Ultra-short term HRV features as surrogates of short term HRV: a case study on mental stress detection in real life

Figures

• 

  Fig. 1. HRV processing workflow. ECG: Electrocardiogram; NN/RR is the ratio of the total RR intervals labelled as NN (normal-to-normal beats); short term: HRV is analyzed in 5 min excerpts; ultra-short term: HRV is analyzed in excerpts of 3, 2, 1 and 0.5 min length

• 

  Fig. 2. Segmentation process. The ultra-short HRV features were extracted from the central position of the 5 min NN excerpts (left-hand side). This procedure was repeated for the shortest significant length of NN excerpts. The shortest excerpts were extracted from different positions, without overlapping (right-hand side)

• 

  Fig. 3. Methodological workflow for the identification of the good surrogates. This process was repeated for each HRV feature at each time scale. The complete list of feature computed at each time scale is reported in Table 1. p: p-value; trend analysis: ↓↓ (↑↑): significantly lower (higher) under stress (p < .05), ↓(↑) lower (higher) under stress (p > .05); rho: Spearman’s rank coefficient, prho: Spearman’s rank p-value

• 

  Fig. 4. Framework of feature selection