Mosaddequr Kazi, Rahman Tanzilur
Department of Electrical & Computer Engineering, North South University, Dhaka 1229, Bangladesh.
Heliyon. 2023 Sep 1;9(9):e19553. doi: 10.1016/j.heliyon.2023.e19553. eCollection 2023 Sep.
We propose a completely non-invasive and highly accurate portable blood glucose estimator, which is simple hardware that anyone, regardless of their prior experience or knowledge, can use to obtain an immediate reading of their blood sugar level. Glucose levels can be monitored in real time and displayed on the device thanks to its infrared LED light source, sensor with built-in amplification unit, processing unit with blood glucose regression model, power management unit for autonomous operation, and display. The device was initially used to create a dataset of photoplethysmography (PPG) signals collected from the fingertip of 50 subjects. The extracted signal features were correlated with the subject's glucose level, which was measured at the same time using a commercial glucometer, and several regression models were constructed. The models were evaluated using signals from up to 110 subjects across three datasets, and the most optimized model was implemented in the device to predict the subject's blood glucose level solely based on PPG in real-time. The device with the built-in model has been subjected to extensive testing to gauge its efficacy. The device's clinical accuracy is encouraging. The pricey strips and needles that must be purchased along with the hardware in the conventional method will no longer be necessary with this device.
我们提出了一种完全非侵入性且高度准确的便携式血糖估算器，它是一种简单的硬件设备，任何人无论其先前经验或知识如何，都可用来立即获取自己的血糖水平读数。借助其红外LED光源、带有内置放大单元的传感器、具有血糖回归模型的处理单元、用于自主运行的电源管理单元以及显示屏，血糖水平能够被实时监测并显示在该设备上。该设备最初用于创建一个从50名受试者指尖采集的光电容积脉搏波描记图（PPG）信号数据集。提取的信号特征与同时使用商用血糖仪测量的受试者血糖水平相关联，并构建了多个回归模型。使用来自三个数据集中多达110名受试者的信号对模型进行评估，最优化的模型被应用于该设备中，以仅基于PPG实时预测受试者的血糖水平。内置该模型的设备已经过广泛测试以评估其功效。该设备的临床准确性令人鼓舞。使用这种设备，就不再需要与传统方法中的硬件一起购买昂贵的试纸和针头了。
