Long and short term QT-RR interval co-variability in type 2 diabetes

Masaki Furuya,Yuta Masuda,Kei Sato,Toshihiro Nishibe,Kazuo Yana,Takuya Ono

DOI: https://doi.org/10.1109/EMBC.2014.6943523

Abstract:This paper examines the long and short term co-variability of QT and RR intervals for diabetic patients to explore if the QT-RR co-variability could yield a noble index for the stratification of clinical severity of the disease. Twenty four hour Holter ECG recordings are made for 19 type 2 diabetic (T2DM) patients and 25 normal subjects. RR and QT intervals are extracted from ECG signals sampled at 200 Hz and their co-variability has been examined. To see the long term QT-RR co-variability, correlation coefficients and mutual entropies between QT and RR intervals have been estimated for original beat to beat intervals and smoothed median interval series of successive one hundred beats. Mutual entropy for both beat-to-beat and smoothed median QT and RR interval series showed statistically significant differences between T2DM and control subjects whereas differences in correlation coefficients showed significant difference only for beat-to-beat intervals. Mutual entropy between both beat-to-beat and smoothed median QT-RR interval sequences showed the equally well separation between T2DM patients and control subjects: Mutual entropy and serial correlation coefficients for beat to beat intervals are respectively 1.42 ± 0.33 (bits), 0.856 ± 0.055 for control and 0.752 ± 0.23 (bits), 0.756 ± 0.10 for T2DM patients. Scatter diagram between RR and QT intervals show apparent nonlinearity which validate this result. Short term QT-RR co-variability has been examined by spline smoothed QTc series and sporadic changes have been observed for the control subjects whereas no such changes are found in diabetic patients. This new phenomenon could be a mean for the clinical characterization of diabetes.