Wearable ID system could pave way to passive mHealth interoperability

Researchers at Dartmouth College in Hanover, N.H., have demonstrated the feasibility of a wearable sensor that they designed to passively recognize people. If successful, the prototype they hope to build from their study will curtail mistaken-identity errors in, and malicious hacking of, mobile and wearable medical devices.

Led by Cory Cornelius, a PhD candidate in the school’s department of computer science, the team presented a paper Aug. 7 describing the way their bracelet-like sensor uses bioimpedance—the physiological response to electric current passing through human tissue—to give mHealth sensors the ability to verify the wearer’s identity.

The team studied their approach on 46 adult subjects and found a wearer-recognition accuracy rate of at least 85 percent. Accuracy reached 90 percent when they combined the method with measurements allowing for an error of 1 mm in wrist circumference.

The all-Dartmouth team included four computer scientists, an engineer and an adjunct assistant professor of surgery from Dartmouth’s Geisel School of Medicine.

If an mHealth system knows the identity of its wearer, “the system can properly label and store data collected by the system,” they wrote. “Existing recognition schemes for such mobile applications and pervasive devices are not particularly usable—they require active engagement with the person (e.g., the input of passwords), or they are too easy to fool (e.g., they depend on the presence of a device that is easily stolen or lost).”

The group presented the paper, “Who Wears Me? Bioimpedance as a Passive Biometric,” at a Usenix workshop on health security and privacy.