Fascia Ecosystem: A Step Forward in Sleep Engineering and Research


MIT Media Lab / Guillermo Bernal

Guillermo Bernal 

Abstract— Millions suffer from sleep disorders, and sleep clinics and research institutions seek improved sleep study methods. This paper proposes the Fascia Ecosystem for Sleep Engineering to improve traditional sleep studies. The Fascia Sleep Mask is more comfortable and accessible than overnight stays at a sleep center, and the Fascia Portal and Fascia Hub allow for home-based sleep studies with real-time intervention and data analysis capabilities. A study of 10 sleep experts found that the Fascia Portal is easy to access, navigate, and use, with 44.4% finding it very easy to access, 33.3% very easy to navigate, and 60% very easy to get used to. Most experts found the Fascia Portal reliable and easy to use. Moreover, the study analyzed physiological signals during various states of sleep and wakefulness in two subjects. The results demonstrated that the Fascia dataset captured higher amplitude spindles in N2 sleep (72.20 ffV and 109.87 ffV in frontal and parietal regions, respectively) and higher peak-topeak amplitude slow waves in N3 sleep (93.51 ffV) compared to benchmark datasets. Fascia produced stronger and more consistent EOG signals during REM sleep, indicating its potential to improve sleep disorder diagnosis and treatment by providing a deeper understanding of sleep patterns.


© 2024 Fascia Sleep. All Rights Reserved.

The Fascia Sleep Mask 

The Fascia Sleep Mask provides a comfortable, athome data collection alternative to sleep study limitations. The mask gathers polysomnogram-like data as you sleep using fabric-based sensor technologies. There are two main components to the Fascia Mask hardware: a main sensor PCB, and an ergo-electronics face pad. The main sensor PCB is an analog front end that collects biopotential signals such as EEG (Fp1, Fp2, Cz, Pz), EMG, EOG (vertical and horizontal), Electrodermal Activity (EDA), heart rate (PPG), skin temperature (PPG), and respiration (PPG, IMU). The core technology for sensing has been previously published in the following publication [23].The signals from the facepad are wirelessly streamed to a remote broker using MQTT protocol. The facepad is made up of a flexible PCB with goldplated electrodes and a screen-printed electrode array on a breathable fabric cushion for comfort. Figure 2 section 1, depicts one of the Fascia Sleep Mask prototypes.

The Fascia Hub 

The Fascia Hub is an advanced solution for IoT interactions that is designed to manage auxiliary functions on a Raspberry Pi board [24]. It has several components, such as relays, microphones, lights, speakers, infrared cameras, and a diffuser atomizer, that allow it to perform a variety of tasks. One key feature of the hub is its privacy button, which gives users control over when their data is recorded, ensuring privacy and reducing ethical concerns. The speaker of the Fascia Hub provides controlled auditory stimulation for sleep studies, allowing researchers to control the volume and tone of the stimulation and study its effects on sleep. The Fascia Hub includes a diffuser atomizer for studying sleep and olfactory interventions. It can control humidity levels and diffuse different scents, giving researchers the ability to study their effects on sleep quality. A

The Fascia Portal

 The Fascia Portal enables the collection, labeling and analysis of sleep study data for use in diagnostic machine learning models. Researchers can monitor patient signals in real-time, store experiment data, and utilize a machine learning API for accurate identification of sleep stages, spindles and slow-waves. Traditional sleep staging methods are visually-based and prone to inconsistencies due to human variability, but the Fascia Portal offers fast, reproducible and accurate automatic sleep scoring. Unlike other systems, it also allows for realtime classification of sleep stages in 5-second intervals. Our API uses YASA [25] and SLEEPNET [26], trained and tested on a large set of polysomnographic recordings from healthy and sleep disordered subjects. Compared to expert consensus scoring, YASA performs as well as human inter-rater agreement (85%). The playback section automates sleep scoring to aid human scorers and speed up sleep staging. Until the model is more general and robust, a trained sleep scorer should visually verify the ML model’s predictions, paying special attention to low-confidence and/or N1 sleep epochs, which are most often misclassified. For sleep studies, the YASA and SLEEPNET models now include a confidence and salience map as part of API.

Researchers may gain detailed, accurate, and useful information into patients’ sleep patterns with the Fascia Ecosystem, improving sleep disorder diagnosis and treatment. 

How does the Fascia Sleep Mask compare to a PSG?

We validated data collected and processed by the Fascia Ecosystem against clinical-grade PSG databases.

Sample: 4 healthy participants slept 12 nights with the Fascia Sleep Mask in their homes.  Participants had a video call with a remote researcher before putting on the sleep mask and setting up the Ecosystem. 

Device validation against PSG: Results collected with the Fascia Sleep Mask were compared to healthy participants’ data from 2 clinical-grade PSG databases: the Physionet CAP Sleep Database(9) and the Sleep-EDF Expanded Database (10). •W stage measures: EMG epoch energy •N2 stage measures: absolute power based on median absolute power, frequency based on the frequency of spindles, symmetry based on the average location of the most prominent peak across spindles •N3 stage measures: mean slow waves visualization •REM stage measures: absolute magnitude of EOG peaks, fall slope from peak to trough of the EOG signal  


MIT Media Lab / Guillermo Bernal