Measuring Blood Pressure with a Smartphone Camera

Chronically high blood pressure (BP) or hypertension is the key risk factor for cardiovascular diseases - the primary cause of death globally, with approximately ten million deaths annually. By 2025 it is predicted that there will be almost 1.5 billion hypertensive people around the world.

The key to addressing this epidemic is an urgent need for accurate, accessible screening and monitoring of blood pressure to ensure the early detection of hypertension.

One potential solution to this issue lies with smartphones - readily available devices for mobile health, with over one third of global consumers owning one already.

Smartphone cameras offer the potential to measure a photo-plethysmographic (PPG) signal at the fingertip, and, when coupled with recent advances in PPG-based BP monitoring,¹ the potential of smartphones for hypertension prevention, diagnosis and management is clear.

These ubiquitous devices could play a leading role, but it is first important to successfully demonstrate their ability to provide accurate blood pressure measurements.

This article outlines a validation study performed on 85 patients from a hypertension clinic, whereby the pulse wave analysis algorithm - oBPM^® - was used to provide estimated blood pressure by using a smartphone camera to obtain optical signals acquired from patients’ fingertips.

The results are compared to auscultatory measurements to highlight compliance with the ISO 81060-2 standard for non-invasive sphygmomanometers.

The patented oBPM^® optical blood pressure monitoring algorithm² was recently validated during general anesthesia induction using a series of PPG signals from a standard pulse oximeter.³

The goal of the study outlined here was to assess the accuracy of oBPM^® applied to smartphone-derived PPG signals acquired via a dedicated app (OptiBP^®).

The study⁴ was comprised of two arms (ClinicalTrials.gov identifier: NCT03875248). The first arm involved training the algorithm’s parameters using data from 51 patients at the Geneva or Lausanne (CHUV) University Hospitals in Switzerland that were scheduled for an elective surgery that required general anesthesia and invasive arterial BP monitoring.

The decision to use general anesthesia induction in this instance was due to its tendency to prompt significant BP variations, therefore, enabling thorough training of the algorithm to ensure sensitivity.

The second arm of the study saw the algorithm used to estimate BP values of 85 patients scheduled for an elective visit at CHUV’s outpatient hypertension clinic. These were then compared to reference auscultatory measurements acquired by two independent observers via a double stethoscope.

Seven pairs of measurements were acquired for each of the validation group’s 85 patients. An initial calibration procedure was applied for each patient. This consisted of an offset correction with any unreliable BP estimates being automatically rejected.

The accuracy of the BP estimates was evaluated against the reference auscultatory measurements.

This evaluation used metrics from the ISO 81060-2 standard for non-invasive sphygmomanometers, which stipulates that the cohort-wise mean error must be within ±5 mmHg with standard deviation greater than 8 mmHg.

Auscultatory measurements only provide systolic and diastolic BP values, meaning that reference mean BP values were obtained as 2/3 diastolic + 1/3 systolic. Table 1 displays these results whilst also highlighting compliance with the ISO 81060-2 standard for each BP value.

Table 1. Accuracy (mean error and standard deviation of the error) of the smartphone-derived BP values estimated by the oBPM^® algorithm compared to reference auscultatory measurements. Source: CSEM

Enhancing the accessibility of BP measurements on a global scale is an essential step in reducing the mortality and morbidity linked to hypertension.

Since smartphones are so widely used around the world - including in developing and low-income countries – they present an ideal solution to this issue.

A clinically acceptable level of accuracy must be met to make this feasible. This has been confidently demonstrated in this study of 85 patients where BP estimates were found to be compliant with the ISO 81060-2 standard.

References

1. M. Proença, P. Renevey, F. Braun, et al., "Pulse Wave Analysis Techniques". In: J. Solà, R. Delgado-Gonzalo (eds) The Handbook of Cuffless Blood Pressure Monitoring. Springer, Cham, 2019.
2. M. Proença, J. Solà, M. Lemay, C. Verjus, "Method, apparatus and computer program for determining a blood pressure value", WO 2016138965 A1, 9th of September 2016.
3. M. Proença, Y. Ghamri, G. Hofmann, et al., "Automated pulse oximeter waveform analysis to track changes in blood pressure during anesthesia induction: a proof-of-concept study", Anesthesia & Analgesia, 130(5):1222-1233, 2020.
4. P. Schoettker, J. Degott, G. Hofmann, et al., "Blood pressure measurements with the OptiBP^® smartphone app validated against reference auscultatory measurements", manuscript in prep., 2020.

Acknowledgments

M. Proença, G. Bonnier, A. Lemkaddem, and M. Lemay from the Swiss Center for Electronics and Microtechnology; J. Degott, G. Hofmann and P. Schoettker from the Department of Anesthesiology, Lausanne University Hospital and University of Lausanne, R. Schorer from the Department of Acute Medicine, Geneva University Hospital and University of Geneva; U. Christen, J.F. Knebel, from Biospectal SA; and A. Wuerzner, M. Burnier, G. Wuerzner from Service of Nephrology and Hypertension, Lausanne University Hospital and University of Lausanne.

This information has been sourced, reviewed and adapted from materials provided by CSEM.

For more information on this source, please visit CSEM.