Computational Study of Dual Sensor Design For Noninvasive Diabetics Detection Device Based on Infrared And Ultrasound PIEZOELECTRIC MEMS Technologies

ABSTRACT

Continuous and noninvasive method of monitoring and regulating the blood glucose level in diabetic patients is essential to avoid complications that may result from the condition. Over the last few years, efforts have been made to replace the existing invasive method of glycemic monitoring with the novel noninvasive approach by taking advantage of recent developments in biosensors and enhanced computing power of microprocessors. This work sought to explore the combined technologies of Infrared and Ultrasound Microelectromechanical Systems (MEMS) for glucometer sensor design in view to address the clinical inaccuracies and reliability challenges encountered by sensors. Combined technology approach allows the possibility to compensate aberration of one modality by another. The method for optimizing transducers for sensor design has traditionally been through trial and error guided by one-dimensional simulations. This is costineffective and time-inefficient. However, carrying out simulation for the design of a device allows tweaking of parameters for optimal design with least amount of materials and efforts. Finite Element Analysis (FEA) is broadly employed in engineering to understand, control and predict the design or operation of a device or physical process. Thus, the modelling and simulation for the Ultrasound MEMS in this work is carried out using FEA in COMSOL Multi-physics, while experimental results reported in existing literatures were adopted to incorporate Near Infrared data for the proposed dual sensor approach. The calibration for the output of the two sensors (Ultrasound and Near Infrared) was implemented using multiple regression in MATLAB software environment. Comparisons were made between the proposed system and actual glucose measured data. The calculated residuals between the actual glucose and the predicted values were found to be marginal (< 0.01%) and the Error Grid Analysis (EGA) plot showed the values fell within the acceptable region (REGION A) of clinical acceptance. Thus, suggesting the performance of the xii proposed dual sensor system as more promising when compared to existing standalone NIR technology design. In terms of material selection, Barium Titanate (BT) showed superior performance as compared to Lithium Niobate (LN) and Barium Sodium Niobate (BNN). The optimized design parameters obtained per the setup utilized in this study were: 0.1142 mm cylindrical thickness, 0.634 mm radius (width) and 2.35MHz operating frequency.