FOI release
Patient safety and regulatory enquiry – Rapid Health Smart Triage medical device
This request was refused in part, so we didn't provide some of the information the requester asked for. This may include information where we can neither confirm nor deny that we hold it.
Case reference FOI2026/00985
Received 29 August 2026
Published 13 October 2026
Request
I am writing as a member of the public with concerns regarding the regulatory status, clinical safety and post-market surveillance of Rapid Health Smart Triage, which is being deployed within NHS primary care. I understand that Smart Triage is software used to assess information supplied by patients and to determine or influence the urgency, routing and/or timeframe in which patients receive healthcare. I would be grateful if this correspondence could be forwarded to the MHRA team responsible for software as a medical device, medical-device compliance and post-market safety surveillance. I am not requesting confidential patient information or proprietary source code. I am seeking to understand the regulatory safeguards that apply to a system capable of influencing when patients receive clinical attention. 1. Medical-device registration Please confirm whether Rapid Health Smart Triage is currently registered with the MHRA as a medical device. If so, please provide or confirm: * the registered manufacturer; * device name; * MHRA registration/reference number; * stated intended purpose; * medical-device classification; * UKCA/CE status; and * date of registration. 2. Device classification Please explain which medical-device risk classification applies to Smart Triage and the regulatory basis for that classification. In particular, does its intended purpose encompass assessing symptoms and determining or recommending the urgency, routing or timeframe for medical attention? If so, I would be grateful for clarification as to how that functionality was taken into account when determining the appropriate device classification. 3. Nature of the technology Rapid Health and NHS material have described Smart Triage using terms including AI and automated/autonomous triage. Please advise what information the MHRA holds concerning whether Smart Triage uses: * artificial intelligence; * machine learning; * a large language model; * generative AI; * deterministic/rules-based clinical decision logic; or * another form of automated clinical decision-making. If the device is predominantly rules-based rather than machine-learning based, please explain whether this affects its regulatory classification or the clinical evidence required. 4. Clinical evidence What clinical evidence is required for software of this type to demonstrate that it performs its intended purpose safely? In the case of Smart Triage specifically, does the MHRA hold or has it reviewed evidence concerning: * sensitivity for urgent and emergency presentations; * false-negative rates; * under-triage rates; * over-triage rates; * red-flag detection; * comparison with assessment by qualified clinicians; and * performance in children, elderly patients and other vulnerable groups? If such evidence is held and can be disclosed, please identify or provide it. If the MHRA does not itself review this evidence before registration of a device of this classification, I would be grateful if you could explain this clearly and identify who is responsible for assessing it. 5. Clinical evaluation Is Rapid Health required to maintain a Clinical Evaluation Report or equivalent clinical evaluation for Smart Triage? If so, what type and level of clinical evidence is required to support the manufacturer's conclusion that the benefits of the system outweigh its clinical risks? Does the MHRA independently review that Clinical Evaluation Report? 6. Under-triage as a device safety risk I am particularly concerned about the possibility of under-triage. If software determines that a patient does not require urgent medical attention, but the patient subsequently proves to have a serious or time-critical condition, would this potentially constitute a medical-device adverse incident or near miss reportable to the MHRA? Similarly, would a failure by the software to recognise a clinically significant red flag potentially constitute a reportable medical-device incident? 7. Adverse incidents involving Smart Triage Has the MHRA received Yellow Card reports, manufacturer vigilance reports or other reports concerning Rapid Health Smart Triage? In particular, have reports concerned: * under-triage; * delayed medical attention; * incorrect urgency assessment; * failure to recognise red-flag symptoms; * incorrect routing; * emergency hospital treatment following a lower-urgency triage outcome; * serious patient harm; or * death? I appreciate that information about individual patients cannot be disclosed. However, I would be grateful if the MHRA could provide aggregate numbers and categories of reports concerning this product where legally possible. 8. Manufacturer's reporting obligations Please explain Rapid Health's obligations to report adverse incidents and safety signals concerning Smart Triage to the MHRA. In particular: * What types of under-triage incidents must be reported? * What are the applicable reporting timescales? * Is Rapid Health required to monitor trends in less serious incidents or near misses? * Is the manufacturer required to notify the MHRA when clinical decision rules are changed in response to safety incidents? 9. Post-market surveillance What post-market surveillance is required for Smart Triage? Is Rapid Health required to monitor the real-world performance of its triage system after deployment? Does that surveillance specifically need to measure or identify: * under-triage; * false negatives; * red-flag failures; * differences between automated triage and subsequent clinician assessment; and * differences in performance between patient groups? 10. Changes to clinical decision rules If Rapid Health changes the clinical rules determining urgency or patient routing, what regulatory process applies? At what point would a change to those rules constitute a significant modification requiring further clinical evaluation, regulatory assessment or notification to the MHRA? 11. Patients and Yellow Card reporting Please confirm whether a patient or member of the public can submit a Yellow Card medical-device report where they believe Smart Triage: * incorrectly assessed their urgency; * delayed access to appropriate medical care; * failed to identify serious symptoms; or * otherwise affected their diagnosis or treatment. If so, I would be grateful if you could confirm the appropriate reporting route. 12. Regulatory responsibility Finally, I would appreciate clarification of an important point of accountability. When medical-device software autonomously determines or influences the urgency with which an NHS patient receives medical attention, which organisation is responsible for demonstrating that the algorithm or clinical decision rules are sufficiently safe for that intended purpose? In the case of Smart Triage, is that determination made by: * Rapid Health as manufacturer; * the MHRA; * an Approved Body; * NHS England; * the NHS organisation deploying the software; or * some combination of these organisations? I would be particularly grateful if you could explain which organisation has actually examined the clinical performance evidence demonstrating that Smart Triage can safely distinguish patients who can wait from those requiring urgent medical attention. 13. Regulatory review Given the increasingly widespread deployment of automated triage systems within NHS primary care, has the MHRA undertaken, or does it intend to undertake, any specific review of the clinical safety of Rapid Health Smart Triage or comparable automated triage systems? If concerns about a particular deployment or patient experience should instead be submitted through the Yellow Card scheme, please advise accordingly. Thank you for considering these questions. My purpose in contacting the MHRA is to understand the regulatory chain by which software capable of influencing the urgency of NHS patient care is demonstrated to be safe, and how possible failures are detected and investigated once that software is in widespread clinical use. I would be grateful for a substantive response or for this correspondence to be forwarded to the appropriate medical-device safety and compliance team.
Response
See attached
Documents
This is Medicines and Healthcare products Regulatory Agency's response to a freedom of information (FOI) or environmental information regulations (EIR) request.
You can browse our other responses or make a new FOI request.