Insights from the FAMOS deployment at Leeds
As AI tools become more common in clinical settings, a key challenge remains unsolved: how can hospitals ensure these tools remain safe, effective, and trustworthy once deployed? At Newton’s Tree, a healthtech company building a vendor-neutral platform for AI selection, evaluation, and monitoring, we have developed the Federated AI Monitoring Service (FAMOS) to address this challenge.
In partnership with Leeds Teaching Hospitals NHS Trust (LTHT), Newton’s Tree has piloted FAMOS to demonstrate the feasibility of monitoring third-party AI applications in real time. The project makes the case for proactive AI monitoring as a foundational requirement for safe and scalable digital health.
Strategic vision and objectives
The strategic aim of the FAMOS project is to provide NHS Trusts with a real-time monitoring framework that supports safe, long-term use of AI. As AI becomes embedded in everyday care, hospitals need infrastructure that detects risks early and supports responsible deployment at scale.
FAMOS enables hospitals to monitor AI across multiple workflows and vendors, offering a system-wide view of performance. This aligns with the NHS’s broader vision of safe, patient-centred digital transformation, supporting innovation without compromising quality or safety.
Addressing a growing health challenge
As AI adoption accelerates, post-deployment oversight remains inconsistent. Current approaches rely on end-user feedback and adverse event reporting – reactive measures that often fail to capture slow performance degradation or shifts in clinical use.
There’s also confusion around who is responsible for monitoring: vendors, clinicians, or IT teams? FAMOS addresses this by creating shared visibility and accountability. It collects relevant metrics continuously, making it possible to spot issues – such as performance drift or input data anomalies – before they impact patients.
Patient-centric innovation
Newton’s Tree built FAMOS with patients and clinicians in mind. Radiologists at LTHT helped co-design the dashboard, ensuring the metrics surfaced were clinically meaningful and actionable.
These insights are surfaced in real-time on an intuitive dashboard, allowing clinicians to adjust workflows or escalate concerns quickly. Early results from the Leeds pilot show that this level of visibility could support safer, more responsive care.
Impact of AI and data on care
FAMOS illustrates how AI and real-time data can enhance care delivery. Its federated design means hospitals can collaborate without moving patient data offsite – maintaining privacy while enabling performance comparisons across sites.
During the pilot, the system revealed differences in how AI tools were being used and trusted by clinicians. For instance, one site showed higher automation bias – clinicians accepting AI outputs without sufficient scrutiny. This insight helps guide training and workflow changes to reduce risk.
Real-time data also uncovered patterns in imaging artefacts and other quality issues, allowing the clinical team to address upstream problems. These are insights that would not surface through static reporting alone.
Collaboration and stakeholder engagement
Success at LTHT was driven by strong collaboration across clinical, technical, and vendor teams. From image processing to dashboard validation, stakeholders contributed actively to shaping how FAMOS was deployed.
Radiologists reviewed cases at regular intervals, comparing their interpretations to AI outputs. Hospital IT ensured fast system integration, with metrics displayed within 20 seconds of AI output generation. Third-party vendors, including Annalise.ai and Qure.ai, supported transparent performance evaluation – something increasingly important in NHS procurement.
This shared effort created a model of how Trusts and vendors can work together to improve AI accountability in real-world settings.
Scaling and future expansion
If successful, the next phase will see FAMOS extended to more AI tools across LTHT, including modalities like CT and MRI. Its modular, vendor-neutral architecture means other NHS Trusts can adopt the platform without custom development.
Scaling will require investment in integration and training, as well as clearer standards for AI monitoring. Newton’s Tree is working with NHS England and other partners to help define those standards and support national rollout.
Broader implications for healthcare digitisation
Real-time monitoring infrastructure like FAMOS ensures that Trusts can identify risks early, maintain oversight across vendors, and adapt to changes in clinical practice. As the NHS moves toward more proactive, tech-enabled care, tools like FAMOS will be essential for maintaining trust and safety. Newton’s Tree believes this is the future of AI in healthcare: not just intelligent, but accountable.
About the author: Haris Shuaib is CEO of Newton’s Tree and an NHS Consultant Clinical Scientist.