Calibration of the Photoplethysmogram to Arterial Blood Pressure: Capabilities and Limitations for Continuous Pressure Monitoring

Abstract

It is well known that the photoplethysmogram (PPG) is related to the underlying arterial blood pressure (ABP). However, there is ambiguity regarding the precise relationship of these two circulatory signals: some investigators have described a static relationship, while others have used more complex dynamic characterizations. This paper attempts to reconcile these models, employing results obtained from controlled human subject tests. Our data indicate that there is no simple relationship between the continuous PPG and ABP. However, when several serial beats are aggregated, a static, reproducible relationship is observed. This suggests that, in a stable experimental environment, most of the factors that decouple the ABP and PPG operate over short time scales. When these factors are eliminated through beat averaging, calibrated plethysmography offers a near-continuous estimation of ABP. Our data suggest that this PPG/ABP relationship is not stationary, since it is less consistent over time scales greater than 20 minutes. Variable local ABP was generated by raising and lowering the subjects' hands; this method also offered the means for oscillometric BP measurement without any additional actuator such as a conventional pneumatic cuff.

Continuous and Noninvasive Blood Pressure Measurement: A Novel Modeling Methodology of the Relationship Between Blood Pressure and Pulse Wave Velocity

“…The model coefficients are identified by individual calibration. 1,3,4,6,7,[9][10][11][12]15 Theoretical modeling of the relationship between BP and PTT/PWV is a crucial part in the methodology above. So far, the proposed theoretical models fall into two categories: linear models and nonlinear models.…”

Data-driven estimation of blood pressure using photoplethysmographic signals

“…PPG signal and discrete wavelet transform PPG and arterial blood pressure are coupled in mid-term signals, meaning between four to thirteen heartbeats [16], [5]. This observation hints at an optimal sampling window of twenty to ninety seconds.…”

“…The placement of the PPG device is crucial: finger BP may be different from arterial BP in the arm [6], but strong correlation is apparent if the circumstances are given -for instance, during sleep [7]. Even given a high correlation with the BP wave, explicit mathematical models get complex as they attempt to incorporate free parameters, such as arm position and motion [5]. Machine learning methods address this problem by automatically extracting the relationship between BP and the PPG waveform.…”

“…Peripheral volumetric changes and arterial BP are linked [4], but there is no simple, continuous relationship [5]. The placement of the PPG device is crucial: finger BP may be different from arterial BP in the arm [6], but strong correlation is apparent if the circumstances are given -for instance, during sleep [7].…”

“…Until more data become available, sampling is restricted to a calm, seated position. The arm position relative to the heart is especially important [9], [5], [12]. As measurements leave a clinical setting and smartphones enable PPG recording, patients need more guidance to obtain a clean signal, as even the contact force with the sensor matters [13].…”