0
Evaluating the use of AI in the deployment setting of a primary healthcare triage use case
22.11.2024
Introduction
AI in medical imaging heralds a new era, where machine learning serves as a complement to human expertise, offering a "second set of eyes" to enhance the interpretation of radiological studies and improve diagnostic precision. AI algorithms have shown remarkable capabilities to quickly analyse complex imaging data and detect subtle abnormalities that might be overlooked by the human eye.
That said, the path from innovation to implementation is not without hurdles. Recent studies highlight the need for rigorous assessment of AI’s real-world effectiveness—not just its technical capabilities, but also how seamlessly it integrates into existing clinical workflows.
About PRIME-CXR
Led by the National Healthcare Group (NHG), PRIME-CXR is an initiative to speed up the review process of chest X-rays (CXRs) at National Healthcare Group Polyclinics (NHGP).
Currently, all CXRs performed at NHGP are reviewed, analysed, and interpreted by radiologists. Patients would have to wait for the review of the CXR to be completed before they are discharged or referred for further investigation in the case where abnormal findings were discovered. With imaging volumes having increased over the years, there is a need to leverage technology and innovation to complement the skillsets of radiologists to improve patient care.
The initiative uses an AI-powered CXR analysis solution called Lunit INSIGHT CXR to help enhance the diagnosis of lung and heart conditions by rapidly identifying and prioritising CXRs with abnormalities. This, in turn, reduces patient wait times for discharge or referral to higher acuity facilities.
As an independent, third-party AI assurance firm, Resaro is proud to contribute to PRIME-CXR by evaluating the feasibility of deploying Lunit INSIGHT CXR for CXR assessments under the proposed triage approach. We compared the performance of the AI solution with that of a practising radiologist and provided recommendations for the operational deployment, including the optimal threshold settings for the AI predictions.
Associate Professor Tan Cher Heng, Group Chief Research and Innovation Officer of NHG (Singapore) says, "Resaro's AI assurance team delivered crucial insights. Their thorough testing of a third-party AI solution helped us grasp its value and limitations. Their meticulous validation process helped us understand how best to maintain the highest standards of patient care and optimise healthcare delivery as we prepare to adopt AI in clinical settings."
An initial pilot phase for Lunit INSIGHT CXR will run at Geylang Polyclinic from October 2024 until August 2025. Based on the findings from the pilot, the AI system may be fully deployed to additional NHGP, marking a milestone in primary care capabilities.
In partnership with Resaro, the PRIME-CXR team will also be establishing an evaluation framework for radiology AI solutions, allowing health systems to evaluate and procure solutions that could bring about the highest clinical value.
Images: Minister of Health Ong Ye Kung dropped by our PRIME-CXR booth at the Singapore Health & Biomedical Congress on October 10th, 2024, to hear how an AI augmented chest X-ray workflow can help patients reduce their waiting time and expedite reporting for abnormal cases.
19.11.2024
Navigating Generative AI: the 3 worldviews on Innovation, Safety, and Security
01.11.2024
Securing AI: A Collective Responsibility
29.10.2024