Technology
February 13, 2025
3 min read

Understanding the Limitations and Accuracy of rPPG Measurements

Remote photoplethysmography (rPPG) has emerged as a non-contact method for monitoring vital signs, revolutionizing the way we approach health monitoring.

Remote photoplethysmography (rPPG) has emerged as a non-contact method for monitoring vital signs, revolutionizing the way we approach health monitoring.

However, as with any technology, understanding its limitations and accuracy is crucial for effective application. This article examines the challenges and advancements in rPPG measurements, specifically in the context of Shen.AI’s SDK.

The rPPG Technology

Shen AI’s technology leverages rPPG, which captures blood volume changes in the skin using advanced computer vision algorithms. This method allows for the estimation of vital signs such as blood pressure and heart rate without direct contact​​. Such a non-invasive approach is particularly valuable in the current healthcare context, where minimizing contact is often necessary.

Challenges in rPPG Measurement

The accuracy of rPPG measurements can be influenced by various factors, including environmental conditions, subject demographics, and technical aspects of the measurement process itself. Two critical factors affecting rPPG accuracy are lighting and stability. For instance, Shen.AI’s engineers have developed algorithms for stabilization and normalization to widen the spectrum of conditions under which accurate measurements can be taken​​.

However, despite these advancements, rPPG methods are subject to biases due to environmental conditions and subject demographics. Datasets often used for developing and testing these methods, such as BP4D and Multi-Pie, tend to lack diversity in subject skin tones and age, and are usually collected under optimal laboratory conditions​​.

Accuracy and Limitations in Diverse Conditions

A study on the biases in rPPG methods reveals that while rPPG methods show agreement with ground truth measurements to some extent, they also exhibit variability. For instance, the mean difference between pulse rates measured by rPPG methods and ground truth was found to be approximately 2%, with a variability of about 15%​​. This variability is significant and can be attributed to the features present in the dataset, such as environmental lighting changes, head movements, and facial marks or headgear, which can lead to incorrect pulse rate prediction​​.

Shen.AI SDK: Enhanced Accuracy and Application

In the context of blood pressure measurement, Shen.AI SDK adheres to the same principles as traditional blood pressure monitors. However, the SDK has been significantly improved to provide more accurate and reliable measurements. Shen.AI’s approach involves continuous measurement over 60 seconds to yield an average value, acknowledging that blood pressure can fluctuate significantly over short intervals​​.

Conclusion

While rPPG technology, as implemented in Shen.AI SDK, offers a promising non-invasive method for health monitoring, understanding its limitations and working towards minimizing biases are crucial for its effective application. Future developments and research should focus on enhancing the accuracy of these measurements, especially across diverse demographics and in varying environmental conditions.

Share this post

More blog posts

Sagittis et eu at elementum, quis in. Proin praesent volutpat egestas sociis sit lorem nunc nunc sit.

Blog
July 30, 2025
5 min read

AI in healthcare: How tech makes healthcare more human

What if AI didn’t replace clinicians — but helped them see more, listen better, and connect deeper? In this expert conversation, we explore how AI is enhancing care by improving diagnosis, enabling prevention, and restoring the human touch in medicine.

Read more
Blog
July 22, 2025
2 min read

Shen AI Vitals is now ANVISA approved

Shen AI Vitals has officially received ANVISA approval, confirming its readiness for use in regulated digital health solutions across Brazil.

Read more
Blog
July 17, 2025
6 min read

Real-time health data: Turn insights into action - webinar recap

See how Shen AI and Thryve are bridging the gap between fragmented data and actionable insights, combining real-time vitals from facial scans with continuous health data to support prevention, population health, and more.

Read more