Nividia - OneAdvanced Partnership
Sovereign AI, built for NHS care navigation
OneAdvanced and NVIDIA. Trained on NHS data. Hosted in the UK. Benchmark tested against leading AI models.
The problem with general-purpose AI in healthcare
AI adoption across the NHS is accelerating, but global, general-purpose models aren't necessarily built around the clinical context, workflows and data requirements of UK healthcare.
Clinical context gap
Patients don't describe symptoms in clinical terminology. General-purpose models can draw on vast medical knowledge, but accurate triage requires interpreting real-world patient language in the right clinical context.NHS workflow complexity
General-purpose models aren't trained on the NHS workflows and pathways that shape how patients are assessed, directed and referred, making it harder to map patient needs to the appropriate pathway or service.Data governance risk
Processing confidential NHS patient data through AI architectures without appropriate UK data controls can create risks around data residency, security, governance and regulatory compliance.OneAdvanced and NVIDIA: Building sovereign AI for NHS Care Navigation
OneAdvanced and NVIDIA partnered to develop and validate a contextual sovereign AI model for NHS Care Navigation, combining OneAdvanced’s UK sovereign healthcare data and NHS workflow expertise with NVIDIA AI infrastructure. Built using NVIDIA Nemotron open models and NeMo libraries, the model was trained on pseudonymised NHS patient triage requests. The pilot demonstrates the potential of ‘Care Navigator LLM’ to improve triage accuracy, support faster access to appropriate care and reduce NHS resource wastage
What the pilot demonstrated
2x
more accurate than a GP control group
150x
lower inference cost than leading frontier AI models
The value of specialist AI for the NHS
UK sovereign
Built and governed within the UK, patient data is stored and hosted under UK law, supporting NHS requirements around data residency, resilience and regulatory compliance. Model weights, fine-tuning, hosting and inference remain within the UK perimeter, keeping the AI capability under UK governance.
Clinical grade
Fine-tuned on pseudonymised NHS Primary Care patient triage requests, the model is grounded in real clinical and NHS workflow context. GP corrections made during use of the system can provide the basis for continuous training data, supporting the model’s evolution with clinical feedback over time.
Cost efficient
A smaller, specialist model architecture focused on care navigation is designed to deliver the required clinical capability without the cost of running much larger general-purpose models. With lower inference costs than leading frontier models, it can support more cost-efficient AI deployment at NHS scale, while reducing resource wastage and increasing clinical capacity.
OneAdvanced's work demonstrates how NVIDIA technology can be applied to create practical AI systems that support clinicians, improve patient experience and operate safely at scale.
Anthony Hills, Director UK & IrelandNVIDIADetecting clinical topics in patient requests is key to enabling the AI Care Navigator to ask the right follow-up questions and provide the right response to patients.
Dr Ben Brown, GP Partner Langworthy Medical Practice and Medical DirectorOneAdvancedWe are focused on delivering structural advantages to the UK NHS, along with compliance to medical device and UK sovereign AI model standards.
Simon Walsh, Chief Executive OfficerOneAdvanced
Built on more than 35 years of NHS experience
40m+
NHS patients supported annually
4000+
GP Practices
85%
of NHS 111 services
Sovereign AI for NHS Care Navigation
Ready to explore the potential of AI in NHS care navigation?
Curious about how AI could improve triage accuracy, reduce admin burden or support care navigation? We'd like to talk.