Telemedicine: RIGHT-BODY DETECTION
Lead Institution: Dartmouth
Project Leader: David Kotz
The goal of this project was to develop a biometric that could be passively measured and show that it is feasible to integrate into a wearable device. Specifically, we examined bioimpedance as passive biometric – bioimpedance is a measure of the resistance of biological tissue to tiny electrical currents. We showed that we could build a wearable device that is capable of sensing bioimpedance and demonstrate an algorithm for recognizing users wearing the device.
Focus of the research/Market need for this project
There are no wearable devices on the market that include passive biometrics in part because most biometrics are unsuitable for wearable devices. A device with this capability, however, would provide significant security benefits by ensuring that medical devices could only be used by authorized personnel, or to ensure that the data collected by the device is entered into the correct person’s medical record. We have filed a patent application for our technique and already had some inquiries from industry.
Our research looked at methods for determining whom a sensor is sensing. Such a solution is important in health-related applications where the data needs to be correctly labeled with the identity of the patient or individual being measured. If there is some mechanism for determining identity, then the collected data can automatically be labeled with that person’s identity. We conducted an initial set of in-lab experiments to validate the use of bioimpedance as a biometric.
We then prototyped a wrist-worn solution and developed an algorithm to recognize the individual that is wearing it. Our wearable device consists of a custom-designed daughterboard for the Shimmer Wireless Sensing Platform (a commercially available research-oriented device). This choice allowed us to concentrate on the sensing hardware itself while leveraging the computational (MSP430), storage (SD) and communication (Bluetooth) resources of the Shimmer platform. (Ultimately, a fully custom design would be smaller, cheaper, and more power-efficient.) In addition to the daughterboard, we also developed a sleeve with electrodes that connects the device. Using this device we then conducted an out-of-lab study. In our out-of-lab study, we instructed users to wear the device over the course of a day. We programmed the device to collect bioimpedance samples periodically from the subject. We also developed better recognition algorithms by trying new features and machine-learning classification algorithms. In concert with this new dataset, these recognition algorithms yielded better recognition rates than our previous in-lab study. We also showed that bioimpedance is potentially viable as a long-term biometric as shown in Cornelius’ dissertation.
In the course of our research we also investigated other recognition methods (a person’s wrist circumference measurements and a person’s vocal resonance, which is the sound of a person’s voice as it is heard through their chest as opposed to the open air), but these methods did not yield satisfactory recognition results.
Key Conclusions/Significant Findings/Milestones reached/Deliverables
We believe bioimpedance can become a useful biometric for integration into wearable devices. However, there remain some barriers to overcome in terms of motion artifacts and perspiration.
Materials Available for Other Investigators/interested parties
Peer-reviewed published papers, and a PhD dissertation.
Market entry strategies
We are in early discussions with a major vendor of consumer electronics, which hopes to refine and incorporate this method as a possible means for recognizing the user of their electronics. Our methods are under patent protection.
A Wearable System That Knows Who Wears It
Cory Cornelius, Ronald Peterson, Joseph Skinner, Ryan Halter, and David Kotz
In Proceedings of the International Conference on Mobile Systems, Applications, and Services (MobiSys), pages 55-67, June, 2014. DOI 10.1145/2594368.2594369
Vocal Resonance as a Passive Biometric
Cory Cornelius, Zachary Marois, Jacob Sorber, Ron Peterson, Shrirang Mare, and David Kotz
Dartmouth Computer Science Technical Report TR2014-747, February 2014
Usable Security for Wireless Body-Area Networks
Cory Cornelius; PhD dissertation
Dartmouth Computer Science Technical Report TR2013-741, September 2013
Who Wears Me? Bioimpedance as a Passive Biometric
Cory Cornelius, Jacob Sorber, Ronald Peterson, Joe Skinner, Ryan Halter, and David Kotz
USENIX Workshop on Health Security and Privacy (HealthSec), Bellevue, WA, August 2012.
Various media outlets, including Technology Review, reported on the results of this research.
Wearable Computing Device for Secure Control of Physiological Sensors and Medical Devices, with Secure Storage of Medical Records, and Bioimpedance Biometric
David Kotz, Cory Cornelius, Ryan Halter, Jacob Sorber, Minho Shin, Ronald Peterson, Shrirang Mare, Aarathi Prasad, and Joe Skinner
Patent Application (same as the Amulet patent application) 61/655893, filed December 2012