Google Cloud and the German health-care company Bayer on Tuesday announced they are building an artificial intelligence-powered platform that aims to help radiologists diagnose patients and work through cases more quickly.
The platform’s generative AI flags anomalies within images for radiologists to look at, and it can also pull up relevant information from that patient’s medical history, Thomas Kurian, CEO of Google Cloud, told CNBC. If a patient comes in for an annual breast cancer screening, for example, the platform can detect current problems, compare the image to prior screenings and summarize that information, he said.
Health-care companies like Bayer will be able to use the platform to develop radiology-specific applications that carry out functions like these more easily, Google said.
A radiologist is a doctor that uses medical images like CT scans, MRIs and X-rays to identify and treat conditions. But like physicians across many specialties in the U.S., radiologists are facing a growing labor shortage, according to the Association of American Medical Colleges. As of early April, there are more than 1,800 vacant job postings on the American College of Radiology’s website, compared to around 220 listings in April 2014.
Many radiologists are also fighting burnout as an aging population and easier access to imaging technologies have led to mounting caseloads. Google Cloud said its new platform could help alleviate these ongoing workforce challenges.
“That whole process flow is designed to help radiologists get through their task with assistance more quickly,” Kurian said in an interview. “It makes them more efficient so they can actually see more images and service more patients.”
Kurian said the platform does not replace radiologists, as the doctor is in “sole control” of the recommendation they will make. Instead, he wants people to look at the platform as an assistive tool, like a microscope. The goal is to easily give radiologists the information they need and save them from spending 15 or 20 minutes searching through patient records, Kurian said.
Google Cloud and Bayer are not the only companies exploring AI applications for medical imaging. In 2021, the Netherlands-based health-care company Philips and Amazon Web Services said they are working to use AI to analyze medical imaging data. Similarly, GE HealthCare published a blog post in 2022 about the various AI tools it has developed for radiology.
Keith Kirkpatrick, research director at The Futurum Group, said there’s not one clear leader in the medical imaging AI market yet since the technology is still so new.
“It’s really wide open,” Kirkpatrick told CNBC. “We’re still fairly early in the game right now.”
Kirkpatrick, who was briefed on Tuesday’s announcement, said Google Cloud and Bayer’s radiology platform will have to demonstrate high levels of technical accuracy, offer strong privacy and security controls and be easy to use in order to win in the space. Establishing trust with radiologists will be the key, he added.
“Google is going to have to make sure that their technology is as close to foolproof as possible,” Kirkpatrick said.
Google Cloud has been working with Bayer on the radiology platform for around five years. The foundation was built using existing Google Cloud solutions like Vertex AI, Healthcare API and BigQuery, and Kurian said the platform’s data is encrypted.
The companies drew on Bayer’s expertise in radiology to make sure that the product is easy for the doctors to use. Bayer said its radiology products generated around €2 billion ($2.16 billion) in sales last year, according to a release.
Even so, the platform represents a foray into an entirely new business model for Bayer, according to Guido Mathews, Bayer’s vice president of radiology.
“We don’t offer a new pill — we offer a service for which we will charge users accordingly,” Mathews told CNBC in an interview. “To help develop models and also to help deploy models for radiology, that’s a big step forward for us.”
Google Cloud and Bayer are exploring a number of different pricing models for the platform, he said. Other health-care organizations will begin testing and providing feedback on the platform this year.