With Covid-19 cases again climbing, health tech companies and researchers are renewing their pitch for wearables and apps as a cutting-edge way to catch new cases and detect when patients are growing sicker.
The flood of tech tools — and the marketing machinery playing up their potential — promises to give users more timely information and fill key gaps in testing and tracing cases. But it is not altogether certain that these devices will benefit patients. It’s not just a basic question of whether a device or algorithm is accurate, health technology experts say, but whether the information provided is actually helpful in delivering better care or stemming the spread of the virus.
It is easy to take an off-the-shelf monitoring device, slap a Covid-19 label on it, and tell the world the device can be used to help lift us out of a public health crisis. It is far more difficult to ensure the product can home in on the unique signature of this virus and improve outcomes for patients, especially when it affects people so differently.
“I can tell you for a given system, it may be 80% accurate. But for me to show you it made someone’s care better is actually much harder to accomplish,” said Karandeep Singh, a physician and professor at the University of Michigan who studies the use of technology in health care.
But in certain clinical settings and populations, apps and wearables might be able to provide significant assistance during the pandemic, experts said. Here are a few questions to ask when trying to differentiate between empty promises and valuable tools.
Is it providing information specific to Covid-19?
Plenty of apps designed to monitor vital signs can accurately detect a fever and changes in respiration, but that’s not the same thing as correctly diagnosing Covid-19.
“That kind of app is not going to be nearly specific enough,” said Singh. “We’re heading into flu season. You can’t tell apart flu from cold from anything else.”
John A. Rogers, a biomedical engineer at Northwestern University, has spent months trying to tackle this problem with a wearable he developed for the university’s health system in Chicago. It is a Band-Aid-sized patch that attaches to the user’s throat to help monitor coughing and respiratory symptoms, such as shortness of breath.
One of the planned uses was to monitor signs of possible infection of frontline health care workers. So far, however, none of the health workers who have tested the device with Northwestern’s health system has become sick. It’s not clear whether none has contracted the virus, or whether some did but were asymptomatic, which points to a challenge facing any tech tool designed to track Covid-19 symptoms.
“You have to have some type of symptoms in order for us to pick anything up,” Rogers said. “If you’re completely asymptomatic we’re not going to be able to see it. This is not a molecular scale test.”
That’s not to say it can’t be helpful for other purposes. The wearable, which is experimental and has not been approved by regulators, is also being used to monitor symptoms in hospitalized patients. In one case, Rogers said, it flagged periods where a patient was experiencing a dangerous heart arrhythmia. It also picked up respiratory interruptions at night, helping providers spot signs of sleep apnea.
“It turned out to be pretty severe and we could see it pretty clearly,” Rogers said.
He said the impact of the wearable is still being evaluated and that his partners at the Shirley Ryan Ability Lab are seeking to develop an AI model that would use the data to help predict infections from symptom data.
Is the product targeted toward a particular population?
A major shortcoming of most wearables is that they are deployed in populations with very low risk of developing the problem they are designed to detect. The Apple Watch, for example, is often used by young, healthy people unlikely to benefit from its ability to detect the heart arrhythmia known as atrial fibrillation.
In Covid-19, that means many symptom tracking apps meant to flag the onset of illness in broad populations are likely to flag perceived problems that don’t amount to much. This results in a low positive predictive value, or the probability that a subject who tests positive truly has the illness.
“It’s going to be crying wolf a lot,” Singh said. He said that’s a significant drawback in a health care system trying to contend with a pandemic.
“With any of these apps, if you identify a problem, usually that problem results in a connection to the health care system, which has a time and a cost value to it. We don’t have unlimited resources,” he added.
However, the problem of false positives is mitigated in higher-risk populations, such as people who live in nursing homes or whose immune systems are compromised. In those defined user groups, it is helpful to provide caregivers with alerts about sudden changes in vital signs or a fever, because those are more likely to be associated with medical emergencies.
Will having the information support better care?
Apps and wearables can collect massive amounts of biological data from patients. But that doesn’t mean the information is going to be helpful to doctors who are trying to treat them.
A Covid-19 symptom tracker developed earlier this year by researchers at King’s College London, Harvard University and Stanford compiled symptoms reported by more than 2.6 million people, such as fever, cough, shortness of breath, and loss of taste and smell.
While the researchers are hopeful that the smartphone app can help inform individuals of their risks, and potentially flag infection hot spots, they are not arguing that it would significantly improve the care of infected patients.
That’s because it’s not clear that providing that information, through this app or another, will help doctors triage patients or change the way they are treating them.
“That’s something that remains to be seen,” said Andrew Chan, a Harvard professor who helped develop the app. “There’s a lot of hope this approach could be used in the setting of Covid because it is so highly infectious and there is a need to keep distance between patients and providers.”
But so far, there is no evidence that apps or wearables used to collect biological information on Covid-19 patients is improving their care. Singh said proving a positive effect on care is likely to take years, even in the case of products that have demonstrated an ability to accurately measure changes in symptoms and predict a patient’s deterioration.
“This is all experimental,” he said. “Studying the impact of a technology like this takes a ton more time than studying the validity of a technology.”