Non-invasive blood glucose concentration level estimation accuracy using ultra-wide band and artificial intelligence

Abstract

Diabetes becomes a rapidly increasing global epidemic and getting serious health concern worldwide. There is no remedy except systematic management to keep blood glucose level under control. To achieve that regular glucose level monitoring is a routine task for a patient. This involves collection of blood physically from body with some discomfort and measuring using some device. To overcome this disadvantages and distress, non-invasive blood glucose measurement system is in demand. This article presents an ultra-wide band (UWB) microwave imaging and artificial intelligence based prospective solution to detect blood glucose concentration level non-invasively (without physical blood). The system consists of a pair of small UWB biomedical planar antenna, UWB transceiver as hardware and an artificial neural network with signal acquisition and processing interface as software module. The UWB signal with center frequency of 4.7 GHz was transmitted through ear lobe and forward scattering signals were received from other side. Characteristics features of received signal were extracted for pattern recognition and detection through deep artificial neural network. The system exhibits around 88% accuracy to detect glucose concentration in blood plasma. Besides, it is affordable, safe, user friendly and can be used with comfort in near future.

Feasibility Study of Glucose Concentration Measurement of Aqueous Solution Using Time Domain Reflected Signals

“…These days, electromagnetic waves have been used for various medical applications, including blood glucose measurement [ 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 ], skin cancer [ 13 ], and breast cancer [ 14 ]. Blood glucose measurement has been researched in different frequency ranges, from radiofrequency [ 11 ] to the infrared range [ 12 ].…”

“…In light of further measurement errors, this sensitivity is very weak. Furthermore, performing WBMS using co-planar and planar sensors has been introduced as a suitable approach for detecting glucose variation recently [ 3 , 7 , 19 ]. In all of these researches, the material under test (MUT) has to be in direct contact with the sensor.…”

“…However, some convenient setups were proposed recently in both FD and TD measurements [ 16 , 20 , 21 , 22 , 23 ]. A non-invasive time-domain WBMS system based on an artificial neural network for detecting glucose content of blood was proposed in [ 7 ]. A pair of custom-made ultra-wideband (UWB) rectangular patch bio-antennas were placed on a human earlobe, and TD transmission signals with a central frequency of 4.7 GHz were measured using a transceiver [ 7 ].…”

“…A non-invasive time-domain WBMS system based on an artificial neural network for detecting glucose content of blood was proposed in [ 7 ]. A pair of custom-made ultra-wideband (UWB) rectangular patch bio-antennas were placed on a human earlobe, and TD transmission signals with a central frequency of 4.7 GHz were measured using a transceiver [ 7 ]. Complex processing and a large duration of measurement are the weaknesses of this system.…”

Non-Invasive Glucose Monitoring Using Optical Sensor and Machine Learning Techniques for Diabetes Applications

“…The complexity of non-invasive glucose measurements necessitates the use of more effective algorithms capable of dealing with higher dimensional data in order to improve the accuracy of non-invasive glucose concentration measurements. Various machine learning techniques are used for glucose predictions, such as multiple linear regression [21], partial least square regression [22], [23], principal component regression [22], feed forward neural networks [24], [25], deep neural networks [26], [27], support vector machines [27], [28], random forest regression [18], etc. These techniques are used to build regression based models that predict continuous values of glucose concentration.…”