A couple weeks ago, we got the exciting news that the FDA had approved an artificial intelligence (AI) algorithm that does reads on images from a portable X-ray in order to screen for collapsed lung. The teams at GE Healthcare and UC San Francisco are behind this technology and research.
This is really exciting news because I believe it’s the first time the FDA has approved an AI algorithm for this type of radiology read. No doubt this is the first of many and shows the promise of what’s going to be possible when it comes to technology reading images.
Many large organizations may look at this and wonder why it’s such a big thing since they have a radiologist and radiology equipment available to do these reads and avoid adverse outcomes. While this is true at many large organizations, it’s not true for many resource limited care settings including late at night when the radiology staff might be sleeping.
This last situation is one that excites me most about AI reading images. AI never sleeps and so it has no problem reading any image at any hour. Given that this AI has shown an accuracy exceeding 96%, it’s the perfect option for these late night reads. If it discovers something abnormal, it can kick off a workflow for the onsite staff and the on call radiologist to do further analysis and treatment. However, if it finds that everything is fine, the radiologist doesn’t need to be disturbed. That’s a powerful outcome that will no doubt save lives and improve radiologist burnout.
When talking with a colleague about this technology, he provided some fascinating regional insight. While this technology is powerful at real time reads of chest X-rays, it can also be applied to all your past X-rays where it could discover missed diagnosis and other relevant findings that could help patients. Turns out that the European healthcare system loved this idea. The US healthcare system wanted nothing to do with it. Knowing how healthcare is paid for and the liability issues in the US, this shouldn’t come as a surprise, but is a sad reality of our healthcare system today. Maybe value based care can help adjust this perspective.
No doubt this is an exciting time for AI in healthcare. I can’t wait until we have more AI algorithms that are not only better than humans, but dramatically better. Turns out that this future is coming whether you want it or not. It’s more a question of whether healthcare will embrace this new technology as a tool for its success or if we’ll fight it and it will find other avenues of success. This technology just cleared by the FDA seems like it’s been cautious and will be the former.