Stanford researchers working on wearables for early detection of infectious diseases

Stanford researchers are working with Fitbit and Scripps Research Institute to develop wearables that can detect infectious diseases such as COVID-19 and help contain their spread. Early evidence has suggested that wearable devices, such as smart watches, rings and clothes, can help predict the onset of infectious disease even before symptoms begin.

The consortium brings together ongoing research at Stanford and Scripps. Led by Michael Snyder, professor and chair of the genetics department at Stanford School of Medicine, the Stanford Healthcare Innovation Lab launched the COVID-19 wearables study with the goal of determining whether information from wearable devices can be used to track infectious diseases.

You Might Like

“Smartwatches and other wearables make many, many measurements per day — at least 250,000, which is what makes them such powerful monitoring devices,” Snyder told Stanford Medicine News. “My lab wants to harness that data and see if we can identify who’s becoming ill as early as possible.”

Wearable devices run by algorithms can alert a user when their heart rate, skin temperature or other physiological indicators signal that their body is fighting an infection. One study has suggested that Fitbits can accurately detect flu hotspots faster than the Centers for Disease Control and Prevention (CDC).

However, more work needs to be done before the wearables can be used to detect COVID-19, the researchers say.

“[The current algorithm] is more like a flagging algorithm,” said Xiao Li, a former postdoc in Snyder’s lab who worked on the 2017 study upon which the COVID-19 wearables study is based. “So I think it’s too early to say this is a diagnosis, it’s more like a warning. An algorithm should be developed for a pattern specific to COVID-19.”

“We do not know if we can tell the difference between COVID-19 and other viral infections, but we think that it is still useful to know if you are getting sick in a pandemic,” Snyder told The Daily.

Additionally, he said people will need to consider context when using wearables for monitoring infection.

“If they are watching a scary movie and their heart rate goes up, they should ignore it,” he said.

In March, Scripps Research Institute launched DETECT, a study that monitors heart rate and records symptoms like fever and coughing. For the COVID-19 wearables study, Snyder is collecting data from five different brands of wearable devices and developing algorithms to detect when someone is getting sick. 

“The goal is early detection before they [the users] are symptomatic,” Snyder wrote in an email to The Daily. “We are enrolling 1) People who are confirmed positive for COVID-19 who have already been wearing a smart-watch and 2) High risk people (healthcare workers, family members of sick people) who are getting regularly screened for COVID-19.”

Snyder and his group are giving people in the high-risk group smartwatches while they are healthy and are continuing to track them while they are sick.

“As soon as we announced the study, 2000 people signed up,” Snyder wrote. “Moreover, millions of people are wearing smartwatches.”

Russ Altman, professor of bioengineering, genetics, medicine and biomedical data science, said it is important that patients wear wearables early on for the technology to be useful.

“You need to make sure you get them very early, because if you get them when they are clinically sick, that’s not useful, because a doctor can figure out they’re sick by looking at them,” Altman said.

He also said it is important to have a large number of participants and a robust training set — the data that the algorithm will be trained on — to accurately predict who will be sick.

“They need to be very careful about the sample size and about making sure that they have an unbiased sample and that they validate in an entirely new population so that they don’t overfit to one particular population,” Altman said.

Snyder said it is possible to use a small number of people to train the algorithms to detect subtle physiological changes.

“The algorithms are trained on the individual so we are looking for a shift on that person that is abnormal,” he wrote. “However, we do not know how often we will miss asymptomatic people who might have subtle changes.”

His work is based on a report he published in 2017, which showed a correlation between deviation in physiological signals and inflammatory response. The algorithm in that study showed that it was possible to detect infection using physiological data measured by wearables.

Li, the first author of the 2017 report, said she was inspired to develop the idea for the algorithm when an individual whose physiological data the lab was tracking was found to have extremely low levels of SpO2, a measure of blood oxygen saturation.

“It turned out to be early onset Lyme disease,” Li said.

Upon further analysis, researchers discovered the person’s heart rate and skin temperature were also abnormal, and that the signals had started very early. That study inspired Li to develop the algorithm that was used to detect disease using physiological data.

She expressed optimism about the use of wearables in the COVID-19 pandemic, particularly for early detection of disease, while highlighting that wearables’ results should be taken into consideration with other data.

“I think this is a cool alternative and should be integrated with other tests and models,” Li said. “With early detection using wearables, we can guide people to hospitals for blood tests.”

Altman is hopeful about the use of wearables to combat viral infection, if not for COVID-19 then at least in the future.

“If they can do all that, it will be very useful because they might be able to learn things that would be relevant for the next pandemic,” he said.