In emergency medicine, medical practitioners are often required to make fast and accurate diagnoses multiple times each day to ensure patients have the best possible outcomes. Undiagnosed fractures are a common concern in this setting, and the consequences for the patient can lead to delayed care resolution and a spectrum of avoidable complications. Fracture detection solutions are particularly critical in this context, as radiographic fracture identification remains highly complex and challenging. To address this persistent issue, the National Institute for Health and Care Excellence (NICE) in the UK established a committee to assess the role of AI in radiology and shortlisted top fracture detection AI solutions, including AZmed’s AZtrauma, one of the verticals of Rayvolve® AI suite, as part of its Early Value Assessment (EVA).
Why Missed Fractures Remain a Critical Challenge in Radiology
Emergency departments at the NHS are frequently inundated with patients who need to be seen immediately. Diagnostic determinations have to be made quickly and with the information at hand. Unfortunately, this can sometimes result in missed fractures. Misdiagnosis in traumatology appears in 3-10% of cases1.
Several factors contribute to this situation:
- Time pressures: This cause arises from limited time for analysis due to the high volume of patients.
- Cognitive load fatigue: Working long shifts and treating a continuous stream of patients can also cause fatigue, increasing the risk of errors.
- Pathophysiological progression of fractures: Certain fractures may not be easily identifiable on X-rays.
- Inconsistent expertise: Varied expertise across medical professionals is also a cause for concern.
NICE considers AI a sustainable solution to fundamental healthcare challenges. This EVA program is important progress in the ongoing process of getting AI into routine medicine in the United Kingdom.
Key Points from NICE
The diagnostics advisory committee evaluated evidence from various sources and considered key factors related to the adoption of fracture detection solutions powered by AI. These include:
Patient and Carer Considerations
Patients worried about missed fractures leading to complications and expressed the physical and psychological strain of delayed diagnoses, including time lost from work or school. They noted that AI tools could help improve the accuracy of diagnoses and reduce misdiagnoses, bringing peace of mind during emergency medicine visits. However, the aspect of human interaction remained crucial. Patients emphasized that AI should augment, and complement rather than replace healthcare providers. The committee said it was important to educate patients about the role of AI, making sure that they were aware of its advantages and limitations, as well as involving them in shared decision-making.
Clinical Efficacy
The committee assessed 16 studies that investigated the sensitivity and specificity of AI-assisted radiographic fracture identification in adults and pediatric patients. AZtrauma part of the Rayvolve® AI suite was able to significantly improve sensitivity from 86.5% to 95.5% when used by healthcare professionals. AI can definitely play a very useful second-opinion role during clinical assessments. However, the studies were mainly retrospective, and there was a dearth of real-world evidence in UK emergency medicine clinical care contexts. The committee recommended more real-world evidence to verify these findings and assess AI X-ray tools across a wide array of clinical settings.
System Impact
AI technologies could decrease variability in X-ray interpretation, especially in smaller and less well-resourced healthcare settings. The tools showed consistency regardless of cognitive load fatigue or time pressures. Widespread adoption may reduce the geographical difference in the quality of care, the committee acknowledged. However, they are also concerned that over-reliance on AI may lead to de-skilling of healthcare providers. To counter this, clinical radiologists and emergency physicians can interpret X-rays without looking at AI results first. And automated AI analysis should not replace a physician’s diagnosis.
Cost-Effectiveness
Initial models indicated that fracture detection solutions are cost-effective, when looking again at implementation costs at this stage, estimated at £1 per x-ray scan. Based on outdated estimates of the accuracy, AZtrauma initially generated a negative incremental net health benefit, but upon integrating up-to-date data it became evident that, when incorporated into clinical workflows, the AI had the potential to be cost-effective.
NICE Requirements for AI Tools in Healthcare
NICE has defined some criteria to support the generation of data for these devices:
- Regulatory Compliance
- Obtaining required approval(s), e.g., CE mark, FDA clearance.
- NHS Digital Technology Assessment Criteria (DTAC).
- Evidence-Based Validation
- Demonstrate enhanced diagnostic accuracy over traditional approaches.
- Show favorable impacts on diagnostic determinations and patient outcomes.
- Demonstrate cost-effectiveness and system-wide benefits.
