← All topics
08
Reliability
6 min

Measurement quality & reliability

Designed for reliable measurements in real-world conditions

Camera-based physiological measurement depends on the quality of the signal captured during a scan. Lighting, movement, face position and camera conditions can all affect measurement quality.

Shen AI continuously evaluates these conditions throughout the measurement and provides feedback to help users improve the signal. Applications can also access signal-quality information and use it to make informed decisions about measurement results.

Quality starts before the measurement

Before a scan begins, the SDK analyzes the camera view to determine whether conditions are suitable.

When something needs to be adjusted, users receive specific guidance. Depending on the situation, this may include moving to better lighting, avoiding strong backlight, uncovering the forehead, removing glasses or changing the distance from the camera.

The guidance updates as conditions change, helping users start the measurement under suitable conditions rather than discovering a problem after the scan is complete.

Signal quality is assessed continuously

Once the measurement starts, Shen AI continuously evaluates the quality of the captured physiological signal.

With the built-in interface, users see this through the on-screen quality indicator. The SDK also exposes a Signal Quality value through the API, ranging from 0 to 1, with higher values indicating greater confidence in the captured signal.

The value updates throughout the measurement as conditions change.

When Blood Pressure is included, Signal Quality reflects the environmental conditions relevant to blood pressure estimation. Because Blood Pressure requires stricter conditions, a lower Signal Quality value can be expected when it is part of the measurement.

What affects measurement quality?

The SDK considers several factors that can affect the captured signal, including:

  • lighting conditions
  • camera characteristics
  • face positioning
  • forehead visibility
  • movement
  • the quality and consistency of the physiological signal

These factors can change during a scan. By continuously assessing them, Shen AI can provide more useful feedback than simply marking a measurement as successful or unsuccessful.

Helping users improve over time

Measurement quality is not only about the result of one scan. Clear feedback can also help users develop better measurement habits.

After a measurement, Shen AI can identify the single most important factor that could improve the next scan. For example, a user may be advised to improve lighting, keep the device steadier or make sure their forehead remains visible.

Providing one clear recommendation helps users understand what to change without overwhelming them with multiple warnings.

Giving applications more control

Signal Quality is also available to the application through the SDK.

Depending on the use case, developers can use this information to:

  • show confidence information alongside results
  • decide whether to ask the user to repeat a measurement
  • filter low-quality measurements before further analysis
  • monitor measurement quality across users or devices
  • build analytics around measurement conditions over time

There is no single Signal Quality threshold that is appropriate for every application. The right approach depends on the use case and the balance between measurement quality and user experience.

Reliability in real-world conditions

Reliable camera-based measurement requires more than performance under ideal conditions. Users measure themselves with different cameras, in different environments and with varying levels of movement and lighting.

Shen AI's continuous signal-quality assessment is designed to account for these changes during the measurement and help the application respond when conditions are less suitable.

Combined with appropriate user guidance and customer-side testing, this provides a way to manage measurement quality as part of the overall product experience.

Learn how to evaluate Shen AI's accuracy →
Explore clinical evidence & validation →