Non-invasive methodology for detection of Blood Pressure from PPG obtained from the finger of the individual from the smartphone camera.
Introduction:
This analysis presents the findings of a comparison study conducted to evaluate the accuracy and performance of blood pressure (BP) measurements obtained using two different methods: a gold standard BP monitor and PPG (Photoplethysmography) technology-based BP monitoring using a smartphone camera. The study aimed to assess the agreement between the two measurement methods and identify any potential differences or discrepancies.
Methodology:
- Participants: The study involved 388 participants. Participants’ BP was measured using both the gold standard BP monitor and a smartphone application utilising PPG technology.
- Measurement Procedure: BP measurements were obtained sequentially using each method following standard clinical protocols. Measurements were taken in a seated position after a period of rest.
- Data Collection: BP measurements were recorded in mmHg for systolic blood pressure (SBP) and diastolic blood pressure (DBP) using both methods. Each participant underwent three measurements with both devices to ensure reliability.
- Statistical Analysis: The collected data were analysed using statistical methods to assess agreement between the two measurement methods.
The participants in the study were recruited from a diverse demographic background, representing a broad spectrum of age groups, genders, and socioeconomic statuses. The sample size was sufficient to ensure statistical power and generalisability of the findings. The study included participants ranging from young adults to elderly individuals, ensuring representation across different age groups. Additionally, efforts were made to include participants with varying levels of physical activity and lifestyle factors to capture the real-world variability in BP measurements. Overall, the participants’ diverse characteristics contribute to the robustness and relevance of the study findings in evaluating the accuracy and performance of the BP measurement methods in a real-world setting.
The Technology:
Photoplethysmography (PPG) technology, when utilized through a smartphone camera for finger scanning, holds promise for predicting blood pressure (BP) through a non-invasive and convenient method. PPG involves the measurement of changes in blood volume within tissues, typically detected optically. When a finger is placed over the smartphone camera lens, the emitted light penetrates the skin and interacts with blood vessels, resulting in fluctuations in light absorption corresponding to cardiac activity.
Machine learning and artificial intelligence techniques are employed to process the PPG signal and extract relevant features for BP prediction. Our algorithms are trained on large datasets containing PPG signals and corresponding BP measurements obtained from reference devices, such as mercury sphygmomanometers or oscillometric monitors. Through pattern recognition and regression analysis, the algorithms learn to correlate specific features of the PPG signal with BP values, enabling accurate prediction.
The predictive capabilities of this technology using smartphone cameras offer several advantages, including accessibility, portability, and ease of use. Users can perform BP measurements conveniently using their smartphones, without the need for specialized equipment or expertise. Additionally, the non-invasive nature of PPG-based BP prediction enhances user comfort and compliance, facilitating regular monitoring of BP levels.
Statistical Analysis:
The statistical analysis of the comparison study between the gold standard blood pressure (BP) monitor and smartphone photoplethysmography (PPG) technology involved several key analyses to evaluate the agreement and accuracy of the two measurement methods.
The mean age of the participants was 39.4± 12.65 years, with a range of 18–60 years. Among the participants, 282 (24.23%) were men and 882 (75.77%) were women.
The mean differences between the CarePlix App and reference device BPs were 1.58 ± 6.41 mm Hg for systolic BP (SBP) and -1.6 ± 4.76 mm Hg for diastolic BP (DBP). The Bland-Altman analysis demonstrated a bias of 1.58 with limits of agreement ranging from -17.32 to 20.48 mm Hg for SBP and a bias of -1.6 with limits of agreement from -16.38 to 13.19 mm Hg for DBP.
Class-wise Accuracy: Accuracy metrics were calculated separately for different BP categories, including hypotensive, normotensive, and hypertensive categories. This analysis allowed for the evaluation of the performance of each method across different BP ranges and clinical scenarios.
Hypotensive (≤110 mmHg for Systolic) : The mean differences between the CarePlix App and reference device BPs in Hypotensive range were -9.02 ± 6.77 mm Hg for systolic BP (SBP) and -3.53 ± 4.06 mm Hg for diastolic BP (DBP). The Bland-Altman analysis demonstrated a bias of -9.02 with limits of agreement ranging from -22.63 to 4.6 mm Hg for SBP and a bias of -3.53 with limits of agreement from -14.32 to 7.26 mm Hg for DBP.
Normotensive (111–129 mmHg for Systolic) : The mean differences between the CarePlix App and reference device BPs in Normotensive range were -1.05 ± 4.26 mm Hg for systolic BP (SBP) and 2.11 ± 4.02 mm Hg for diastolic BP (DBP). The Bland-Altman analysis demonstrated a bias of -1.05 with limits of agreement ranging from -14.88 to 12.78 mm Hg for SBP and a bias of 2.11 with limits of agreement from -10.72 to 14.95 mm Hg for DBP.
Hypertensive (≥130 mmHg for Systolic) : The mean differences between the CarePlix App and reference device BPs in Normotensive range were 8.23 ± 9.66 mm Hg for systolic BP (SBP) and 7.8 ± 6.67 mm Hg for diastolic BP (DBP). The Bland-Altman analysis demonstrated a bias of 8.23 with limits of agreement ranging from -15.06 to 31.53 mm Hg for SBP and a bias of 7.8 with limits of agreement from -8.67 to 24.28 mm Hg for DBP.
Conclusion:
In conclusion, BP measurements using the CarePlix App were consistent with the reference BPs, thereby showing its promise as a non-invasive vitals monitoring for BP self-monitoring.
