A mathematical analysis of heartbeat patterns can detect certain heart disorders with unerring accuracy. Previous mathematical tools for diagnosing heart disease have generally not been reliable enough to use clinically. But the new technique, which is based on variations in the time interval between heartbeats, had a 100% success rate in a test aimed at identifying patients whose hearts are not pumping properly.
Electrical engineer Malvin Teich and his colleagues at Boston University analyzed about 20 hours of heartbeat records from 15 patients who had heart failure and 12 healthy controls. The team applied a mathematical transformation that measured how much the time between beats fluctuated, a measure known as the wavelet coefficient, to sets of anywhere from two to more than 1000 heartbeat intervals. As they report in the 16 February Physical Review Letters, they found that the wavelet coefficients for batches of 16 to 32 beats were clearly different for healthy versus sick hearts, and there was no overlap at all between the two groups. Curiously, every healthy heart showed larger beat-to-beat fluctuations than the failing ones did.
The technique is so simple and foolproof that it "can be used right away" in hospitals to diagnose heart conditions, says Teich, who is now collaborating with cardiologists. However, he notes that the method still needs to be tested with larger numbers of patients. He also hopes to get a physiological understanding of why the window of 16 to 32 heartbeat intervals so neatly distinguishes between normal and damaged hearts.
"A 100% result is unusual and very remarkable," says Ming Zhou Ding, a biophysicist at Florida Atlantic University in Boca Raton, who uses wavelet analysis to study brain rhythms. Teich and his colleagues attribute their success to the technique's ability to pick up both short-term and long-term fluctuations in heart rhythms, as opposed to previous methods, which focused either on neighboring heartbeats or on long-term changes.