GE HealthCare taps Amazon Web Services to build generative AI for medical use

GE HealthCare on Thursday announced it is teaming up with Amazon Web Services to build new generative artificial intelligence models and tools that can efficiently analyze complex medical data.

The health-care industry is responsible for generating nearly one-third of all data globally, but much of this information isn't easily accessible. Since patients' medical records, images, scans and insurance records are stored across different file formats and systems, it can be challenging for doctors and researchers to sort through this mountain of information – especially on a larger scale. 

For instance, as much as 97% of the data produced by hospitals goes unused, according to a report from Deloitte. GE HealthCare, which offers medical imaging, ultrasound, patient care and pharmaceutical diagnostic solutions, believes generative AI can help. 

The company is collaborating with AWS to build models that clinicians can use to leverage data more efficiently across health-care operations, including within screenings, diagnoses, decision support and workflows like scheduling. 

"The tools we expect to build as a result of this will be aimed at helping hospitals and clinicians make the most of the data that they have," Dr. Taha Kass-Hout, GE HealthCare's global chief science and technology officer, told CNBC in an interview. 

Kass-Hout said AWS will likely help GE HealthCare speed up its development and deployment of web-based medical imaging applications, for instance, which would provide radiologists and other doctors with easier access to analytics. 

GE HealthCare offers its own AI tools, but its partnership with AWS will supply the company with the technical infrastructure required to quickly build generative AI models and tools at scale. GE HealthCare will use AWS' solutions like Amazon Bedrock and Amazon SageMaker, according to a release Thursday. 

"Training these models requires a lot of compute, requires a lot of data, requires a lot of expertise, and we're collaborating in that regard," Matt Wood, vice president of AI at AWS, told CNBC in an interview. 

In addition to building applications for health care more broadly, Kass-Hout said GE HealthCare is also exploring how to use generative AI to streamline the company's internal productivity. He said one of its initial priorities will be using an assistive tool called Amazon Q Developer to generate real-time code suggestions for its software developers, which should help them work more efficiently. 

Kass-Hout said GE HealthCare maintains rigorous testing and standards before bringing products to market, and the same will be true with the generative AI applications it develops. GE HealthCare does not train models on customer data, he added.

The company's new models and applications will initially be available to GE HealthCare employees and customers, but it plans to make them more widely accessible in the future.