Coarctation of the aorta (CoA) is a relatively common congenital heart defect that narrows a major artery connected to the heart, affecting thousands of newborns each year and connected to serious issues like hypertension, stroke and heart failure. Now, a large team of researchers is using a variety of cutting-edge technologies – like machine learning, 3D printing and supercomputing – to evaluate risk factors and activities for people with CoA.

The researchers began with high-resolution simulations of an aorta affected by stenosis (which narrows the left side of the heart), aiming to understand the impact of various activities, conditions and lifestyles – such as elevation or pregnancy – on an aorta under varying degrees of stress. 

“You can take these simulations and really understand the realistic range of effects on people with this condition, beyond the factors present when the patient is sitting at rest in a doctor’s office,” said Erik Draeger, a computer scientist at Lawrence Livermore National Laboratory (LLNL), in an interview with LLNL’s Jeremy Thomas. “It also describes a protocol where, although you still need to do simulations, you don’t need to do all the configurations there are. One of the things that’s really interesting about this type of study is that, until you can do this level of simulation, you have to go by average results. Whereas with this, you can take an image of the aorta of that specific person and model the stress on the aortic walls.”

To run the staggering number of simulations required to obtain these meaningful results, the researchers needed efficient code and powerful computers. Amanda Randles, a professor of biomedical sciences at Duke University and winner of the Grace Murray Hopper Award from the Association of Computing Machinery (ACM), rewrote HARVEY, the fluid dynamics software used in the research, to optimize it for use on the supercomputer of choice: Vulcan. Hosted by LLNL, Vulcan is an IBM Blue Gene/Q system equipped with IBM PowerPC A2 processors across nearly 25,000 nodes. While Vulcan was decommissioned in 2019, its 4.3 Linpack petaflops of performance still placed it 58th on the most recent Top500 list of the world’s most powerful publicly ranked supercomputers. All in all, the team clocked over 70 million compute hours on Vulcan.

Using the results from the simulations, the researchers built a predictive model powered by machine learning, with the aim of enabling the creation of risk profiles for real-world patients.

“The ideal is that in the future, when a new patient comes in you wouldn’t have to run 70 million compute hours, you would only have to do enough to get those few simulations,” Randles said. “It’s the first step to not requiring a supercomputer in the hospital. We want to be able to give enough training data and a machine learning framework they can employ to do just a few simulations that maybe would fit on a local cluster or something much more accessible, while also leveraging results from the large-scale supercomputing.”

After validating the simulation results using 3D-printed models of the same conditions, the team is looking ahead to the future. For them, that means running more simulations using the same framework, but different diseases – like coronary artery disease.

Header image: a simulation of arterial blood flow. Image courtesy of LLNL’s Liam Krauss.

To learn more about this research, read LLNL’s Jeremy Thomas’ article here.