Recently World Health Organization published a guidance document regarding AI governance and ethics in healthcare. The document outlines key principles for AI deployments in daily medical practice and is a welcome addition to the discussions on the vanguard field of medical AI development. The paper is well summarized in this article by Nicole Wetsman.
In addition to calls for AI transperancy, accountability, equity and accountability, the paper also calls to protect human decision-making autonomy, with additional human oversight of AI made decisions.
This area mostly relates to AI in medical imaging, as this field is getting ever closer to autonomous operations. It serves as a cautionary tale, to discourage technology companies from the “break first, fix later” mentality, as well as stating that “emergencies do not justify deployment of unproven technologies”.
Few months ago at ECR Oxipit previewed ChestLink, the first autonomous medical imaging solution for healthy patient report automation. The product will automate (no findings) chest X-ray reports for healthy patients, where AI is highly confident of the results. The solution is currently undergoing medical device certification.
At Oxipit we strongly believe that automation of radiology is imminent. Yet it must proceed at a step-by-step pace, ensuring the steadfast reliability of underlying technology, as well as adhering to all applicable regulations.
To paraphrase, autonomous AI medical imaging will not come with a bang, but with a whisper.
In the coming weeks we will lay out our story of autonomous AI development. Stay tuned!