Non-Contact Sleep, Vitals, and Behavior Sensing

Emerald is a novel technology that allows for monitoring people's health and physiological signals in a passive way without asking them to wear sensors on their body or measure themselves. The technology is based on a wireless device that looks like WiFi router. It transmits very low power radio signals, which bounce off people and objects in the environment. The device analyzes these reflected radio signals using machine learning algorithms and infers people's sleep, respiration, and movements. The Emerald health metrics are validated in comparison to Federal Aviation Administration (FAA) approved methods for measuring respiration, sleep, and motion. Our team has developed the technology at Massachusetts Institute of Technology (MIT) and created a spin out to commercialize it for improving health and wellness.

The Emerald technology is well-suited for a physical health monitoring system measuring indicators of health and human performance of astronauts on long duration deep space missions. Specifically, NASA has Sleep loss and circadian desynchronization as key risks to space mission that affect astronauts' performance and lead to adverse health outcomes. It has also highlighted the need to identify a set of validated and minimally obtrusive tools to monitor and measure sleep-wake activity and associated performance changes for spaceflight.

The Emerald technology promises to address this risk and deliver the unmet need. Specifically, it delivers a simple contactless sensor that monitors sleep-awake activities, and reports sleep stages, sleep efficiency, and changes in the circadian cycle. The sensor is unobtrusive and does not require astronauts to wear sensors or actively report their sleep. They can go about their mission while the Emerald sensor passively monitor their sleep.

This project focuses on technical development that allows the use of Emerald technology for monitoring sleep during space missions. While the key algorithms have been proven to accurately infer sleep stages and other sleep parameters, prior to this project, the technology was not ready for real-world deployment. Specifically, past research required a significant amount of manual processing where our engineers and data scientists manually collect the data, reformat the collected data, run the algorithms, schedule different processes, and manage computation and storage processes. Thus, in this project we developed the key infrastructure to automate all of these processes and transform the Emerald technology from a system capable of only short term, non-interactive studies to a platform that can be used for continuous, long-term, longitudinal sleep monitoring. To this end, we have developed the following three components: 1) an automated pipeline for data collection and processing, 2) a storage system and access mechanisms that operate on per-user sleep analytics, 3) and a secure user interface to allow users to access their data.

The project was also extended to address new needs that arose from the COVID pandemic. Emerald's ability to monitor physiological signals, e.g., respiration, became particularly relevant for monitoring COVID-19 patients from a distance without any body contact, and hence without incurring contagion risks. Thus, the project was extended to develop tools to track respiration and provide an alert system that allows a medical doctor to configure the system to remotely monitor COVID patients, and to send a notification when certain conditions of abnormal breathing are detected. The extension also covers an observational study in which the Emerald sensor is used to monitor actual COVID patients to check the viability of using Emerald for passive monitoring of COVID patients and tracking their recovery process and identifying recovery problems.

More »