When Dr. Ary Goldberger looks at an EKG, he doesn’t just see the signal of a beating heart, he also sees a musical score, with its diversions and sudden shifts in tempo. Or at least that’s what he’d like to see. Because, according to Goldberger, an EKG that lacks such variations reveals a sick heart, which can’t respond to the body’s demands.
Back in the early 1980s, Goldberger, then a young cardiologist at the University of California San Diego, expected healthy hearts to beat in steady, metronomic patterns. But his data showed that only diseased hearts beat this way, while healthy patients generated unpredictable EKG patterns.
Goldberger, now a professor at Harvard Medical School, concluded that heart disease triggers such simplicity. In retrospect, he said, it’s not surprising that health requires complexity. “This makes sense when you consider that healthy physiology needs to be nimble and adaptive. It’s only sick, aging, or premature systems that get locked into overly rigid patterns.”
Learning from Complexity
Over the intervening decades, links between signal complexity and health have been demonstrated for many other biological measures, including brain waves, breathing rate, balance, and gait. More than scientific curiosities, these measures offer powerful medical opportunities, Goldberger believes. “The body’s output signals are highly predictive,” he said. “What we’d like to do is probe those signals for encoded information telling us that the body’s physiology is about to drive off a cliff.”
Phys Rev Lett 2002; 89 : 068102
Goldberger, a Wyss Institute core faculty member, is working with other members of the Anticipatory Medical and Cellular Devices Platform to develop algorithms and other computational tools to expose when signal complexity is breaking down. Goldberger said he couldnÕt do the work alone. The non-linear equations needed to map changes in biological systems are so complex, he said, that it makes a huge difference to involve physicists – who are more familiar with such modeling.
“That’s what the Wyss Institute is all about: uniting scientists around shared problems,” Goldberger said. “We’re developing and refining computational tools to pull useful information out of extremely complex data sets. That’s a huge challenge that exceeds the capacity of conventional science and bioengineering as usual.”
Potential Applications
Goldberger hopes to use these tools to prevent sleep apnea — a dangerous pause in night-time breathing that often kills premature infants and leaves exhausted adults prone to accidents, heart attack, and stroke.
Goldberger has found that the best predictor for sleep apnea is an EKG signal. Reflecting cardiopulmonary coupling, EKG signals change in response to the amount of air in the lungs. By mapping those changes on a sleep spectrogram, Goldberger and his colleagues assess how the heart and lungs talk to each other during the night. With those data, they’ve developed algorithms that detect when breathing rates get unsteady. Now measured with paste-on electrodes attached to the body, Wyss Institute faculty members are developing miniaturized, wireless systems that could monitor EKGs for apnea detection remotely, in patientÕs homes.
In another line of research, Goldberger is collaborating with the United States Army Institute of Surgical Research, at Fort Sam Houston, Texas, to triage badly injured trauma patients. Working with Army Surgeon Lee Cancio, as well as Madalena Costa, an instructor in medicine at Harvard Medical School, Goldberger is trying to identify how characteristic losses in EKG complexity—measured with wearable sensors on a soldierÕs uniform—might predict an acute risk of death.
“The more a soldier loses heart rate complexity, the more in need he or she is of life-saving care,” Goldberger explained.
Goldberger said he’s excited about sharing these efforts with other Wyss researchers. “We’re just glimpsing the future of anticipatory medicine,” he said. “The Wyss Institute promises to provide a Galilean new window into worlds that are not currently visible with conventional monitoring tools.”