Electroencephalogram Spectral Moments for the Detection of Nocturnal Hypoglycemia

Abstract

The potential of using an electroencephalogram (EEG) to detect hypoglycemia in patients with type 1 diabetes (T1D) has been investigated in both time and frequency domains.Under hyperinsulinemic hypoglycemic clamp conditions, we have shown that the brain's response to hypoglycemic episodes could be described by the centroid frequency and spectral gyration radius evaluated from spectral moments of EEG signals. The aim of this paper is to investigate the effect of hypoglycemia on spectral moments in EEG epochs of different durations and to propose the optimal time window for hypoglycemia detection without using clamp protocols. The incidence of hypoglycemic episodes at night time in five T1D adolescents was analyzed from selected data of ten days of observations in this study. We found that hypoglycemia is associated with significant changes (P<0.05) in spectral moments of EEG segments in different lengths. Specifically, the changes were more pronounced on the occipital lobe. We used effect size as a measure to determine the best EEG epoch duration for the detection of hypoglycemic episodes. Using Bayesian neural networks, this study showed that 30 second segments provide the best detection rate of hypoglycemia. In addition, Clarke's error grid analysis confirms the correlation between hypoglycemia and EEG spectral moments of this optimal time window, with 86% of clinically acceptable estimated blood glucose values. These results confirm the potential of using EEG spectral moments to detect the occurrence of hypoglycemia. IndexTerms-Electroencephalogram (EEG), hypoglycemia, optimal time window, spectral moment.

Detection of Hypoglycemia Using Measures of EEG Complexity in Type 1 Diabetes Patients

“…The obtained accuracy, using 21 complexity features as inputs for the NN, was almost 90%. This result seems remarkable, given that other recent works aimed at detecting nocturnal hypoglycemia by EEG spectral moments reached a sensitivity and specificity of 85% and 52%, respectively [ 32 ]. In addition, a classifier fed by the sole power spectrum features on our data led to an accuracy of 71% only.…”

Ability of Current Machine Learning Algorithms to Predict and Detect Hypoglycemia in Patients With Diabetes Mellitus: Meta-analysis

“…Table 1 shows the summary of study characteristics. Of the 33 studies, 19 studies (58%) [26][27][28][29][30][31]33,35,36,[38][39][40][41][42][44][45][46][47]54] predicted hypoglycemia, and the remaining 14 studies (42%) detected hypoglycemia [15,20,25,32,34,37,43,[48][49][50][51][52][53]55]. As much as 25 of the 33 included studies (76%) [15,20,[25][26][27]29,30,32,35,36,38,39,[41][42][43][44]…”

“…Regarding the time of day when hypoglycemic events occurred, nocturnal hypoglycemia was the most frequently reported (14 studies of the 33 included studies; 42%) [15,20,26,30,32,35,36,41,44,[49][50][51][52][53]). As to the place of the supposed hypoglycemic episode, 16 of the 19 studies that predicted hypoglycemia (84%) [26][27][28][29][30]35,36,[38][39][40][41][42][44][45][46][47] supposed the event took place in an out-of-hospital setting.…”

“…The remaining 3 studies (16%) [31,33,54] supposed hypoglycemia occurring in an in-hospital setting. Of the 14 studies that detected hypoglycemia, 11 studies (79%) [15,20,25,32,43,[48][49][50][51][52]55] detected hypoglycemia in an experimental setting, where hypoglycemia was induced by a hypoglycemic clamp procedure. In 20 of the 33 included studies (61%) [15,20,25,27,29,31,32,35,36,38,41,[43][44][45][49][50][51][52]54,55]), a hold-out method was used to separate the information in the database according to training and test data.…”

“…In the majority of the 19 studies for predicting hypoglycemia (13 studies; 68%) [26][27][28][29][30]35,36,38,[40][41][42]46,47], historical CGM data were input into the ML algorithm while the remaining 6 studies (32%) [31,33,39,44,45,54] did not use CGM. Of the 14 studies that detected hypoglycemia using ML, 7 studies (50%) [20,25,32,49,50,52,55] used information from electroencephalograms (EEGs) and 4 studies (29%) [15,43,51,53] used results of electrocardiography (ECG).…”