Clinical evidence & validation
Evidence behind the measurements
We believe physiological measurements should be evaluated against appropriate reference methods, under clearly defined conditions. Shen AI's measurement technology is supported by clinical studies, defined validation methods and ongoing evaluation across different measurement conditions.
Our evidence covers the performance of individual physiological parameters and provides a framework for customers who want to evaluate Shen AI in their own products and environments.
Clinical studies
Shen AI's technology has been evaluated in clinical studies comparing camera-based measurements with established reference methods.
A clinical trial conducted in Poland in November 2025 included 132 adult participants and evaluated Heart Rate, Heart Rate Variability, Breathing Rate and Blood Pressure across two visits. The study included hundreds of measurements for each parameter and used established reference methods for comparison.
The study design was based on ISO 81060-2:2018 for the Blood Pressure evaluation, as there is currently no dedicated standard for remote, contactless blood pressure measurement.
View the Clinical Trial Report →
Measuring against appropriate references
The reference method matters when evaluating accuracy.
For Heart Rate and Heart Rate Variability, Shen AI recommends comparison with an ECG-based reference. For Blood Pressure, evaluation should use a clinically validated blood pressure monitor.
Consumer wearables such as smartwatches, fitness bands and smart rings can be useful for everyday health tracking, but they should not be treated as ground truth in an accuracy evaluation. They are measurement technologies themselves, and differences between two estimates do not necessarily show that either one is incorrect.
Read the accuracy evaluation guidelines →
What the evidence shows
Clinical evaluation results show strong agreement for Heart Rate and Heart Rate Variability, with performance evaluated against ECG-based reference measurements.
Breathing Rate has also been evaluated against a respiratory reference method, with results demonstrating agreement across different measurement durations.
Blood Pressure is more challenging. It naturally varies from moment to moment and is affected by factors such as posture, movement, stress and talking. Clinical results show lower agreement than for Heart Rate and Heart Rate Variability, while calibration can improve some performance measures.
For this reason, Blood Pressure results should be interpreted within the conditions and limitations of the relevant evaluation rather than compared directly with the performance of other parameters.
View clinical results and methodology →
Accuracy depends on measurement conditions
Clinical validation is only part of understanding measurement performance. The quality of the captured signal can also depend on factors such as lighting, camera characteristics, face position and movement.
Shen AI continuously evaluates signal quality during a measurement and provides feedback when conditions may affect the reliability of the captured signal. This helps users improve their measurement conditions and gives applications additional information about measurement confidence.
Evaluate Shen AI in your own environment
Many customers perform their own validation before deploying Shen AI. This can help establish how the SDK performs with your specific application, devices and target users.
For meaningful results, testing should use appropriate reference equipment, consistent measurement conditions and multiple measurements. For Heart Rate and Heart Rate Variability, comparing individual heartbeat intervals over the same time period provides a more reliable evaluation than comparing only the final summary values.
Blood Pressure requires additional care because of its natural variability. Participants should rest before reference measurements, and Shen AI and cuff measurements should be taken as close together as possible.
A single measurement rarely provides enough information to assess performance. Looking at repeated measurements under controlled conditions gives a much clearer picture of agreement and repeatability.
Download the accuracy evaluation guidelines →
Evidence with clear limitations
We aim to present measurement performance in context rather than as a single accuracy number.
Results depend on the parameter being measured, the reference method, measurement duration and conditions. Clinical studies also have defined participant populations and testing protocols, so their results should not automatically be interpreted as performance in every population, device or use case.
For customers preparing a formal validation study, the Shen AI team can provide guidance on designing an evaluation protocol appropriate to the intended use.