − 0.003 × WAZ / −0.010
PLN / 9762.467
− 1027.470 × WHZ + 2581.475 × HAZ − 4195.872 × WAZ / 0.034