The American Heart Association and the Duke Clinical Research Institute on Monday unveiled a partnership to try out machine-learning techniques to find new ways to analyze data and drive scientific discoveries and treat cardiovascular diseases.
The two organizations will use the heart association's cloud-based Precision Medicine Platform—hosted by Amazon Web Services—which includes data related to cardiovascular health from the Duke Clinical Research Institute, AstraZeneca, Intermountain Healthcare and others.
"This allows us to bring data from all different sources together, which makes it easier for scientists to get access to the data," said Jennifer Hall, chief of the heart association's Institute for Precision Cardiovascular Medicine.
So far, the association has facilitated the exchange of data from 23 million people via the platform.
With this new partnership, scientists from the organizations hope to accelerate how they learn from the data, which come from epidemiological studies, imaging, wearables and other sources.
Researchers might apply machine learning and artificial intelligence to the data to look for new drug targets and pathways, Hall said. Those might sometimes come from fields such as immunology or cancer research rather than cardiology, she said.
"It's through this type of discovery that we can break down the walls and start exploring outside our field," Hall said.
As the organizations put their energy into digital technologies, they'll also keep an eye on the analog, also focusing on people and processes, including vetting results, Hall said.
That vetting is particularly important in healthcare, according to Mark Morsch, vice president of technology for Optum. "It's not just about getting the right answer—it's about being able to explain the answer and show the evidence behind it."
Morsch expects other organizations will attempt similar projects. Earlier this year, GE Healthcare and drugmaker Roche announced they would work together to apply machine learning for clinical decision support.
AI is a favorite buzzword in healthcare lately. But progress has been slow, with healthcare organizations struggling to gather enough clean data to train machine-learning algorithms. They're also wary of adopting an untested technology.