Evaluating heart rate coupling with Multi-dimensional Recurrence Quantification Analysis (MdRQA) and Cross-Correlation (CC): Analytical decisions and implications

Interpersonal heart rate coupling appears as a promising marker of group performance and coordination. However, discrepancy in methodological approaches challenges the interpretation of the results across studies. A more consistent and informed approach to preprocessing and analysis techniques could produce a more coherent picture on the value of heart rate coupling. In this manuscript, we examine the effects of different methodological decisions (resampling, analysis methods and their parameters) on the ability to capture coupling. For this purpose, we utilize both experimental dyadic heart rate data collected during a collaborative drumming task, and synthetic data with similar characteristics. Starting from preprocessing, we compare heart rate resampling window sizes, juxtapose Cross-Correlation (CC), windowed CC and Multidimensional Recurrence Quantification Analysis (MdRQA) and elaborate on the MdRQA parameter estimation and form of quantification. Our findings imply that CC as well as unembedded MdRQA recurrence rate (%REC) better distinguish coupled heart rate data from uncoupled ones, or, in other words, identify heart rate coupling. Moreover, for this objective, we suggest using Beats per Minute applying a small data-driven resampling window size, and, for MdRQA setting a moderate threshold yielding %REC between 15–25.