Information in a Heartbeat

In fact, efforts to do so by measuring unique anatomical and physiological traits — biometrics — have grown steadily in popularity. This includes technologies that scan fingerprints, retinas, and facial features.

Some of these technologies, however, have limitations. For instance, fingerprint scanning can fail in a dirty environment and facial scanning can be thwarted by disguises.

However, a new type of biometric — heartbeat patterns — is difficult to disguise because everybody's electrocardiogram (ECG) trace is unique. But information from an ECG trace is often a visual expression of traits, rather than data about the physiology of somebody's heart.

In their ESTC Award-winning article, SAIC's Steven Israel, John Irvine, and Mark Wiederhold propose a more extensive set of ECG descriptors to better characterize a heartbeat trace. The article details a series of data processing and testing experiments to show that an ECG trace contains unique information about the physiology of someone's heartbeat.

The authors started by collecting data from different individuals, then characterizing the sources of noise (extraneous signals) in a raw data stream. From this, they designed a data filter to separate the cardiac information from the noise. According to the paper, the authors identified reference points on the filtered data and digitally extracted them to define stable features (that is, the unique information that characterized individuals.) The data was then tested to ensure that location of the ECG sensors did not affect it.

Since a person's heart rate can vary with mental and emotional state, the authors developed a data collection protocol in which subjects performed a variety of tasks designed to elicit different stress levels. (Low-stress levels included the subjects' baseline state, meditative, and recovery tasks. High-stress level tasks included reading aloud, mathematical manipulations, and driving in virtual reality.) This was necessary to identify features in the heart signals unique to individuals, but not affected by a person's mental and emotional state.

In addition, the authors used the data set, including 15 attributes that characterize a heartbeat, to identify several individuals.

Researchers such as Israel, Irvine, and Wiederhold, have helped make SAIC a leader in developing and deploying new biometric technologies. Their research into heartbeat patterns is funded by DARPA and designed to detect and identify humans at a distance. The Pattern Recognition journal published their article, "ECG to Identify Individuals".