Modeling how cars deform in a crash, how spacecraft respond to extreme environments, or how bridges resist stress could be done thousands of times faster thanks to new artificial intelligence that enables computers people to solve huge mathematical problems that usually require supercomputers.
The new AI framework is a generic approach capable of quickly predicting solutions to the ubiquitous and time-consuming mathematical equations needed to create models of the propagation of fluids or electrical currents through different geometries, like those involved in standard engineering testing .
Search details appear in Nature Computational Science.
“This is a solution that we believe will generally have a massive impact on various areas of engineering because it is very generic and scalable.”
Natalia Trayanova
Professor of biomedical engineering and medicine
Called DIMON (Diffeomorphic Mapping Operator Learning), the framework solves ubiquitous mathematical problems known as partial differential equations that are present in almost all scientific and engineering research. Using these equations, researchers can translate real-world systems or processes into mathematical representations of how objects or environments will evolve over time and space.
“While the motivation to develop it comes from our own work, it is a solution that we believe will generally have a massive impact on various areas of engineering because it is very generic and scalable,” said Natalia Trayanovaprofessor of biomedical engineering and medicine at Johns Hopkins University who co-led the research. “It can work on basically any problem, in any field of science or engineering, to solve partial differential equations on multiple geometries, such as in crash testing, orthopedic research or d “other complex problems where shapes, forces and materials change.”
In addition to demonstrating the applicability of DIMON to solve other engineering problems, Trayanova’s team tested the new AI on more than 1,000 cores. “digital twins” highly detailed computer models of the hearts of real patients. The platform was able to predict how electrical signals propagated through each unique heart shape, achieving high prognostic accuracy.
Image caption: DIMON revolutionizes modeling by eliminating the need to recalculate grids each time they change shape. Instead of breaking complex shapes into small elements, it predicts the behavior of physical factors such as heat, stress and motion on various shapes, significantly speeding up simulations and optimizing designs.
Picture credit: Minglang Yin / Johns Hopkins University
Trayanova’s team relies on solving partial differential equations to study cardiac arrhythmia, which is a misbehavior of electrical impulses in the heart that causes irregular beats. Using their cardiac digital twins, researchers can diagnose whether patients might develop this often-fatal disease and recommend ways to treat it.
“We’re bringing new technology into the clinic, but many of our solutions are so slow that it takes us about a week from the time we analyze a patient’s heart to solving the partial differential equations to predict whether the patient has a high risk of sudden cardiac death and what is the best treatment plan,” said Trayanova, who heads the. Johns Hopkins Alliance for Innovation in Cardiovascular Diagnostics and Treatment. “With this new approach to AI, the speed at which we can find a solution is incredible. The time needed to calculate the prediction of a cardiac digital twin will go from several hours to 30 seconds, and this will be done on a desktop rather than a supercomputer, which allows us to integrate it into the daily clinical workflow.”
Partial differential equations are usually solved by dividing complex shapes like airplane wings or body organs into grids or meshes made up of small elements. The problem is then solved on each simple part and recombined. But if these shapes change, such as during accidents or deformations, the grids must be updated and the solutions recalculated, which can be slow and computationally expensive.
DIMON solves this problem by using AI to understand the behavior of physical systems on different shapes, without needing to recalculate everything from scratch for each new shape. Instead of dividing shapes into grids and solving equations over and over, AI predicts how factors like heat, stress, or movement will behave based on the patterns it has learned, making it much faster and more efficient in tasks such as optimizing designs or modeling shapes. -specific scenarios.
The team integrates the cardiac pathology which leads to arrhythmia into the DIMON framework. Due to its versatility, the technology can be applied to shape optimization and many other engineering tasks where solving partial differential equations on new shapes is required repeatedly, said Minglang Yin, postdoctoral researcher in biomedical engineering at Johns Hopkins University who developed the platform.
“For each problem, DIMON first solves partial differential equations on a single form, then maps the solution onto multiple new forms. This shape-changing capability highlights its tremendous versatility,” Yin said. “We are very excited to use it to solve many problems and provide it to the broader community to accelerate their engineering design solutions.”
Other authors are Nicolas Charon of the University of Houston, Ryan Brody and Mauro Maggioni (co-lead) of Johns Hopkins, and Lu Lu of Yale University.
This work is supported by NIH grants R01HL166759 and R01HL174440; a scholarship from the Leducq Foundations; the Heart Rhythm Society Scholarship; The U.S. Department of Energy grants DE-SC0025592 and DE-SC0025593; NSF grants DMS-2347833, DMS-1945224, and DMS-2436738; and the Air Force Research Laboratory awards FA9550-20-1-0288, FA9550-21-1-0317, and FA9550-23-1-0445.