Table 3 Variables that significantly contributed to explained variance in the binary logistic regression model (i.e., prediction whether a person would be below or above cut-off for clinically relevant depressive symptoms)
From: Detecting subtle signs of depression with automated speech analysis in a non-clinical sample
Predictor | Coefficient | Standard error | Wald | p-value |
|---|---|---|---|---|
Speech ratio negative | 426.79 | 169.30 | 6.36 | 0.012 |
Mean f0 positive | -0.55 | 0.22 | 6.26 | 0.012 |
Standard deviation f0 positive | -0.64 | 0.25 | 6.57 | 0.010 |
H1 a3 harmonic difference positive | -0.24 | 0.09 | 7.02 | 0.008 |
Average mfcc 2 negative | 0.17 | 0.09 | 3.54 | 0.06 |
Average mfcc 3 negative | 0.53 | 0.20 | 6.77 | 0.009 |
Average mfcc 5 negative | -0.98 | 0.34 | 8.27 | 0.004 |
Average mfcc 7 negative | -1.26 | 0.45 | 7.66 | 0.006 |
Average mfcc 7 positive | 1.32 | 0.46 | 8.16 | 0.004 |
Mean pause duration positive | -18.34 | 7.19 | 6.50 | 0.011 |
Local jitter positive | 27.19 | 11.53 | 5.56 | 0.018 |
Local absolute jitter positive | -6326.19 | 2537.08 | 6.22 | 0.013 |
Power spectrum ratio negative | 89.27 | 32.68 | 7.46 | 0.006 |
Mean power positive | -5389.42 | 2067.18 | 6.79 | 0.009 |
Intensity standard deviation negative | 1.04 | 0.40 | 6.63 | 0.010 |
Apq5 shimmer positive | -2.35 | 1.05 | 4.99 | 0.025 |
Local shimmer negative | 2.57 | 0.97 | 6.96 | 0.008 |
Average dependency distance positive | 6.60 | 2.53 | 6.79 | 0.009 |