Throat wearable tracks Covid-19 symptoms
May 19, 2020
Researchers at Northwestern University and Shirley Ryan AbilityLab in Chicago have developed a throat wearable device and are creating algorithms tailored to catch early signs and symptoms associated with Covid-19 and to monitor patients as the illness progresses.
The more researchers learn about the novel coronavirus, the more unknowns seem to arise. These ever-emerging mysteries highlight the need for more data to help researchers and physicians better understand and treat the contagious and deadly disease.
Capable of being worn round the clock, the device produces continuous streams of data and uses artificial intelligence to uncover subtle, but potentially life-saving, insights. Filling a vital data gap, it continuously measures and interprets coughing and respiratory activity in ways that are impossible with traditional monitoring systems.
Developed in an engineering laboratory at Northwestern and using custom algorithms being created by Shirley Ryan AbilityLab scientists, the devices are being used in a study at AbilityLab by Covid-19 patients and the healthcare workers who treat them. About 25 affected individuals began using the devices last month. They are being monitored both in the clinic and at home, totalling more than 1500 cumulative hours and generating more than one terabyte of data.
About the size of a postage stamp, the soft, flexible, wireless, thin device sits just below the suprasternal notch – the visible dip at the base of the throat. From this location, the device monitors coughing intensity and patterns, chest wall movements that indicate laboured or irregular breathing, respiratory sounds, heart rate and body temperature, including fever. From there, it wirelessly transmits data to a HIPAA-protected cloud, where automated algorithms produce graphical summaries tailored to facilitate rapid, remote monitoring.
“The most recent studies published in the Journal of the American Medical Association suggest that the earliest signs of a Covid-19 infection are fever, coughing and difficulty in breathing,” said Northwestern’s John Rogers, who led the technology development. “Our device sits at the perfect location on the body – the suprasternal notch – to measure respiratory rate, sounds and activity because that’s where airflow occurs near the surface of the skin.”
Arun Jayaraman, a research scientist at Shirley Ryan AbilityLab, who is leading the algorithm development, added: “We anticipate that the advanced algorithms we are developing will extract Covid-like signs and symptoms from the raw data insights and symptoms even before individuals may perceive them. These sensors have the potential to unlock information that will protect frontline medical workers and patients alike, informing interventions in a timely manner to reduce the risk of transmission and increase the likelihood of better outcomes.”
Many patients’ symptoms fully disappear before they suddenly and unexpectedly begin deteriorating, sometimes within a matter of hours. Other patients have recovered and tested negative but later test positive again.
The unknowns underscore the need for continuous patient monitoring to ensure that physicians can intervene at the slightest sign of trouble. This device provides around-the-clock monitoring for Covid-19 patients and those exposed to them.
“Having the ability to monitor ourselves and our patients and being alerted to changing conditions in real time will give clinicians a new and important tool in the fight against Covid-19,” said Mark Huang, a physician at Shirley Ryan AbilityLab, who has worn the sensor. “The sensor also will offer clinicians and patients peace of mind as it monitors Covid-like symptoms, potentially prompting earlier intervention and treatment.”
The device can monitor hospitalised patients and then be taken home to continue round-the-clock supervision. The real-time data streaming from patients give insights into their health and outcomes that is not being captured or analysed by traditional monitoring systems.
Although the wearable device is unable to measure blood oxygenation levels, which is an important component of lung health, the team plans to incorporate this capability in its next round of devices. The lab has already successfully incorporated this capability in its previous work to produce clinical grade-monitoring devices for intensive care units.
Not only can the device monitor the progress of Covid-19 patients, it could also provide early warning signals to the frontline workers who are most at risk for catching this disease. The device offers the potential to identify symptoms and to pick up trends before the workers notice them, thereby providing an opportunity to engage in appropriate precautionary measures and to seek further testing as quickly as possible.
“People with obvious, severe symptoms are going to the hospital, being tested or being told to self-isolate,” Jayaraman said. “For those who have symptoms they perceive as mild or seasonal allergies, there is no warning system. They could be in contact with others and unknowingly spread infection.”
In the future, this sensor package could help researchers and physicians quantify which therapeutics are working best.
“At the simplest level, our systems allow assessments based on data, in a quantitative way, without relying on human judgment of whether a patient is coughing more or less,” Rogers said.
The device builds on recent research from a collaboration between Rogers’ and Jayaraman’s labs, first published in February 2020, with a focus on monitoring swallowing and speech disorders in patients recovering from stroke. These sensors work by precisely measuring vibratory signatures from the throat and chest. By measuring vibrations rather than acoustics, the team avoids noise from background sounds and it bypasses privacy issues.
Jayaraman’s team is developing custom signal processing and machine-learning algorithms to train the device how to recognise coughs in the data.
“As the algorithm becomes smarter, our hope is that it will begin to discriminate among which coughs are Covid-like and which are from something more benign,” Jayaraman said. “The most basic approach, already deployed on Covid-19 patients and health care workers, simply counts coughs and their intensity.”
More advanced analytics packages should be available within the next few weeks.
In the coming weeks, the teams will continue collecting patient data to strengthen their algorithms through deployments both in the clinic and at home. They also are responding to other requests for access to the technology, across the medical complex in Chicago. Additional deployments have started.
Rogers and Jayaraman also are examining data from patients recovering from Covid-19, attempting to determine when they are no longer contagious. Some patients wearing the device have been dismissed from the acute-care hospital and are rehabilitating at Shirley Ryan AbilityLab. In the future, this device could help determine whether post-Covid patients still have minor, perhaps imperceptible symptoms.
Rogers hopes the device will not just tell physicians how to best treat Covid-19 but also inform researchers about the nature of the virus itself.
“The growing amount of information and understanding around Covid-19 as a disease will be critically important to containing and treating the current outbreak as well as those that might occur in the future,” he said. “We hope, and we believe, that these devices may help in these efforts by identifying and quantifying characteristics and essential features of cough and respiratory activity associated with this disease.”
To accelerate the deployment of this device, the team recently launched a lean engineering-centric company, Sonica Health, based on intellectual property jointly developed by Northwestern and the Shirley Ryan AbilityLab.