Correlating Short-Term Heart Rate Variability and Instantaneous Blood Glucose Measurements

摘要

In this research, we aim to analyze the correlations between instantaneous blood glucose measurements and short term (30 seconds to 5 minutes) heart rate variability (HRV) parameters and check the research hypothesis that these HRV parameters indicate the ability of a patient to control the blood glucose level. Methodology: Pearson and Spearman's Rank correlations are used as methods within this paper. Although 155 patients with heart problems and arrhythmia were included in the study, only intervals between normal heartbeats were evaluated. Data: The HRV parameters correlated to the blood glucose levels are divided into time-domain (TD) HRV parameters: SDNN, SDNN Prima (SDNN-1), ASDNN, ASDNN Prima (ASDNN-1), ASDNN Secunda (ASDNN-2), ASDNN Tertia (ASDNN-3), SDANN Secunda (SDANN-2), SDANN Tertia (SDANN-3), NN50, NN50 Prima (NN50-1), pNN50, pNN50 Prima (pNN50-1), rMSSD and rMSSD Prima (rMSSD-1); and non-linear HRV parameters: SD1, SD2, SD1/SD2. Several methods were utilized to process the ECG signal strip and eliminate the ectopic beats, artifacts, lost signals, noise, and other segments that influence HRV metrics. Conclusion: We found that SD1/SD2 showed the strongest positive correlations for 5 minute ECG recordings, 45–40 minutes before ($r = 0.36$) and 20–25 minutes after ($r = 0.31$) the fasting glucose measurement satisfying the condition $p$ ≤ 0.05. We conclude that short term HRV parameters can't indicate a strong linear or monotonic correlation to the blood glucose regulation ability, but can motivate further research including deep learning.