Relationship between heart-rate variability and glycemia

Both heart-rate variability (HRV) and blood glucose levels are important information for assessing the individual's state of health. The aim of the present study is to assess the relationship that exists between HRV computed in 5-minutes intervals and glycemia values in the same 5-minutes intervals, in a population of subjects affected by type 1 diabetes mellitus while carrying out activities of daily living. The electrocardiogram (ECG) is a diagnostic tool used in clinical practice to record electrical heart activity. HRV is the measure of the time variation in R– R intervals during sinus arrhythmia, denoted as “N–N” intervals; it represents the influence of the autonomic nervous system on the sinoatrial (SA) node. While the activity of the parasympathetic nervous system increases HRV, the activity of the sympathetic nervous system decreases it. The time domain estimates of HRV that are more commonly used are: the standard deviation of all NN intervals (SDNN), the root mean square of successive NN interval differences (RMSSD), the heart rate (HR), the HRV triangular index (TRI), the triangular interpolation of the NN interval histogram (TINN), Approximate Entropy (ApEn), the ratio of successive NN intervals whose difference is > 50ms (NN50), the ratio between NN50 and all NN in the defined time period (pNN50), the measure of the standard deviation of instantaneous RR interval variability derived from the RR Poincarè plot (SD1), the measure of the continuous long-term RR interval variability derived from the RR Poincarè plot (SD2), and the ratio SD1/SD2. HRV frequency domain features are the power of high frequency (HF) component of HRV, the power of low frequency (LF) component of HRV, and the ratio LF/HF. All these HRV estimates could give information on the body's ability to regulate blood glucose levels and on the trend of the glycemic profile. In the present study the dataset used is publicly available and composed of 9 patients with type 1 diabetes mellitus; they were monitored with the wearable device Zephyr BioHarness3.0 for ECG recording and with iPro2 CGM system for glycemia measurements for four days in everyday life conditions. The ECG signals were visually analyzed with MATLAB R2020a and classified as having at least 5 consecutive minutes of ECG uninterrupted by noise or not. Also, CGM data were incomplete or lacking in some of the four days. Then the analysis was restricted to the subjects presenting at least one day with both glycemic tracing and ECG recording and not only noise. Then all the 5-minutes ECG segments in which the ECG was visible and suitable for the calculation of the HRV parameters were given in input to the HRVTool, a software developed as a Matlab Graphical User Interface to compute a list of HRV parameters. The Pearson’s correlation coefficient (r) was used to study the strength of the correlation between each HRV parameter obtained and the glycemia value recorded in the same 5 minutes in each subject and considering all the subjects grouped. In conclusion, from this study it emerges the existence of weak correlations between HRV parameters computed in 5-minutes intervals and glycemia values in the same 5-minutes intervals in a population of subjects affected by type 1 diabetes mellitus while carrying out activities of daily living.