A study by Stanford University School of Medicine and Unanimous AI has found that the new Swarm AI system can make more accurate diagnoses than individual doctors or machine learning algorithms alone.
Developed by Unanimous AI, the Swarm AI technology connects a group of doctors with artificial intelligence (AI) algorithms and combines their real-time individual insights into an optimised output.
In the study, the researchers analysed the accuracy of diagnosing pneumonia by using chest X-rays, a common imaging procedure in the US.
It was observed that diagnoses with the Swarm AI system led to a 33% decrease in the average error rate, when compared to conventional approaches by individual practitioners.
Based on this finding, the research team concluded that AI technologies can retain the direct participation of human practitioners as well as increase their diagnostic accuracy.
The Swarm AI technology was also compared to the CheXNet system, which uses only software algorithms for automated diagnosis of pneumonia from chest X-rays.
Results showed that Swarm AI was 22% more accurate in binary classification compared to the CheXNet system.
The researchers attributed this enhanced accuracy to the Swarm AI system’s functionality to combine real-time human insights with AI technology.
Stanford University Radiology assistant professor Matthew Lungren said: “Diagnosing pathologies like pneumonia from chest X-rays is extremely difficult, making it an ideal target for AI technologies.
“The results of this study are very exciting as they point towards a future where doctors and AI algorithms can work together in real-time, rather than human practitioners being replaced by automated algorithms.”
Furthermore, it is expected that the new AI technology can help in generating more accurate ‘ground truth’ datasets, which depend on human judgement, to train machine learning algorithmic systems.