Evaluating the impact of Artificial Intelligence (AI) for chest radiograph interpretation on clinical decision making in emergency medicine

Start Date

18 July 2022

End Date

25 July 2024

Project Participants

Annalise AI, Macquarie University, Sydney Local Health District

Status: Completed

Tags: acute care, artificial intelligence (AI), chest radiography, clinical decision support, emergency medicine, hospital, medical imaging

Opportunity 

Emergency doctors frequently interpret chest x-rays and make time-critical clinical decisions before radiologist reports are available. AI offers the potential to support these decisions, but robust evidence is needed to understand if AI-assisted detection of pathological findings can improve care delivery and patient outcomes. 

Objectives 

This project evaluated Annalise Enterprise CXR AI system in the emergency department (ED) setting. with the hypothesis that Annalise CXR can support ED clinician decision-making and improve patient outcomes by increasing diagnostic accuracy, reducing treatment delays, and ensure reporting of non-emergent findings. The project also examined clinician experience with Annalise CXR and the implications on workflow, confidence and safety. 

The randomised controlled study involving nearly 200 Australian emergency doctors showed that AI support significantly improved diagnostic accuracy of most doctors in interpreting chest x-rays, with moderate improvements in patient management decisions. The AI system helped reduce cognitive load in complex scenarios and increased diagnostic confidence. The findings also identified important safety considerations relating to incorrect AI findings and clinician experience levels. 

The project provided an evidence base for the safe deployment of AI-enabled clinical decision support in Australian hospitals for emergency doctors interpreting chest x-rays.  

The research highlights opportunities to improve clinical decision-making while demonstrating the importance of governance, validation and clinician oversight in future AI implementation.

Integrity, Excellence,
Teamwork and Authenticity

Hand holding a smartphone against a colourful, defocused background