New AI outperforms supercomputers: solves scientific problems faster than ever

The team tested DIMON on more than 1,000 digital computer heart models of real patients.

Researchers from Johns Hopkins University created a new AI model called DIMON. It can solve complex equations thousands of times faster, just on a personal computer. Interesting Engineering writes about it.

Partial differential equations are common mathematical problems. They help transform real-world scenarios into mathematical models to predict future changes in objects or environments. However, solving these large mathematical problems is usually a job for supercomputers. Things are getting easier with the advent of artificial intelligence (AI).

A new AI model, DIMON — Diffeomorphic Mapping Operator Learning, — was tested on more than 1,000 digital computer models of the hearts of real patients. It accurately predicted the paths of electrical signals through different structures of the heart.

In a demonstration, researchers used partial differential equations to study cardiac arrhythmia. This happens when the human heart beats irregularly due to distorted electrical signals.

A new AI significantly speeds up the prediction of heart disease, reducing calculation time from hours to 30 seconds. It can be done using a simple computer, making it more practical for everyday clinical use. This demonstrates the potential of AI for various engineering applications, including medical diagnostics.

DIMON uses AI to predict the behavior of physical systems in various shapes. It learns patterns to predict factors such as heat, stress and motion, which speeds up tasks such as design optimization and shape-specific modeling.