Journal: Advanced Science

Article Title: Health Monitoring via Heart, Breath, and Korotkoff Sounds by Wearable Piezoelectret Patches

doi: 10.1002/advs.202301180

Figure Lengend Snippet: Performance characterizations and physiological signal measurements of a prototype piezoelectret sensor. a) Illustration of the piezoelectret sensor patches at different body locations. b) The schematic diagram of a prototype sensor, showing key aspects such as i) laser processing for FEP grooves; ii) internal electrodes for electromechanical conversion; iii) external electrode for guarding and shielding structure. c) Result of transferred charges versus applied pressure showing the dynamic sensitivity variations between the applied pressure of 0–18 kPa at a constant frequency of 220 Hz. Inset (top): experimental setup for characterizing the dynamic sensitivity of the piezoelectret sensor. Inset (bottom): measured transferred charges as the applied pressure increases from the initial state (P0) to extra + 5 Pa, + 10 Pa, and + 15 Pa, and decreases from extra + 15 Pa, + 7 Pa to the initial state (P0), where P0 is the initial reference pressure chosen arbitrarily in the low linearity region. d) Recorded data from the neck position of a volunteer including: (top) the original data; (bottom) the respiratory wave (0–0.4 Hz), the pulse wave (0–10 Hz), and the audio wave (> 20 Hz) by processing the original data with filters of various frequency ranges. The volunteer is asked to perform activities sequentially to record the normal heart beating, talking, deep breathing, snoring, coughing, and swallowing. e) Enlarged time series (top) and STFT (short‐time Fourier transform) spectrograms (bottom) corresponding to different physiological activities.

Article Snippet: [ ] The 36‐positions of 6×6 sensor array from the same volunteer is further examined using a medical physiological recorder BIOPAC and results show similar characteristics (Figure , Supporting Information).

Techniques:

Journal: Advanced Science

Article Title: Health Monitoring via Heart, Breath, and Korotkoff Sounds by Wearable Piezoelectret Patches

doi: 10.1002/advs.202301180

Figure Lengend Snippet: Measurement and analysis of heart sounds by placing the sensor patch at the heart location. a) Original signal and filtered signals acquired at the heart location. The respiratory wave is the baseline of the original signal without high frequency noises; the pulse wave and heart sound are obtained by using the 10 Hz low‐pass filter and the 20–200 Hz band‐pass filter after removing the baseline drift, respectively. b‐i) Enlarged view of the dotted box in a, showing the S2 split during the inspiration process. ii) Enlarged view of the dotted box in i), highlighting the aortic (A2) and pulmonic (P2) components during the S2 split. c) The STFT spectrum of the original signal (Figure ) within 0–4 Hz, showing the fluctuations of heart rate (HR) and respiration rate (RR). d) The Hilbert spectrum corresponding to the heart sound in b (ii). e) The comparison of the recorded heart sounds from the piezoelectret sensor (HS Sensor) and a medical physiological recorder (HS BIOPAC). The timing of the cardiac cycles is verified by the ECG reference. f) Comparison of the instantaneous cardiac cycles between the piezoelectret sensor (P‐P Sensor ) and i) ECG (R‐R ECG ), and ii) BIOPAC (P‐P BIOPAC ). g) Calculation of the i) systolic period and ii) diastolic period length based on the data recorded by the ECG, piezoelectret sensor, and BIOPAC.

Article Snippet: [ ] The 36‐positions of 6×6 sensor array from the same volunteer is further examined using a medical physiological recorder BIOPAC and results show similar characteristics (Figure , Supporting Information).

Techniques: Comparison

Journal: Advanced Science

Article Title: Health Monitoring via Heart, Breath, and Korotkoff Sounds by Wearable Piezoelectret Patches

doi: 10.1002/advs.202301180

Figure Lengend Snippet: Acquisition and classification of breath sounds. a) The typical sound record from the left chest location during the deep breath is separated to various physiological signals of different frequency bands, including respiratory wave, pulse wave, heart sound, and breath sound. b) Comparison of the breath sounds and STFT spectrum obtained from: a piezoelectret sensor, and a medical recorder BIOPAC. c) Time‐series morphologies, and d) frequency components of the bronchial, bronchovesicular, and vesicular sounds. The original breath sound data is processed by a digital comb filter to suppress the power frequency (50 Hz) component and the high‐order harmonic components. e) Similar breath patterns (n1, n2 for normal breathing and p1, p2 for panting) in the magnitude versus time plot (top) can be distinguished in the MFCCs plot qualitatively (bottom). f) Quantitative analyses by using a parameter ( σ h 2 / σ v 2 ) to quantify the MFCCs similarity between two breath sounds, respectively. g) Confusion matrix of the classification results for three breathing patterns (normal breathing, panting, and snoring).

Article Snippet: [ ] The 36‐positions of 6×6 sensor array from the same volunteer is further examined using a medical physiological recorder BIOPAC and results show similar characteristics (Figure , Supporting Information).

Techniques: Comparison