The Government’s Department of Science, Innovation and Technology has included Newton’s Tree Federated AI Monitoring Service (FAMOS) to its AI Assurance Techniques portfolio.

The portfolio has been developed to support anyone developing, deploying or using AI by providing them with appropriate techniques for trustworthy AI.

FAMOS gives hospitals, AI developers, and regulators the ability to observe and monitor AI performance in near real time, which helps them to tackle the challenge of drift; where changes in clinical environments, input data or human-computer interactions can degrade AI performance.

“It is excellent to see FAMOS included in this portfolio. AI that starts safe does not necessarily stay safe, and existing frameworks in healthcare arguably do not mitigate this well enough. Our monitoring dashboard aim’s to turn the tide of this risk management issue, and ultimately improve the care of patients - who are at the heart of this”.
- Robin Carpenter, Head of AI Governance and Policy at Newton’s Tree

With the portfolio is a description of how FAMOS aims to support the UK’s 5 regulatory principles:

1. Safety, Security and Robustness: it helps users detect early performance issues so manufacturers and healthcare organisations can intervene in time.
2. Transparency and Explainability: collecting and collating information so performance information can be understood and communicated.
3. Fairness: if appropriate data is presented to the stakeholders at the right time, discriminate model or user behaviour can be mitigated with corrective action.
4. Accountability and governance: an AI vendor needs the right information in order to uphold the quality of their product, and likewise a healthcare organisation needs the right information to uphold the quality of the care provided.
5. Contestability and Redress: FAMOS will provide relevant metrics in one place to support an investigation into contestability concerns such as automation bias.

Ensuring AI applications remain safe and effective after they are deployed is critical for both hospitals and vendors. Get in touch to see if we can support your team keep AI applications safe within your organisation.