TY - GEN
T1 - Strain Plethysmography at the Radial Artery
T2 - 2023 IEEE Biomedical Circuits and Systems Conference, BioCAS 2023
AU - Shokouhmand, Arash
AU - Jiang, Xinyu
AU - Ayazi, Farrokh
AU - Ebadi, Negar
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - This study introduces a novel framework based on radial strain plethysmography (SPG) for cuffless estimation of blood pressure (BP). For this purpose, a sensitive, custom-built micro-electromechanical systems (MEMS) strain sensor is used to capture pulse information at the radial artery. A comprehensive feature set consisting of the timing intervals between the fiducial points of the SPG segments, their derivatives, and integrals was extracted and concatenated with the demographic information of the subjects. An XGBoost regressor is then trained on the extracted features to estimate the systolic and diastolic blood pressures, SBP and DBP, respectively. The model is evaluated on 15 subjects using a 5-cross-validation (5-CV) approach. Mean error (ME), mean absolute error (MAE), and root mean square error (RMSE) of -0.22, 3.59, and 5.12 mmHg, respectively, are reported for the estimated DBP values, and ME, MAE, and RMSE of -1.00, 6.23, and 8.96 mmHg, respectively, are achieved for SBP estimation. It is also indicated that significant features for BP estimation encompass both demographic information and timing intervals, highlighting their combined importance. The results demonstrate that radial SPG is a reliable alternative to optical-based technologies for accurate BP estimation.
AB - This study introduces a novel framework based on radial strain plethysmography (SPG) for cuffless estimation of blood pressure (BP). For this purpose, a sensitive, custom-built micro-electromechanical systems (MEMS) strain sensor is used to capture pulse information at the radial artery. A comprehensive feature set consisting of the timing intervals between the fiducial points of the SPG segments, their derivatives, and integrals was extracted and concatenated with the demographic information of the subjects. An XGBoost regressor is then trained on the extracted features to estimate the systolic and diastolic blood pressures, SBP and DBP, respectively. The model is evaluated on 15 subjects using a 5-cross-validation (5-CV) approach. Mean error (ME), mean absolute error (MAE), and root mean square error (RMSE) of -0.22, 3.59, and 5.12 mmHg, respectively, are reported for the estimated DBP values, and ME, MAE, and RMSE of -1.00, 6.23, and 8.96 mmHg, respectively, are achieved for SBP estimation. It is also indicated that significant features for BP estimation encompass both demographic information and timing intervals, highlighting their combined importance. The results demonstrate that radial SPG is a reliable alternative to optical-based technologies for accurate BP estimation.
KW - MEMS
KW - Strain plethysmography (SPG)
KW - blood pressure (BP)
KW - demographic information
KW - radial artery
KW - timing intervals
UR - http://www.scopus.com/inward/record.url?scp=85184978471&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85184978471&partnerID=8YFLogxK
U2 - 10.1109/BioCAS58349.2023.10388913
DO - 10.1109/BioCAS58349.2023.10388913
M3 - Conference contribution
AN - SCOPUS:85184978471
T3 - BioCAS 2023 - 2023 IEEE Biomedical Circuits and Systems Conference, Conference Proceedings
BT - BioCAS 2023 - 2023 IEEE Biomedical Circuits and Systems Conference, Conference Proceedings
Y2 - 19 October 2023 through 21 October 2023
ER -