Smart Sock

Project Overview

Development of a novel smart sock for epilepsy monitoring using e-textile technology in collaboration with Instituto de Telecomunicações, University of Aveiro, and Meia Mania Lda. This project aimed at improving patient outcomes and quality of life through continuous and unobtrusive seizure monitoring.

Motivation 

Wearable devices offer potential benefits for epilepsy management, especially for drug-resistant patients. However, current options are obtrusive and may have limited long-term acceptability, thus smart socks have the potential to provide patients with a discreet and less intrusive alternative for epilepsy monitoring.

Highlights of the Project

1. Designed and validated a proof-of-concept wearable device for epilepsy monitoring.

2. Incorporated Electrodermal Activity (EDA), Photoplethysmography (PPG), and motion sensors (tri-axial acceleration sensor).

3. Explored the feasibility of conductive textiles as EDA electrodes and ankle-based PPG monitoring.

4. Evaluated the acquired signals from the foot compared to the hand (reference location).

5. Presented the research prototypes at various events [1] [2] and received the award "Best Poster by Accenture" at the entrepeneurial event E.Awards@Técnico 2022, where I present an entrepeneurial project based on the Smart Sock [3].

Here are two pictures of the concept of the prototyped smart sock and its biomedical sensors:

• The sensorized sock has EDA electrodes made of e-textiles that extend down to the medial arch. Note that this foot location is an alternative body location for EDA monitoring and is reported in the literature as having similar responsiveness to emotional stimuli compared with the hand. 

• The processing unit contains all hardware components for signal acquisition and processing, and integrates the acceleration and the PPG sensors. As shown, the PPG sensor fits through the connector "sock-unit interface" that is permanently crimped on the e-textile.  Moreover, it establishes the electrical connectivity with the EDA electrodes through a two-pin connector. 

As for the prototypes, I developed one with EDA electrodes made of 1) conductive lycra located at the medial arch (light gray sock), and another made of 2) conductive ink located at toes I and II (dark sock; toes were considered as a second alternative site for EDA monitoring within the foot). Both these prototypes have the electromechanical attachment of the processing unit on the side of the ankle for PPG monitoring. 

To validate the prototypes, more than 30 signal acquisitions were performed in different test subjects under low movement conditions. The experimental setup is shown below, and consists of the smart sock optically synchronized with a BITalino board, which performs simultaneous reference EDA acquisitions at the hand (i.e. standard location for reference acquisition).

The foot-EDA signals acquired from the smartsock (1) (orange signal) were visually very similar to the ones acquired from the hand (blue signal). 

If we consider only the phasic Skin Conductance Response (SCR) of the EDA signal, then, for all tested subjects, varying Pearson Correlation values were obtained for selected time windows corresponding to detected EDA peaks, as shown in the boxplot below. Surprisingly, for some subjects the obtained correlations were approximately equal to 1.0 for all extracted windows, whereas in some subjects negative correlations were obtained, which can be interpreted as high inter-subject variability regarding the similarity between hand- and foot-EDA signals.

Comments

EDA

EDA signals are valuable means to gather information about the state of the Autonomic Nervous System (ANS). However, the ANS has highly complex neuronal pathways leading to differently innervated body parts. It is also known that not all SCR events occur in all body parts and that they may occur with a temporal lag in different body parts (*). Moreover, the Pearson correlation metric (which is commonly used when comparing simultaneously acquired EDA signals) does not account for delay in events between the two time-series. Therefore, the fact that low and even negative correlations were obtained should not discourage further research exploring the foot as a body location for EDA monitoring.

For instance, even if the signals have opposite trends in amplitude, it does not necessarily mean that the foot-EDA is worse or unusable for detecting emotional or health states. Without confirming this, I would risk suggesting it could be the case that EDA signals measured at the foot provide a different type of information that would not be obtainable from the hand. In other words, it is possible that foot-EDA signals can potentially enrich and diversify the information that is extracted from hand-EDA signals when studying the ANS. 

PPG

As for the PPG signals acquired from the ankle, they displayed similar waveforms compared to the PPG signals acquired from the index finger. The Heart-rate estimation error obtained for this body part was clinically significant, but I used an off-the-shelf pulse sensor, I would not disregard the possibility to use a tuned PPG sensor to overcome this issue, which I'm working on! 

(*) I like to think of this as if the neuronal impulses descending from the central nervous system down the spine do not do so or spread evenly in the afferent nerves of the limbs. Possibly, these impulses may arrive with different intensities in different body parts, and the eccrine sweat glands there may also be triggered differently (i.e. different sensitivities to cholinergic stimulation).

Ongoing Tasks

• Design of an EDA sensor with improved measurement performance.

• Design of a PPG sensor with improved sensitivity.

• Miniaturization of the acquisition unit.

Please note that this project is currently a work in progress, and additional project materials will be available soon.  In the meantime, you can check out a detailed description of the project and read the latest publication from this year.