Authors:
Cleber França Carvalho
;
Thilini Savindya Karunarathna
and
Zilu Liang
Affiliation:
Ubiquitous and Personal Computing Lab, Kyoto University of Advanced Science (KUAS), Kyoto, Japan
Keyword(s):
Continuous Glucose Monitoring, Multiscale Entropy, Diabetes, Prediabetes, Approximate Entropy, Attention Entropy, Dispersion Entropy.
Abstract:
The advent of continuous glucose monitoring (CGM) has made it possible to measure glucose frequently in daily life. This availability of glucose time series enables advanced analysis to uncover patterns in glycaemic dynamics that were previously undetectable with traditional blood-sample-based measurements. One such analytical method is multiscale entropy (MSE), which assesses the complexity of time series data across varying time scales. In this study, we performed a comparative analysis of MSE across three cohorts: individuals with type 1 diabetes (T1D), type 2 diabetes (T2D) and prediabetes (PRED). Our goal was to identify potential differences in glucose dynamics across these groups. We applied three base entropies, including approximate entropy (ApEn), attention entropy (AttnEn) and dispersion entropy (DispEn). We found that AttnEn and DispEn were useful in distinguishing between individuals with diabetes (both T1D and T2D) and those with prediabetes, whereas ApEn did not show s
ignificant discriminative power. Furthermore, we observed no substantial differences between T1D and T2D in terms of their MSE profiles. These results suggest that MSE, with appropriate base entropy measures, holds promise as a tool for developing biomarkers to differentiate between diabetes and prediabetes. Future studies could explore additional base entropy measures and analysing larger, more diverse datasets.
(More)