Event Abstract

A Model for Diabetic Blood Glucose Prediction Based on Electroencephalography Signals Using Deep Learning

  • 1 Başkent University, Ekin Pre-Incubation Center & Yucelen Group, Turkey
  • 2 Başkent University, Vocational School of Technical Sciences, Turkey
  • 3 Başkent University, General Surgery Department, Turkey
  • 4 Stuttgart, Germany
  • 5 Başkent University, Ekin Pre-Incubation Center, Turkey

In the last decade, with the emerge of Deep Learning concept applying machine learning to big data has become possible. Unlike shallow neural networks which require pre-processing techniques for feature extraction, Deep Learning is capable of automatic feature extraction. Therefore, Deep Learning can be used for classification of big raw data such as physiological signals. One of the main interests in neuroergonomic and neuroadaptive is electroencephalography (EEG) signal processing and classification. Electroencephalography is related to various vital and mental indicators. Change of these indicators directly affect Electroencephalography signals. Conversely processing the electroencephalography signals, clues on the related indicators can be gained. One of the related indicators is blood glucose since it is main source in physiologic conditions. Although brain is approximately %2 of body weight, %20 of glucose-derived energy is consumed in brain. Therefore electroencephalography signals can have potential for the classification of blood glucose status. In this study Deep Learning is proposed to be used to classify electroencephalography signals as hyperglycemia, hypoglycemia and normal status. This provides continous monitoring of blood glucose in varying conditions using electroencephalography signals. Fluctating levels of blood glucose are critical to diagnose diabetes which is a condition of an incapability of regulating blood glucose levels and carbohydrate metabolism in human body. Hormones such as insulin and glucagon are exreted by pancreas in order to reciprocate the blood glucose levels. It is a common knowledge that hyperglycemia is related to long-term devastation and function disorder of the nerves [1]. Hypoglycemia can usually be defined as unnaturally low blood glucose level and is the most prevalent but hazardous complication of insulin therapy carried out intensively for patients who have type 1 diabetes mellitus (T1DM) [2,3]. T1DM is qualified by glucose level variations ranging from hypoglycemia to hyperglycemia . In order to execute diabetes therapy, hemoglobin (HbA1c) range and inadequecy of hypoglycemic and hyperglycemic episodes can be maintained by monitoring glucose level frequently, management of insulin and balance of carbohydrate [4]. In addition to these methods, diabetic blood glucose fluctuations can be predict by examining EEG signals. EEG is conventionally measured using surface electrodes which are placed on scalp of the patients and capable of measuring the electrical activity in the brain which represents metabolic status of the brain cells [2]. Brain needs glucose as a primary source of energy. Low levels of blood glucose devistate brain function by impressing ionic currents in the brain neurons. Alterations in these ionic currents produce voltage oscillations which can be determined by EEG [5]. Acute changes in glucose consentration and mental fluctuations in diabetes patients which can cause hyperglycemia might also be detected by using EEG signals [4]. In order to estimate blood glucose levels with unconventional methods, various type of studies were presented. One of these studies suggests a non-invasive estimation of blood glucose and blood pressure simultaneously by using a photoplethysmograph sensor via machine learning techniques [6]. In addition, another application also proposes a non-invasive detection method for patients with T1DM by using fuzzy reasoning model. Developed system is able to detect the presence of hypoglycemic episodes based on corrected QT interval of electrocardiogram (ECG) signal and the heart rate [7]. The aim of this study, differently from these studies, is to correlate between diabetic blood glucose and EEG signals by creating a novel model in order to predict diabetic blood glucose based on EEG signals using deep learning. Deep learning, as one of the most currently notable machine learning technique, has achieved prosperity in many applications such as image and speech recognition, also text minnig. It uses supervised and unsupervised strategies to learn multi-level methods and features in hierarchical architectures for the tasks of classification. Recent development in EEG Signal Processing technologies has enabled the collection of big data. Measurement periods and data magnitudes necessitate big data processing. Therefore classification of big data shall be performed by Deep Neural Network, an advanced type of artificial neural network based methodology. In the past few years, deep learning has played an important role in big data analytic solutions [8].

Figure 1


[1] Jens B. Frøkjær, Carina Graversen, Christina Brock, Ahmad Khodayari-Rostamabad, Søren S. Olesen,Tine M. Hansen, Eirik Søfteland, Magnus Simrén, Asbjørn M. Drewes, Integrity of central nervous function in diabetes mellitus assessed by resting state EEG frequency analysis and source localization, Journal of Diabetes and Its Complications 31 (2017) 400–406

[2] Lykke Blaabjerg, Claus B. Juhl, Hypoglycemia-Induced Changes in the Electroencephalogram: An Overview, Journal of Diabetes Science and Technology 2016, Vol. 10(6) 1259–1267

[3] Lien B. Nguyen, Anh V. Nguyen, Sai Ho Ling,Hung T. Nguyen, Analyzing EEG Signals under Insulin-induced Hypoglycemia in Type 1 Diabetes Patients, 35th Annual International Conference of the IEEE EMBS Osaka, Japan, 3 - 7 July, 2013

[4] M. Rachmiel, M. Cohen, E. Heymen, M. Lezinger, D. Inbar, S. Gilat , T. Bistritzer,G. Leshem, E. Kan-Dror, E. Lahat, D. Ekstein, Hyperglycemia is associated with simultaneous alterations in electrical brain activity in youths with type 1 diabetes mellitus, Clinical Neurophysiology 127 (2016) 1188–1195

[5] Fabio Scarpa, Maria Rubega, Mattia Zanon, Francesca Finotello,Anne-Sophie Sejling, Giovanni Sparacino, Hypoglycemia-induced EEG complexity changes in Type 1 diabetesassessed by fractal analysis algorithm, Biomedical Signal Processing and Control 38 (2017) 168–173

[6] Enric Monte-Moreno, Non-invasive estimate of blood glucose and blood pressure from a photoplethysmograph by means of machine learning techniques, Artificial Intelligence in Medicine 53 (2011) 127– 138

[7] Sai Ho Ling∗, Hung T. Nguyen, Natural occurrence of nocturnal hypoglycemia detection using hybrid particle swarm optimized fuzzy reasoning model, Artificial Intelligence in Medicine 55 (2012) 177–184

[8] Zhang, Q., Yang, L., Chen, Z., Li, P., “A survey on deep learning for big data”, “Information Fusion”, Volume: 42, Pages: 146-157, DOI: 10.1016/j.inffus.2017.10.006, Published: JULY 2018.

Keywords: deep learning, Electroencephalography (EEG), Hypoglycemia, Hyperglycemia, Correlation analysis

Conference: 2nd International Neuroergonomics Conference, Philadelphia, PA, United States, 27 Jun - 29 Jun, 2018.

Presentation Type: Oral Presentation

Topic: Neuroergonomics

Citation: Berkol A, Karayegen G, Tartan EO, Ekici Y, Kara G and Eser Z (2019). A Model for Diabetic Blood Glucose Prediction Based on Electroencephalography Signals Using Deep Learning. Conference Abstract: 2nd International Neuroergonomics Conference. doi: 10.3389/conf.fnhum.2018.227.00025

Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.

The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.

Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.

For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.

Received: 22 Mar 2018; Published Online: 27 Sep 2019.

* Correspondence: Mr. Ali Berkol, Başkent University, Ekin Pre-Incubation Center & Yucelen Group, Ankara, Turkey, aberkol@baskent.edu.tr