Can fitness trackers warn us about the next Covid-19 attack?
Xiaomi-backed wearables maker Huami published a study on detecting the spread of infectious disease using data from 1.3 million users
The tiniest changes in your heart rate, temperature and sleeping patterns could help give forewarning for the next big pandemic.
That’s according to researchers from the wearables company Huami after collecting data from 1.3 million users in Europe and China. The researchers hope their prediction model could eventually be used to alert health authorities of future infectious disease outbreaks.
Huami may not be as well known as Fitbit in much of the world, but it’s one of the largest wearables makers in the world. The Xiaomi-backed company is known for the cheap Mi Band line of fitness trackers and Amazifit smartwatches.
The company took a dive into nearly three years of data to see if it could create an algorithm to detect the rise of physiological anomalies, including higher body temperature, shorter sleep cycles and increased heart rate. The scientists then combined the data with official records on Covid-19 infections. The researchers found that more anomalies correlated with more reported infections.
In Wuhan, the epicenter of the outbreak, the rate of infections predicted by Huami’s data peaked around January 28, five days after the city-wide shutdown designed to fight the spread of the coronavirus. The officially reported peak was February 8. (Data after February 12 was omitted because of a change in diagnostic criteria that resulted in a jump in the number of cases.)
Wuhan also had the highest predicted peak among the cities studied. In addition to other cities in China, researchers also looked at cities in Italy, Spain, France and Germany.
Huami’s researchers see one big benefit in using wearables for forecasting the spread of infectious disease: Speed. Models that predict when infections are going to peak rely on reported statistics. A lack of tests, and therefore a lack of data, or waiting for test results can affect the timeliness of the models.
Huami isn’t the first to try to use wearables for insights into Covid-19. A group of researchers at Stanford started a study using Fitbits to see if they could train algorithms to show when a user’s body is fighting an infection.
Huami, however, was careful to avoid talking about the next steps. The company declined to comment on how the findings might be practically utilized by health authorities.
“We all know data sharing is a sensitive topic in the industry,” a Huami representative said, adding that there are benefits to the research in addition to privacy concerns.
The research team points out wearables data has limitations. One obvious problem is that not everyone wears fitness trackers or smartwatches. The rate of anomalies used to create predictions is based on statistics that may not include enough people or a diverse enough sample. The elderly population, for instance, is the most vulnerable to infectious diseases but the least likely to be trying new gadgets.
Another interesting problem is holiday libations. The Covid-19 outbreak overlapped with China’s week-long Spring Festival holiday, when heavy drinking is more common, possibly affecting physiological data.
But the researchers seem optimistic. According to the report, the results show that the model could help shape a nationwide solution for infectious disease surveillance systems. And with more data such as age, gender, and body weight, wearables could also be used in the future to make predictions for individuals.