An algorithm for the heart: PhD student’s work could speed cardiac research

John Jurkiewicz’s algorithm can isolate valuable experiment data in a fraction of the time that scientists previously needed to do the work manually.

A doctoral student in applied mathematics has developed an algorithm that will allow heart cell researchers to get their results in minutes, rather than hours.

John Jurkiewicz is involved in a partnership between UWM’s Department of Mathematical Sciences and the Advocate Aurora Research Institute. His algorithm can isolate valuable experiment data in a fraction of the time that scientists previously needed to do the work manually.

The Advocate Aurora lab studies heart cells, using them to measure cardiac cell function, aid in developmental studies and, eventually, help conduct drug testing. Their work involves giving the cells an electric shock and recording what happens. But that’s when things get tricky.

It’s difficult to separate the electrical signal of activity across some tissue (called field potential, or FP) from the activity of each individual cell (called action potential, or AP). The frequencies connected to action potential are the valuable ones—a heart cell’s AP can say quite a bit about its activity and health.

Before Jurkiewicz created his algorithm, Advocate Aurora lab members had to weed out those valuable signals by hand. It would take an afternoon to parse out one day’s experiment, and they were faced with 1 terabyte of data. “They came to us and said, ‘Can you help?’” says Jurkiewicz, whose faculty advisor is Peter Hinow, mathematics professor in the College of Letters & Science. “So we came up with an algorithm that can segment this data stream automatically.

Originally appeared March 8 in 2021 UWM Research.

Learn about Mathematical Sciences graduate programs at UWM.

“It turns out the AP versus FP classification is ‘easy’ for mathematicians to do,” says Jurkiewicz, whose paper on the work was accepted for publication in Journal of Electrocardiology. And he made a point to run the algorithm on his consumer-grade laptop so researchers wouldn’t need a supercomputer to use the tool. He also wrote the program in the open source language Python.

“We would like to make this as accessible as possible,” Hinow says.