UPMC and other prominent Pittsburgh research organizations announced Wednesday they plan to leverage an Amazon division's machine learning capabilities to accelerate breakthroughs in patient care and product commercialization.
Amazon Web Services will share its machine learning—a type of AI—and cloud computing resources with the Pittsburgh Health Data Alliance, a big data consortium formed in 2015 that includes UPMC, the University of Pittsburgh and Carnegie Mellon University.
"As it relates to cloud computing and where it can go, we see AWS as the leader in this space," said Rob Hartman, director of translational science for UPMC Enterprises.
PHDA collects deidentified patient data from a number of sources including electronic health records, imaging records, prescriptions, genomic profiles and insurance records. That data is used in an effort to transform how diseases are treated and prevented. New technologies like those Amazon Web Services offers can rapidly translate discoveries into treatments and services to improve human health, the organization said in a statement.
Hartman said it's not a formal partnership governed by a contract, but that could be the case in the future. The time duration is open ended.
Amazon Web Services does not receive commercialization rights in exchange for its contribution to any research that ultimately turns into products or services, Hartman said. For Amazon Web Services, the draw may be simply attaching its name to cutting-edge research and ensuring such developments rely on its products, Hartman said.
"There's a lot of wins for them by getting in early and supporting innovative activity in healthcare," Hartman said.
Amazon Web Services' goal is to enable research that accelerates the development of innovative algorithms, publications and source code across a variety of machine learning applications and focus areas, company spokeswoman Cortney Lusignan wrote in an email.
Researchers involved have identified eight projects they will focus on under the arrangement, all of which are designed to impact different areas of healthcare, said Zariel Johnson, UPMC Enterprises' program manager technical.
One such project seeks to develop an algorithm to predict the best surgical interventions for abdominal aortic aneurysms before symptoms appear. Currently, doctors measure an aneurysm's diameter and growth rate to predict the risk of rupture. Another project seeks to use machine learning to predict how tumors are likely to grow in the future.
Amazon Web Services also announced a multiyear research sponsorship with Beth Israel Deaconess Medical Center in March.
The entrance of artificial intelligence into healthcare is one part of the world's fourth industrial revolution, said Pam Arlotto, CEO of the consultancy Maestro Strategies. This revolution combines physical, data and biological assets with digital capabilities, she said.
"When you pull it all together and use some of these machine learning algorithms, it could just take us into the next stratosphere," Arlotto said.