- Continuous Monitoring and Reporting
- Annual submissions to NICE.
- Engage in continual improvement and review processes.
- Assessment Submitted to NHS for Final Approval
- Have a complete reassessment following the two-year evidence-generation period.
- Evidence of preparation for wider NHS rollout.
These requirements help ensure that only safe, effective, and reliable AI tools enter clinical practice.
The Benefits of Fracture Detection AI
Fracture detection AI tools like AZtrauma can revolutionize emergency medicine in many aspects:
- Improved Diagnostic Accuracy
- Missed fractures can be addressed with the support of AI, as studies show it outperforms radiologists in detecting commonly overlooked cases2.
- Reduced Radiological Reporting Times
- AI interpretation only takes seconds to complete, shortening the wait for patients and increasing throughput in the ED.
- Increased Confidence in Clinicians
- AI-backed insights can help providers feel confident in their diagnoses even when they may be dealing with unique or tricky cases.
- Reduced Healthcare Costs
- AI can reduce the chances of missed fractures, which can lead to costly delays in treatment or potential lawsuits.
- Standardized Care
- AI tools could enable less variability in care by providing uniform evaluations across various healthcare organizations.
Conclusion
Artificial Intelligence is transforming healthcare, mainly in radiology, where it is improving diagnostic accuracy and optimizing workflows. A broad global momentum toward AI adoption is bolstered by strong frameworks, such as the EU Artificial Intelligence Act, which is in the process of implementation, and the FDA’s regulatory clearances, to help ensure the ethics and technical standards needed to meaningfully integrate these tools into clinical practice. The NICE assessment highlights a focus on cost-effectiveness and real-world application of systems and is entirely consistent with a commitment to evidence-based deployment of AI. Technologies will not only hit their stride in practicality but will also bridge gaps in diagnostics, ease clinician exhaustion, and help healthcare systems deliver steady, top-tier service as they advance in collaboration. AZtrauma is an example of how AI can fill gaps in care delivery, increasing sensitivity and efficiency while maintaining patient safety and clinician oversight.
About Rayvolve®
Rayvolve®, an AI radiology solution, which now encompasses multiple verticals: AZtrauma, AZchest, AZmeasure, and soon AZboneage (under development), is an AI solution designed primarily for radiology teams. It uses machine learning to quickly process images and help clinicians make accurate diagnoses and reduce workload burden. The solution is trained on one of the largest datasets of X-ray images (>15 million) in radiology AI. Rigorous medical annotations and development processes ensure AI of high performance and reliability, enabling the AI to identify complex and rare anomalies. This exhaustive training transforms the solution into a powerful resource. For its fracture detection solution, which we have discussed in this article, the solution is FDA-cleared for adult and pediatric populations in the US market and CE-marked in Europe. It is currently benefiting over 2,500 medical centers globally. To learn more about its use and implementation, read the full information here: Using AI for Fracture Detection in Radiology
2. Erik L. Ridley. AI algorithms help to spot overlooked fractures. RSNA. 2023
US - Medical device Class II according to the 510K clearance. Rayvolve is a computer-assisted detection and diagnosis (CAD) software device to assist radiologists and emergency physicians in detecting fractures during the review of radiographs of the musculoskeletal system. Rayvolve is indicated for adult and pediatric population (≥ 2 years). EU - Medical Device Class IIa in Europe (CE 2797) in compliance with the Medical Device Regulation (2017/745). Rayvolve is a computer-aided diagnosis tool, intended to help radiologists and emergency physicians to detect and localize abnormalities on standard X-rays. Caution: The data mentioned are sourced from internal documents, internal studies and literature reviews. This material with associated pictures is non-contractual. It is for distribution to Health Care Professionals only and should not be relied upon by any other persons. Testimonial reflects the opinion of Health Care Professionals, not the opinion of AZmed. Carefully read the instructions for use before use. Please refer to our Privacy policy on our website. AZboneage is an uncertified feature currently under development. For more information, please contact contact@azmed.co. AZmed 10 rue d’Uzès, 75002 Paris - www.azmed.co - RCS Laval B 841 673 601© 2024 AZmed – All rights reserved. MM-25-05
.jpg)