AI and NHS MSK triage, explained.
What AI MSK triage means
AI MSK triage usually means software that uses machine learning, or a large language model, to suggest how urgent a musculoskeletal problem is, where the patient should go, or what the diagnosis might be. It learns patterns from data rather than following rules a clinician has written.
The phrase is used loosely. Some products labelled AI are mostly questionnaires with fixed rules; some are genuine learning models. It is worth asking which, because the two behave very differently when something goes wrong.
Where AI helps
Volume
Learning models can read and sort large numbers of free text messages and forms quickly, which matters where services are swamped.
Patterns
With enough good data, a model can spot combinations of features that a simple checklist would miss.
Language
Language models can turn a patient's own words into structured information, and draft letters and summaries.
Where AI struggles
Three questions are harder for a learning model than for written rules, and they matter most at the point where a patient's care is decided.
Can it explain itself? A model's reasoning can be difficult to show in a form a clinician, an information governance reviewer or a coroner can check.
Who is accountable? When a suggestion is wrong, someone must be able to say which part of the tool was wrong and fix it.
Does it stay the same? A model retrained on new data can change its behaviour in ways that are hard to see, which is why regulators treat change control seriously.
In the UK many of these products are regulated as medical devices; the MHRA guidance on software and AI as a medical device sets out how.
Why Orwyn uses written rules
Orwyn is not artificial intelligence. It runs on rules a named consultant surgeon has written, each linked to the published evidence behind it, so every suggestion shows its reasons and can be checked.
AI tools help us build and test the software faster. They never make a clinical decision: every rule Orwyn applies is written by a consultant surgeon.
Rules have limits too: they only know what has been written into them. That is why every clinician's disagreement with Orwyn is recorded and reviewed, and why the rules are versioned and dated like any other clinical document.
Most tools built for MSK triage aim to keep people out of hospital. Orwyn helps the clinician decide when a patient should go in, when they should not, and gives the clinician and the patient a reason to trust the answer.
Orwyn is currently being assessed for an NHS pilot to test its safety, clinical effectiveness and cost effectiveness.
We compare the two approaches in more detail in what MSK clinical decision support is, and explain how NHS MSK triage works for service leads.
Questions we are asked
What is AI MSK triage?
AI MSK triage usually means software that uses machine learning or a large language model to suggest how urgent a musculoskeletal problem is, where the patient should go, or what the diagnosis might be, based on patterns learned from data rather than rules a clinician has written.
Is Orwyn artificial intelligence?
No. Orwyn runs on rules a named consultant surgeon has written, each linked to the evidence behind it. Every suggestion can be traced to the rule that produced it and the source that rule rests on.
Is AI in MSK triage regulated?
In the UK, software that influences clinical decisions, including AI, is often regulated as a medical device by the MHRA. In the NHS it must also meet clinical safety standards DCB0129 and DCB0160.
Where to read more
MHRA guidance on software and AI as a medical device, MHRA.
DCB0129 clinical risk management standard, NHS England. Read it here.
What is MSK clinical decision support, our companion reference page. Read it here.
Surgeon-governed decision support for first-contact MSK clinicians.