Data collection and system design

In total, 10 young healthy volunteers (25 ± 4 years; gender: 6 males and 4 females; weight: 63 ± 8; height: 165 ± 6 cm) with no histories of hypertension or cardiovascular disease were enrolled in our study. Written informed consent was obtained from participants after we had provided them with a complete description of the study. This study was approved by the ethics committee of Pukyong National University and conducted according to the Declaration of Helsinki ethical principles for medical research on human subjects.

In each experiment, the subject followed the same measurement setup and protocol as shown in Fig. 1. Four physiological waveforms were measured simultaneously during the experiments. Our proposed multimodal biosensor was worn on the left wrist of each subject. First, we used the IPG sensor to measure the radial impedance of the radial artery from the inner part of the subject wrist. Second, we employed the PPG sensor for the arterial blood volume pulsation measurement in the area of interosseous arteries from the outer part of the subject wrist (Fig. 1a). Third, an ECG sensor located on the subject chest was used to record the electrical activity of the heart of each subject, which acted as the proximal waveform in the PAT calculation. Lastly, we utilized Oscar 2 (SunTech Medical, USA) for BP reference data acquisition (Fig. 1b). This device is considered the gold standard in 24-hour ambulatory BP monitoring and is certified by the British Hypertension Society (BHS) and the AAMI-SP10 standards.

(a) Multimodal biosensor with an impedance plethysmography (IPG) signal acted as a proximal waveform, and a photo plethysmography (PPG) signal acted as a distal waveform. (b) Measurement setup with chest electrocardiogram (ECG), wrist PPG, and wrist IPG collected simultaneously, and left-arm blood pressure (BP) monitoring with Oscar 2 for BP reference data acquisition. (c) Measurement protocol with three rest sessions (R1, R2, R3) and three arm-exercise sessions (E1, E2, E3) to increase BP.

The details of the proposed multimodal biosensors system are given as follow. The proposed prototype device is attached to the back of the wrist strap and directly contacts the subject’s skin. For the PPG sensor, the overall design was described in our previous report¹². In brief, we use four LEDs to transmit light in different optical wavelengths (in both visible and near-infrared optical spectra) ranging between 660–940 nm. However, only the PPG signal collected from 940 nm is used for analysis, multi-channel PPG signals can be used for heart rate (HR) monitoring during intensive exercise, the motion artifacts reduction algorithm from multi-channel PPG signal can be utilized to separate HR signal and the motion artifacts in the time domain. Moreover, the remaining signal is intended for another clinical experiment such as peripheral capillary oxygen saturation monitoring¹³.

The size of the prototype PPG sensor is 15 × 15 mm² and we utilize a photodiode (PD) from BPW 34 (OSRAM Semiconductor Inc.) to detect reflected light from the subject wrist skin. Moreover, the proposed IPG sensor consists of four conductive silver-plated polyester electrodes that are flexible, stretchable in both directions, and suitable for long-term healthcare monitoring¹⁴. Each electrode is sewed to the strap and acts as a voltage detector and current injector. In this study, we used 500 μA from a 100-kHz current source, which is considered safe for the human body and sufficiently assists in detecting the impedance variance¹⁵. In addition, a three-channel wet electrode with an AD8232 (Analog Devices, Inc., USA) single-lead front end was used for ECG signal collection.

Our experiments were performed repeatedly in an indoor laboratory from April to June 2018 under an ambient temperature controlled at 25 °C and a relative humidity of 60%. For data collection with each subject, a recorded signal was started approximately 5 s before the cuffed BP measurement was taken using Oscar 2 and finished approximately 1 min after the cuff was released from the reference BP device. This protocol was conducted during the three arm-exercise intervention sessions, as shown in Fig. 1b. Our study employed an arm exercise to perturb BP because it was proven easier to perform and to increase subject BP value in an effortless manner. Every subject was instructed to sit relaxed to obtain the first session data (B1). Male and female subjects then performed arm exercises (E1) with 3-kg and 1-kg dumbbells (or handgrips), respectively. The next R2 to E3 followed a similar protocol. Each session was conducted for 20 min and three BP signals were collected every session and averaged for the session BP value.