包括电子皮肤在内的柔性传感技术的革新以及人工智能的广泛应用，使得利用可穿戴设备监测人体生理信号并进行诊断成为了可能。自诞生以来，可穿戴设备的功能逐渐由通讯、娱乐、运动监测等领域向健康、医疗等方向扩展。另一方面，中医可以通过三根手指采集脉搏波进而实现疾病诊断，这一脉诊过程往往需要医生长期丰富的经验积累、较为主观化。本研究的目的在于结合可穿戴传感技术与机器学习方法，模拟中医脉诊的过程，通过现代传感技术揭示脉诊原理，开发传感器采集脉搏波形，使其数字化、可视化和客观化，进而用于健康状况监测与疾病诊断。 针对可穿戴移动健康设备全天候连续监测的特点，以及中医脉诊的实际应用场景，本研究提出了基于三明治结构的压电驻极体柔性动态压力传感器。该传感器具有自驱动、高灵敏度、可工作在大静态压力范围的特点，满足了中医脉诊中施加浮、中、沉不同压力的要求，同时也满足了可穿戴设备对传感器低功耗的需求。首先，通过仿真与实验分析了传感器结构的几何参数和制备工艺对其稳定性与灵敏度等性能的影响。传感器采用了驻极体-柔性间隔层-驻极体的多层结构，得益于驻极体材料保持静电荷的能力以及间隔层良好的柔性，等效压电系数可达到4100 pC/N，灵敏度达到32.6 nA/kPa（低静态压力下）和6.71 nA/kPa（高静态压力下）。 其次，本研究对比了柔性压力传感器与目前可穿戴市场心率监测应用最广泛的光学传感器，在大范围静态压力（20 mmHg至120 mmHg）作用下，采集到的脉搏波形，分析并得出结论：光学传感器在非接触和低静态压力条件下可以较为准确地提取容积脉搏波，但在高静态压力下，信噪比降低，无法刻画出具有细节的脉搏波形；而基于压电驻极体的传感器在模仿中医脉诊的应用场景下具有良好性能。 最后，本研究也展示了所设计开发的基于压电驻极体的脉搏波传感器在健康监测方面的具体应用，包括：心律不齐的监测，非侵入式人体动脉血压的估计，呼吸与心率信号的同时提取，利用脉搏波近似熵反映健康状况，脉搏波作为生物识别信号实现分类识别，以及模拟中医脉诊的“寸、关、尺”三路脉搏波采集等。 本研究为中医脉诊的客观化提供了可行的硬件基础，并探索了多样化的信号处理方法，在可穿戴健康监测领域具有重要的研究意义和实用价值。
Over the past few years, the innovations in sensing technologies such as the electronic skin, and the prevalent implementation of artificial intelligence have become strong foundations for monitoring human physiological signals to make the diagnosis with wearable devices. On the one hand, the functions of wearable devices have gradually expanded from communication, entertainment, and sports to health monitoring and medical assessment. On the other hand, in traditional Chinese medicine (TCM), doctors use three fingers to collect pulse wave and then make the diagnosis, which often requires accumulated long-term experience and is highly subjective. The goal of this work is to use the combination of wearable sensors and big data analytics to unveil the process and mechanism of pulse palpation, which has been hard to interpret so far, to make possible diagnostics of human pulses to imitate the TCM practice for health assessments without well-trained doctors. and to provide a new idea about vital signs monitoring for mobile health, which is becoming a popular tool for efficient and convenient medical services. Real-time and continuous monitoring of physiological signals using sensors with low power consumption is essential for mobile health. For the specific application scenarios of TCM pulse palpation, a pressure sensor embracing high sensitivity as well as a large range is desired to solve the major problem of various holding down levels from superficial, middle to deep. A flexible piezoelectret sensor with FEP/Ecoflex/FEP sandwich structure is proposed in this work, which is self-powered, highly sensitive under a large range of static pressure and thus meets both requirements for wearable devices and TCM pulse sensing. Firstly, the structure of the sensor is introduced along with the theoretical analysis of the output performance. The simulation is conducted and compared with the results of the experiment to show the influence of the geometrical parameters and material properties on the sensitivity and stability of the device. The equivalent piezoelectric coefficient of the FEP/Ecoflex/FEP sandwich structure can reach 4100 pC/N, which is attributed to the abundant electrical dipoles sealed in the formed cavity and the low modulus of the Ecoflex layer. The sensitivity of the piezoelectret sensor with the optimized design can be as high as 32.6 nA/kPa (at static pressure lower than 6 kPa) and 6.71 nA/kPa (at static pressure higher than 6 kPa). Secondly, the performance of pulse wave acquisition under different static pressure (from 20 mmHg to 120 mmHg) is compared between the proposed piezoelectret sensor and the commercial optical sensor, which is the most widely adopted solution for heart rate monitoring in wearable devices. The analysis of the results in both time and frequency domain demonstrates that the piezoelectret sensor is capable to acquire the detailed waveforms under the large range of static pressure, while the optical one fails under high pressure. Thus the piezoelectret sensor is superior to optical one in the application scenario of imitating the TCM pulse palpation where high pressure is an important and indispensable factor. Finally, several applications are demonstrated to illustrate the possible medical assessments using the pulse sensing system, including the diagnosis of arrhythmia, the estimation of blood pressure, the simultaneous monitoring and separation of respiration and heart rate, and the health condition assessment using approximate entropy analysis on pulse wave. A three-channel pulse wave sensor mimicking the TCM doctor’s three fingers corresponding to Cun, Guan, and Chi is implemented and tested. To sum up, this work provides feasible hardware for the objectification, digitalization, and visualization of pulse palpation and diagnosis process in TCM. Diverse data analysis methods on pulse wave signals are also explored, indicating broad application prospects of the piezoelectret sensor in the field of wearable health monitoring.
