Mobile Health Fingerprinting
November 5, 2012
Nowadays there isn't a job a mobile device can't handle, or so it seems. A new field called mobile health, or mHealth, introduces maybe the most useful to date.
Dartmouth's Institute for Security, Technology, and Society is among those developing mobile devices and portable sensors that can be tucked in everything from bracelets to necklaces to wrist watches in order to monitor an individual's health. This new technology is providing an unprecedented level of access to healthcare and helping doctors with early detection—monitoring electrocardiogram signals from a heart monitor, obtaining glucose readings from a glucose meter, or controlling insulin injections, for example.
But these devices aren't foolproof—not yet anyway. That's where Thayer Assistant Professor of Engineering Ryan Halter comes in. Halter is busily working with Dartmouth Professor of Computer Science David Kotz '86 and graduate student Cory Cornelius '07 to ensure that health records of those who are being monitored with mHealth devices will remain private, and safe from security breaches.
Specifically, he went searching for a way to prove that the right medical data is paired with the right patient and not just the person who happens to be wearing the device. What Halter and his team found was a way to analyze the electrical properties of the tissue—or the bioimpedance—of the person wearing an mHealth device.
"Most people are already carrying portable phones with sophisticated computation, and some type of sensor. An insulin monitor or heart rate monitor, for example, can easily wirelessly communicate with these devices, and that information can be transferred to a database and monitored by medical personnel," says Halter. "The difficult part is if someone puts the wrong device on and their heart rate speeds up. There could be a conflict with who is actually treated."
Halter's team discovered they could identify a patient by placing electrodes into the mHealth sensor to measure how easily and accessibly currents flow through their body. Data linked to those currents could then also be sent via cell phone to the medical personnel.
"A person's bioimpedance is not unique across entire population, but usually is within a small group, such as a family who might wear the same device," Halter explains.
Halter and his team completed a successful test run of this electrical fingerprinting technique in a bracelet created by Cornelius on a test group of 50 Dartmouth students last spring. Ninety percent of those individuals were accurately identified using Halter's method. Next, Cornelius, Kotz, and Halter plan to test the robustness and wearability of the identification sensors on a larger sample over a longer period of time. Eventually, they hope to use the bracelet with mHealth devices, specifically one prototype in the works by Dartmouth's Institute for Security, Technology, and Society called Amulet.