← All topics
06
Responsible AI
6 min

Responsible AI

Developing AI-powered health technology responsibly

AI used to measure physiological signals needs more than strong technical performance. It needs a disciplined approach to data, model development, validation and ongoing evaluation.

At Shen AI, we develop our technology with these principles in mind, combining AI development with clinical evidence, quality processes and careful consideration of how the technology is intended to be used.

Developing models with purpose

Shen AI's AI models are developed to extract physiological information from camera-based signals.

The technology combines information from rPPG and rBCG and uses signal processing and AI-based inference to distinguish physiological signals from noise and other variations in camera data.

Our development process considers the conditions in which the technology will actually be used, rather than treating controlled laboratory conditions as the only measure of performance.

Data used responsibly

Shen AI takes a privacy-focused approach to the data involved in its technology.

Physiological processing takes place locally on the user's device or within the web browser, rather than requiring users' facial video or physiological data to be sent to Shen AI servers.

Shen AI also states that personal and sensitive user data is not used to develop its AI models.

This approach helps separate the operation of the measurement technology from the use of personal data for model development.

Learn more about privacy & data protection →

Validation is part of development

AI performance needs to be evaluated against appropriate reference methods and under defined conditions.

Shen AI conducts clinical studies and technical evaluations to assess how its technology performs. These evaluations use appropriate reference measurements and consider factors that can affect camera-based measurement, including measurement conditions and duration.

We also encourage customers to perform their own validation before deployment. Testing the technology in the customer's own product, on target devices and with target users can provide a more meaningful understanding of performance in that specific environment.

Explore clinical evidence & validation →

Evaluating performance in real-world conditions

Camera-based physiological measurement is affected by factors such as lighting, movement, face position and camera characteristics.

Shen AI's technology is developed and evaluated with these variables in mind. The SDK also continuously evaluates signal quality during measurements and provides feedback when conditions may affect the captured signal.

This creates a feedback loop between the technology and the measurement environment: the system can identify when conditions are less suitable, while customers can use signal-quality information to better understand their measurements.

Learn about signal quality & measurement confidence →

Responsible use starts with intended use

Responsible AI also means being clear about what the technology is intended to do — and what it is not intended to do.

Shen AI offers different SDKs for different intended uses and regulatory contexts. Wellness applications are designed for informational health and wellness purposes, while the Medical SDK has a defined medical intended purpose and regulatory framework.

The technology should therefore be integrated, presented and used in a way that remains consistent with the intended use of the selected SDK.

Explore compliance & regulatory requirements →

Continuous improvement

Shen AI continues to evaluate and improve its technology as part of its development process.

Updates to AI-powered measurement technology need to be considered not only from a technical perspective, but also in terms of performance, quality and — where applicable — regulatory requirements.

For regulated medical-device applications, changes are managed within the applicable quality and regulatory framework.

A structured approach to AI-powered healthcare technology

Responsible AI is not a single feature or a one-time assessment. It is an ongoing process that connects data practices, model development, technical evaluation, clinical validation, quality management and intended use.

Our goal is to develop camera-based physiological measurement technology that can be evaluated transparently and integrated responsibly into the products and services that use it.