Machine-Learning Device for In-Home Detection and Diagnosis of COVID-19 Patients’ Health

    By Kathleen Berger, Executive Producer for Science & Technology
    Fourth-year medical students are making home deliveries to COVID-19 patients participating in an observational clinical trial involving sensing capabilities and clinical analysis of a patient’s movements, no matter how big or small. Based on movements, machine learning can lead to a diagnosis and intervention.

    The technology is called Emerald, a touchless sensor and machine-learning platform for health analytics. The Emerald system superficially resembles a Wi-Fi router, mounted to the patient’s wall.

    Emerald is based on award-winning wireless and machine learning research from Massachusetts Institute of Technology (MIT). Using wireless signals, the software allows for remote monitoring of COVID-19 patients’ movements, sleep patterns and breathing using wireless signals. It’s all done while participants are quarantined at home.

    “It’s going to detect their chest wall motion. It helps detect their breathing and changes in their breathing,” said Damien Abreu, MD-PhD candidate at Washington University School of Medicine in St. Louis.

    Machine learning is considered crucial. The more Emerald learns the COVID-19 illness from even the smallest of movements related to symptoms and disease progression, the better Emerald becomes at diagnostics.

    “That might serve as an indicator for a patient to get to the hospital soon because perhaps they are not headed in the right direction,” said Abreu.

    Abreu is project manager over the launch of “Emerald for COVID-19 in St. Louis”. He worked closely with team lead Lena Dang, MD-PhD candidate. They are among several fourth-year medical students from Washington University School of Medicine in St. Louis who helped design and execute the observational trial for in-home patient monitoring.

    “Using a wireless device to monitor physiologic signal of the patient, and then use AI to learn their illness trajectory,” said Dang.

    “It’s shooting out, constantly, these little radio waves that are then bouncing off all the surroundings, bouncing off the chest wall and arriving to the device in different ways,” said Abreu. “The algorithms they are using to interpret the data and analyze the data are configured into an interpretation of changes in motion.”

    Because Emerald has applications for various diseases, the technology was already at Washington University. It was used in a different trial at the Center for the Study of Itch (CSI). CSI’s co-director, Brian Kim, MD, assistant professor of dermatology, anesthesiology, pathology and immunology is principal investigator of the itch-related clinical study. But when the pandemic hit, the lab pivoted.

    “So we thought instead of studying itch, we can use this in patients who are experiencing Covid-19,” said Kim. “Can we start to now detect subtle cues about their breathing, their movement that would predict an adverse outcome? And we can understand more about the disease?”

    Kim is the principal investigator for the Emerald application for COVID-19. Kim explained why putting a clinical trial in the hands of medical students is a unique approach.

    “I already have a lot of things to do and I have to keep my lab going. And I thought one way I can potentially do this is to get the students involved,” said Kim. “In doing so they would know ‘What is it like to design a trial?’ What is it like to deal with devices and collaborate across the country?”

    The data from St. Louis area COVID-19 patients is monitored in real time.

    “The data is getting wirelessly transferred to our collaborators at MIT who have the technical infrastructure to process the data and analyze it,” Abreu said. “And all of that is happening in real time.”

    Medical students in St. Louis also analyze the data and call patients daily.

    “What’s very unique about this study is the fact that we are collecting both clinical information and also motion information at the same time,” said Abreu.

    Anyone interested in participating in the study can reach out to the research team online: