Scripps, NVIDIA to partner on deep learning research

Scripps Research Translational Institute and NVIDIA will collaborate to create deep learning tools aimed at deriving actionable insights from genomic and medical sensor data.

The initiative aims to bring artificial intelligence to process large amounts of data that the partners expect to be gathered through the emerging use of such data in diagnosis, personalized treatment approaches and preventive care.

The partners will focus work at Scripps’ La Jolla, Calif., facility on efforts to apply AI to vast stores of data on individuals. Research will initially focus on developing deep learning-based genetic and digital sensing prediction of atrial fibrillation, an irregular heartbeat which increases the risk of stroke, along with analytics of whole genome sequences, with later expansion to other diseases and datasets.

The goal is to find insights from analyzing populations that can be then applied back to the care of individuals.

For example, using data gathered from tests that continuously measure blood pressure could provide new insights, such as indicating warning signs that an individual is at risk for a stroke, says Eric Topol, MD, founder and director of the Scripps institute and a professor at Scripps Research. That could enable physicians to intervene hours in advance of a medical emergency, thus heading it off.

But sensors that continually track blood pressure over long periods of time produce large amounts of data, and it’s challenging to analyze data from thousands of individuals, Topol says. That’s where NVIDIA’s experience with large amounts of data comes in.

Topol-Eric2-CROP.jpg

Merging technology development with leading-edge scientific research on the use of data in medicine will be critical to this effort, Topol adds. “AI has tremendous promise to transform the future of medicine; with NVIDIA, we aim to establish a center of excellence for artificial intelligence in genomics and digital sensors, with the ultimate goal of developing best practices, tools and AI infrastructure for broader adoption and application by the biomedical research community.”

The use of AI systems is medicine is largely limited to analyzing medical images. Preliminary studies, however, suggest that deep learning techniques could also be applied to big data of whole genomic sequences and continuous physiologic sensors. In deep learning, machine learning happens in layers, forming neural networks with each layer adding to the knowledge of the previous layers.

NVIDIA has helped pioneer the spread of AI across a growing range of fields, including self- driving cars, robotics and healthcare. For example, earlier this year, it started a new initiative to make the power of supercomputing available to legacy imaging scanners currently being used by healthcare organizations. Called Project Clara, the effort is intended to develop a virtualized data center that would be available on a remote basis, able to handle the images from multiple modalities and by several users at the same time, using a medical imaging supercomputer.

“AI is already transforming healthcare by using electronic health records and medical imaging to better diagnose and treat disease,” says Kimberly Powell, vice president of healthcare at NVIDIA. “Our collaboration with Scripps expands these opportunities by tapping into the rapid accessibility of genomic and digital wearable data, and furthers the quest to better predict and prevent disease.”

For reprint and licensing requests for this article, click here.