AWS and GE HealthCare are teaming up to use AI to improve patient care

The collaboration will develop AI models and applications for health care practitioners. The goal? Raising the standard of care for all of us.

Nearly a third of all digital information is generated by the health care sector, but 97% of it isn’t available to physicians or clinicians because not only does it take a variety of unstructured forms–from written notes to images and more–it also tends to get trapped in data silos worldwide.

Today, AWS and GE HealthCare announced a collaboration that aims to unlock that vital information and unleash a new era of wellness possibilities with industry-specific artificial intelligence (AI) foundation models (FMs) and novel applications. GE HealthCare unveiled plans to train and deploy new clinical FMs on AWS’s machine learning and generative AI technologies with the goal to help health care providers not just improve existing protocols and workflows, but invent entirely new approaches to deliver better patient care for everyone.

As with everything involving AI, collecting, analyzing, and utilizing the right data is the key to unlocking new insights and solutions. In the case of health care, the relevant data includes things like doctor’s notes, x-rays, and all the other test results and observations that are gathered during medical visits.

“Health care systems collect large numbers of data points about patients over a lifetime. Unfortunately, very little of it can be accessed efficiently and securely to help inform diagnoses, prognoses, or treatments for patients,” said Dr. Taha Kass-Hout, global chief science and technology officer at GE HealthCare. “Until recently, technology capable of aggregating, analyzing, and interpreting this data securely and efficiently simply didn’t exist.”

Enter AWS, and its growing set of powerful and highly secure AI capabilities and services.

Using Amazon Bedrock, GE HealthCare will both tap the world’s leading foundation models and train new models to develop AI powered application to help hospitals and clinics access and gain insights for improving patient care based on a comprehensive view of their data, rather than just pieces of it. To help develop these future AI applications, GE HealthCare’s internal developers are planning to use Amazon Q Developer, as well as Amazon Q Business to explore the intersection of multimodal clinical and operational data with an aim of reducing the burden on physicians, enabling more personalized care, and increasing efficiency.

“With AWS, GE HealthCare plans to use the cloud to deliver more personalized, intelligent, and efficient care,” said Matt Garman, CEO of AWS. “GE HealthCare is putting generative AI at the heart of their innovation, accelerated by the investments we have made in health care-specific cloud services and generative AI capabilities that provide best-in-class security, data privacy, and access to the latest state-of-the-art foundation models.”

Working towards more personalized care

The foundation models that GE HealthCare will create can be adapted or fine-tuned for specific medical tasks, said Kass-Hout, a trained interventional cardiologist who’s held various technology and medical roles with Amazon, the FDA, and the CDC, among others. He says this makes the models ideal as the basis for creating innovative applications. And because they are in the cloud, doctors could access them online from anywhere, whether from the office or remotely.

For example, one possibility is building a tool to help doctors quickly review a patient’s entire medical history to diagnose the cause of recent symptoms. A cloud-based generative AI application built using these foundation models, could collate and flag notes, records, and images for the doctor in minutes or seconds instead of the days it might take a human, helping the doctor make precise and personalized diagnoses. Similarly, because these apps would have more holistic and detailed views of patient histories, they could help clinicians detect emerging issues early and flag precision treatments.

Kass-Hout believes these foundation models and resulting applications could also lead to “significant” breakthroughs in the broader diagnosis and delivery of specialized treatments for medical conditions. For instance, he says GE HealthCare recently fine-tuned a research foundation model used on ultrasound images to identify anatomical structures without being directly trained for them—underscoring how generative AI can help speed healthcare innovations. Such technology when eventually deployed in clinical practice, could enable cardiologists to, say, easily explore the structures of a particular patient’s ailing heart. It could also reduce clinical application development cycles from years to months.

In short, the technology could usher in a new standard of care that’s less reactive and more predictive and preventive.

Using AWS security to protect patient privacy

Kass-Hout said that AWS’s cloud infrastructure and advanced machine learning (ML) services additionally provide the processing power and security needed to manage the large volume and complexity of data that health care professionals handle each day. This means GE HealthCare can develop and offer generative AI tools to its customers that comply with privacy regulations like the Health Insurance Portability and Accountability Act (HIPAA).

Kass-Hout, who has spent decades in health care, says the ability to collect, aggregate, analyze, and use all of a medical organization’s data has been a pipe dream for most of his career. But with recent advances in the models underlying generative AI, he’s confident the dream is about to become a reality.

“We haven’t seen progress like this since the emergence of the internet,” he said. “These foundation models hold the potential to revolutionize health care data and make precision health care analytics as universal as the web.”

To learn more about GE HealthCare’s digital solutions and healthcare transformation, visit here.

For more on AWS’s generative AI capabilities head here.