The Heart Knows—ECG-Based Biometrics | CardioSource WorldNews
Health Tech | “What the heart knows today, the head will understand tomorrow,” wrote Irish poet James Stephens. Now, flash forward a few years and a company called Bionym has built an ECG-based biometric device called the Nymi, which uses heart rhythm for authentication purposes. We had the opportunity to speak with the founder and CEO of Bionym, Karl Martin, about the device and how ECG biometrics works.
How did you develop the idea for Nymi?
In 2011, our company was founded on the core invention of ECG biometric recognition—this research having come out of the University of Toronto. We were originally developing the technology to be licensed to other companies in a variety of markets: military, corporate IT, consumer electronics, etc. In most cases, the technology was seen as just a “better” or “different” fingerprint or face recognition. However, in early 2013, we realized that the potential was far greater. Rather than just building a better mousetrap, we realized that if we embed the technology into a wearable form, we could use it to enable persistent authentication, which could subsequently enable a whole host of new applications and experiences.
Can you describe the technology behind Nymi?
There are two main technological innovations that make the Nymi possible. The first is a set of algorithms for ECG biometric recognition. These algorithms can take a raw ECG signal and turn it into a robust biometric template for identity authentication. The system is able to extract the features from the signal that make each person unique, and are also consistent from day to day.
The second innovation is the system integration of an ECG sensor, hardware-based cryptography, Bluetooth, and motion sensing, which is used for simple gesture recognition for user input. We have developed a system and protocol that allows the Nymi to be used by a trusted provider of the user’s identity that is both secure (cannot be spoofed) and privacy protective (cannot be read by an outside device or party without explicit opt-in permission from the user).
If the ECG sensor just relies on one point of contact on one wrist, how does the sensor work? Can you describe what leads it may be picking up?
The ECG sensor on the Nymi is a single-lead system with two electrodes. One electrode is on the underside of the wristband, contacting the user’s wrist. The other electrode is on the topside. After the user first puts on the wristband, they touch the top-side electrode with their opposite hand. ECG is captured and their identity is authenticated within a few seconds. After this, authenticated identity is maintained with additional sensors that ensure that the wristband has not been removed from the wrist. ECG is not captured continuously, since it requires the two points of contact; however, we have an [application programming interface] that allows ECG to be captured, for any application, when the user touches the top-side electrode.
Can you describe which features of the rhythm strip are most unique?
Our feature extraction algorithms do not look at time-based features (intervals, amplitudes, etc.) since they are so variable. We extract what are called second-order statistics—essentially the overall shape of the waveform (not particular points). This makes it robust against changes in heart rate.
How do ECG-based biometrics account for natural variation in individual rhythm or pathology? Does the device need to be recalibrated every so often?
Our feature extraction—as well as our proprietary machine learning system—is designed to tolerate natural variations. We have data for a broad, uncontrolled population (with and without cardiac anomalies) that show this. However, in the case of a sudden, significant event, such as a myocardial infarction, it may be necessary for the user to re-enroll. This is similar to other biometrics, such as face (after a palsy or other appearance changing event) and fingerprint (after physical manipulation of the fingerprints). On the other hand, gradual changes can be tolerated and learned by the system.
There are a number of wearable technologies being developed ranging from the Nike Fuelband to the Samsung smart watch. How do you anticipate Nymi fitting into this ecosystem? Do you plan on developing your own enhanced wearable?
Long term, the Nymi will encompass other form factors beyond the wristband (e.g., necklace, waistband, etc.) and can be used as a generic wearable sensor platform. We have the same sensors as the other activity trackers, along with a very open and developer-friendly system that allows others to write similar applications. Additionally, we’re not betting on any single wearable becoming ubiquitous (the smart watch, for example), necessitating that we license our technology to those manufacturers. We haven’t discounted the possibility that we may form partnerships that see our technology embedded into a third-party product, but it’s not on the roadmap right now.
What are the various use cases for the device?
The Nymi is intended to be a tool that not only replaces all forms of identity authentication today (passwords, PINs, key cards, keys, ID documents, etc.), but to enable new hyper-personalized experiences, such as personalized retail, hospitality, smart home, etc. We have an active developer community that is coming up with new applications every day. Essentially, we’re putting the question out to the world: what would you do if identification was easy?
One of the health care-focused applications that has generated interest is the concept of “touch-free” computer interaction for health care professionals. For example, a doctor may need to access medical records or other medical systems in a sterile environment. The Nymi supports touch-free login and interaction via the gesture recognition capabilities.
Shiv Gaglani is an MD/MBA candidate at the Johns Hopkins School of Medicine and Harvard Business School. He writes about trends in medicine and technology and has had his work published in Medgadget, The Atlantic, and Emergency Physicians Monthly.
< Back to Listings