Solving medical mysteries in the AWS Cloud: Medical data-sharing innovation through the Undiagnosed Diseases Network

For many patients, the National Institutes of Health’s Undiagnosed Diseases Network (UDN) is their last hope for finding information. The UDN is a network of doctors and researchers who specialize in diagnosing the tens of millions of people who live with rare diseases for years, or even decades, without a name for their symptoms. Clinicians and researchers of the UDN work together to identify and treat these mysterious conditions.

It takes a medical village to discover and diagnose these rare diseases. To achieve their goal, the UDN is made up of a coordinating center, 12 clinical sites, a model organism screening center, a metabolomics core, a sequencing core, and a biorepository. For many years prior to the UDN, the experts at these sites were limited by antiquated data-sharing procedures. The UDN leadership realized that if they wanted to scale up and serve as many patients as possible, they needed to transform how they process, store, and share medical data—which led the UDN to the Amazon Web Services (AWS) Cloud.

Creating a seamless flow of data between the clinic and the lab

The UDN has two primary goals: to provide answers for patients and families through their medical clinics and to learn more about rare diseases through research. In the medical system more broadly, databases are typically separated—one for clinical work and one for research. But such a system sets limits on what both clinicians and researchers can learn. The experts in the UDN knew that compiling the datasets would allow them to inform one another and even help identify patients with similar symptoms earlier, so they can receive a diagnosis. “From the beginning, we knew we wanted to have all this data in one place,” said Dr. Paul Avillach, assistant professor at Harvard Medical School, who participated in the migration to an AWS Cloud-based dual-model platform, which combines research data with a patient portal. “It was a unique opportunity because we had the funding, knowledge, and willingness of UDN to make this happen.”

Through working with AWS, the UDN designed a custom platform that combined the research database with both a clinician and patient-facing portal. Today, doctors, patients, and researchers all log in to a single platform, and patient medical information is shared seamlessly between researchers and clinicians—all with the patient’s consent. Additionally, building in the AWS Cloud makes sure that all health data shared on the platform is secure and compliant, so patients can rest assured their information is safe.

“This platform is also a huge improvement on the patient side,” added Kimberly LeBlanc, genetic counselor and director of the UDN coordinating center. “Having the data together allows us to know more about our patients. So much of healthcare can benefit from bringing these data sets together. As clinicians, we can learn from our data and improve care as a result.”

Scaling up data-processing to find answers

After the success of their cloud-based platform, UDN leaders realized they could now leverage the AWS Cloud in another way—by scaling up their data-processing capabilities to analyze more genetic sequencing data, faster.

A pivotal part of the UDN’s approach is a data-processing technique called joint variant calling, in which genetic sequencing data is compared against hundreds of other samples to find similarities. When a new diagnosis is made, joint variant calling can help researchers determine if other patients in the UDN system might suffer from the same disease. “If these patients have common variants,” LeBlanc explains, “you can then make discoveries about undiagnosed conditions. Patients have been waiting years and years for this information. These aren’t easy cases. But with joint variant calling, we can compare these patients and find things in common.”

In the past, these types of analyses were conducted via on-premise architectures, which were time-consuming and inefficient. The team leveraged AWS Cloud-based tools that process bioinformatics and genomics data to analyze more than 800 samples at once – an impossibility in their previous on-premise environment. “AWS enabled us to process this data in a timely and much more scalable fashion,” said Dr. Avillach. “Right now we have 800 samples, but in a few years, it could be 5,000 or even 10,000. Moving to the AWS Cloud opens up the opportunity to scale UDN’s services to another order of magnitude.”

Changing patients’ lives with cloud-based data-sharing

For years, six-year-old Elizabeth Nagorniak failed to meet developmental milestones, which baffled doctors and troubled her mother, Mari Hanada. After more than three years without answers, Hanada turned to the UDN team, who diagnosed Elizabeth with a variant of Smith-Kingsmore syndrome, a rare condition connected to a mutation of the MTOR gene. With a diagnosis in hand, Hanada was able to seek treatment for her daughter. “She’s getting new skills weekly now,” Hanada told reporters. “It used to be annually.”

Elizabeth is just one of the hundreds of patients diagnosed by the UDN. And now that the UDN is working with AWS, the number of patients they evaluate can grow—so even more people can find a name for their condition and a path toward healing. “We are working with AWS to create the infrastructure that enables us to do better research and connect it with better clinical care,” Dr. Avillach says. “This is a ground-breaking approach. It’s truly a model of how research should be done in the future.”

Learn more about how AWS can help researchers innovate in healthcare and beyond at the AWS for research and technical computing hub, or contact the AWS Research team directly.