Wearable devices with body sensors have been seen by many as a way to get the general public to be better aware of their overall health. It’s a nice idea, but it still requires people to remember to wear their devices, to check up on the readings via the smartphone, and to recharge the devices when they are low on power. Some people think that wearables still have fundamental limitations that can be overcome with more passive and pervasive monitoring. Engineers at Portugal’s Institute of Telecommunications are working on integrating electrocardiography (ECG) into everyday objects that people touch long enough to get a good reading. We spoke with Ana Fred and Hugo Silva, who have worked extensively on the Bitalino project that allows people to build body monitoring devices, very similar to how Arduino is used by tinkerers making general purpose gadgets. They gave us an interesting rundown of this current project and what it means for the future of body monitoring.
Medgadget: Can you give me some more details about your current project?
Ana Fred and Hugo Silva: Electrocardiography (ECG) is an established standard medical practice and a mainstream diagnostic technique. Although the first practical implementations of devices for human use can be dated back to 1887, measurement methods are still mostly bound to body-worn sensors (or placed on the body) and short-time monitoring settings. With this project we intended to develop an approach in which the sensors do not need to be with the person, but are embedded into everyday use objects instead (ex: a computer keyboard), hence being more pervasive. A major advantage of this approach is the fact that the sensor placement does not require a voluntary action from the user, unlike, for example, a smart watch.
Medgadget: Where do you see the pervasive ECG being used?
Fred & Silva: With prevention being one of the main pillars for managing the risks associated with cardiovascular diseases management, new solutions with the potential to complement current practices, and accelerate early detection of abnormal conditions, can play a major role. We see pervasive ECG being used in all sorts of objects like a computer keyboard, the hand rest of a laptop, a steering wheel of car, and any other objects with which the subjects regularly interact with. The primary use cases are mainly related with preventive healthcare, by means of ECG trace analysis for very early stage detection of cardiovascular diseases, but the features derived from ECG signals are very rich; in particular, building upon the work on Heart Rate Variability (HRV), a number of indicators related with mental workload, fatigue, and overall affective state of the user can be extracted. As such, we also see pervasive ECG being extended to ergonomics, health and safety at work, or user-tuned personalisation of controllable workplace settings, just to name a few.
Medgadget: What are the tech requirements to implement this technology?
Fred & Silva: There are key requirements at multiple levels. At an anatomic level, due to the ECG operating principle, users need to have the left and right limbs in contact with the sensor (e.g. in the case of a keyboard implementation both hand palms would need to be in contact at the same time, even if momentarily).
At the hardware level, specific devices compliant with pervasive ECG are needed, in particular, the sensors have a specific considerations that need to be taken into account even from an industrialization standpoint; these include the use of a virtual ground to simplify the deployment or the electrode materials design, which does not necessarily need to be metallic [1].
At the software level, this is intrinsically a signal processing and big data pattern recognition problem, but the solutions are becoming stabilized. Data acquisition has an intermittence to it (e.g. periodic loss of contact due to hands off events, or noise introduced by friction in the skin/sensor interface), which requires powerful de-noising and outlier detection algorithms. Once the sensors are placed in an interface that can be shared by multiple people (e.g. a steering wheel), determining even if a rough match between the collected data and its owner can also be an important requirement [2]. Furthermore, the sheer volume of data collected that needs to be processed requires suitable frameworks to be dealt with [3, 4].
Medgadget: Anything else interesting you can share would be great.
Fred & Silva: With ECG assessment becoming more pervasive by the day, through wearables such as the Apple Watch 4 or the AliveCor toolset, the next big leap forward is to make it even a bigger part of people’s everyday lives, and in a more widespread way. This can be achieved by integrating the sensors into daily and continued use objects, rather than forcing people to adapt to technology. Furthermore, real world studies have shown that data collected under this “off-the-person” paradigm has a good correlation with clinical grade signals [5], which paves the way to a wide spectrum of applications in health and well-being.
References:
[1] https://www.researchgate.net/publication/254045981_Study_and_evaluation_of_a_single_differential_sensor_design_based_on_electro-textile_electrodes_for_ECG_biometrics_applications
[2] https://www.researchgate.net/publication/261392462_Finger_ECG_signal_for_user_authentication_Usability_and_performance
[3] https://www.researchgate.net/publication/256205837_SignalBIT_A_web-based_platform_for_real-time_biosignal_visualization_and_recording
[4] https://www.researchgate.net/publication/327393703_RepoBIT_Cloud-Driven_Real-Time_Biosignal_Streaming_Storage_Visualisation_and_Sharing
[5] https://www.researchgate.net/publication/273490686_Off-the-person_electrocardiography_performance_assessment_and_clinical_correlation