基于嵌入式系统的穿戴式无创血糖仪研制-仪器科学与技术专业论文.docx

摘要 摘要 I万方数据 I 万方数据 摘要 近年来，糖尿病发病率上升速度加快，目前还没有根治糖尿病的方法，主要是 通过频繁地检测血糖浓度调整降糖药物的用量来预防或减少并发症的发生。目前采 用的手指采血血糖仪有感染的危险，从长远来看，这种做法可能会导致损坏手指组 织。另外，指端采血的血糖仪是一个离散值测量装置，测量时有些高血糖或低血糖 的发生率可能不被记录，因此所得到的监测数据不能完全代表血糖整体状况，因此对 无创血糖检测技术的研究就显得尤为重要。 论文研制了一种基于嵌入式系统的穿戴式无创血糖仪，针对本实验室前期研制 的无创血糖仪的测试精度偏低的情况，在分析血糖与基础代谢率、血氧饱和度、血 流速度等关系的基础上，对实验环境、实验人群以及实验方法都做了系统的规划和 测试。针对现有无创血糖仪便携程度不足的缺点，通过集成多传感器于一体的检测 探头采集能量代谢守恒法无创血糖检测所需的生理信号，实现穿戴式设计。最后搭 建嵌入式平台改善其操作不便，将检测探头检测到的数据通过蓝牙发送至嵌入式平 台，最终将数据通过互联网保存至云数据库。 论文以大量临床试验数据为基础，对各种算法反复对比验证。实验结果显示， 采用随机森林与 BP 神经网络两种算法配合修正后的血糖检测准确率有所提高。与微 创检测方法检测的结果相关性达到了 88.42%，且仪器运行稳定高效。本研究有助于 完善能量代谢守恒法无创血糖检测理论，得出各种人群代谢率、血氧饱和度、心 率、血流速度和血糖的关系，提高无创血糖检测准确率，并有望将该技术应用于临 床，实现无创血糖仪的产业化。 关键词：无创血糖检测；能量代谢守恒法；随机森林；神经网络；嵌入式系统 Ab Abstract II万方数据 II 万方数据 Abstract Recently, the incidence of diabetes reveals rapid increase, and there exists no once- for-all therapy. The regular treatment for minimizing the occurrence of diabetes complications is the frequent glucose test which is to adjust the dose of glucose- lowering drugs. Yet the finger blood glucose meter is at risk of infection. In the long run, it will damage the tissue of fingers. In addition, the value of the fingertip blood glucose meter is supposed to be discrete, leading to missing records of some high blood sugar or hypoglycemia. Thus, the data can’t give a complete report representing the whole situation of blood sugar. Therefore, it is of great importance to investigate the non- invasive blood glucose detection technology. This paper develops a wearable non- invasive glucose meter based on embedded system. For improving the low accuracy of previous noninvasive glucose meter, and deriving from the relationship between the blood sugar and the basal metabolic rate, the blood oxygen saturation as well as the blood flow velocity, the experimental environment, group and methods have been planned systematically. In order to address the inconvenient of carrying the non- invasive blood glucose meter, the paper realizes the wearable design