The Estimation of Brachial-ankle Pulse Wave Velocity (baPWV) Based on the Acceleration Plethysmogram

• Published in: Transactions of Japanese Society for Medical and Biological Engineering Vol. 59; no. 1; pp. 47 - 53
• Main Authors: HIROSE, Kumi, WAKABAYASHI, Satoshi, ONO, Takuya, HIRAHARA, Makoto, YANA, Kazuo
• Format: Journal Article
• Language: Japanese
• Published: Japanese Society for Medical and Biological Engineering 10.03.2021
• Subjects: acceleration plethysmography; health monitoring, ubiquitous healthcare; photoplethysmography; pulse wave velocity
• ISSN: 1347-443X; 1881-4379
• DOI: 10.11239/jsmbe.59.47

This paper proposes a method to estimate brachial-ankle pulse wave velocity (baPWV) based on parameters extracted from acceleration plethysmogram (APG), the second derivative of photoplethysmography (PPG). The baPWV is to measure the arterial stiffness and clinically utilized to probe the presence of atherosclerosis. The measurement is made by the time delay between brachial and ankle pule waves at the clinical bench in a hospital. To explore the possibility to monitor the baPWV in daily basis at home, we examined the relation between baPWV and APG waveform. The simultaneous recordings of baPWV and PPG are made for 74 healthy subjects aged 30-69 (47.3±8.8) in supine posture. We have extracted five parameters from the APG, b/a, c/a, d/a, e/a and Tae, in which a-e are the extreme values of APG and Tae is the time interval between a and e.The new parameter estimation method using the third derivative of PPG waveform is introduced for the accurate parameter derivation, namely b, c and d which sometimes fuse each other to make the separation difficult in APG. Result showed the parameter Tae is the most important index for the estimation. The correlation coefficient between baPWV and Tae showed 0.569 (p＜10－3). With additional parameter d/a, the prediction of baPWV by the multiple regression showed corrected coefficient of determination of 0.430. Setting the risk criteria of arteriosclerosis be baPWV＞1.45 (M/S) the proposed prediction scheme achieved the risk detection rate of 0.917 with the false detection rate of 0.161. Although the estimation accuracy needs further improvement for the clinical use, the method will be utilized for the daily health monitoring for the early detection of the change in arterial stiffness.