Lag‐sensitive cross‐correlation analysis of mother–infant HF ‐ HRV synchrony during free interaction: An exploratory feasibility study

Figures

• 

  FIGURE 1. Representative RRI time‐series signals derived from ECG data for a mother–infant pair after QRS detection, artifact screening, and manual review. The upper two panels show maternal ECG‐derived R peaks and the cleaned maternal RRI time series; the lower two panels show the corresponding infant signals. Five minutes (300 s) of data were used when a complete session was retained; for sessions with shorter analyzable intervals, the corresponding continuous analysis window was used. These RRI time series were used for the frequency‐domain analysis shown in Figure 2.

• 

  FIGURE 2. Representative RRI and HF‐HRV time‐series data. The upper panel shows the RRI time series from a mother–infant pair. The mother's RRI time series is shown in blue, and the infant's RRI time series is shown in orange. The lower panel shows HF‐HRV power derived from spectral analysis of the RRI time series using the Lomb–Scargle periodogram implemented with the plomb function in MATLAB. HF‐HRV was calculated as integrated power within the predefined high‐frequency band and is expressed in msec2. The mother's HF‐HRV power is shown in blue, and the infant's HF‐HRV power is shown in orange. These HF‐HRV time‐series values were used for the subsequent cross‐correlation analyses.

• 

  FIGURE 3. Schematic illustration of the cross‐correlation procedure used to identify synchronization points. For each 30‐s maternal reference window, the infant HF‐HRV window was shifted within the predefined lag range of −30 to +30 s. Positive peaks were detected using the predefined criteria, including r ≥ 0.8, minimum prominence 0.01, and minimum separation 10 s. If multiple eligible peaks were detected, the peak with the highest cross‐correlation coefficient was selected as the synchronization point.

• 

  FIGURE 4. Example graph of cross‐correlation analysis results. Graphical overview of the normalized cross‐correlation analysis using maternal and infant HF‐HRV power. The horizontal axis represents time (s). The cross‐correlation and lag traces illustrate the output from which eligible positive peaks were identified using the predefined criteria. The selected synchronization points were used to calculate the highly correlated time and lag‐category results presented in Table 2.

• 

  FIGURE 5. Session‐level observed synchrony relative to surrogate distributions. Filled circles indicate observed synchrony indices, open squares indicate surrogate means, and open triangles indicate session‐specific surrogate 95th percentiles. The dashed horizontal line indicates the mean observed synchrony index across the 16 sessions. Values are shown for the 16 analyzed sessions using the corresponding analysis‐window denominator for each session. Observed synchrony was considered to exceed the corresponding block‐permuted surrogate distribution when the empirical p‐value was <0.05. Under this criterion, observed synchrony exceeded the surrogate distribution in 13 of 16 sessions.