.jpg)
Everyone, it seems, is sick right now. Walk into an office or school and chances are you’ll find plenty of empty seats, as everyone is laid up with a fever or heavy cold. Rates of flu-like illness are high across the northern hemisphere and don’t appear to have peaked. With one in four flu tests returning positive in the United States and about one in seven in the United Kingdom, a lot of people are out of commission. In the US, it is estimated that at least 6 million people had visited a doctor and 120,000 had been hospitalized by December 3.
If we’d had a more detailed view of how cases were building this year, things might not have gotten so bad. Accurately detecting an outbreak’s early stages can show people—in real time, based on a postal code—where cases are rising and help them avoid exposure. Early signals of flu levels in each community also allow for better predictions of how the entire season will evolve, and what its consequences will be for the health system or businesses.
This can be particularly useful for local doctors and nurses who need to prepare for an influx of patients, or for medical suppliers who need to coordinate regional stocks of drugs and personal protective equipment. With children hit particularly hard by respiratory viruses this season, it is becoming increasingly difficult to find over-the-counter painkillers and fever remedies in northern states, for example, as a national shortage in Canada has prompted desperate parents to seek supplies across the border.
Entrepreneur Inder Singh believes he has a tool that can help. For several years, his California-based technology company, Kinsa, has claimed to be two to three weeks ahead of public health officials when it comes to detecting clusters of illnesses down to the county level. Kinsa achieves this by collecting and analyzing temperature readings from more than a million smart thermometers across the US, data members of the public voluntarily share via their paired smartphones, along with symptom reports. Typically, a thermometer is the only medical device used in a household, Singh says, “so when you use it, it’s like a signal that someone is sick.”
Authorities are often slow to pick up on how a disease is spreading: Usually it takes days before someone’s fever, cough, or sore throat leads them to get tested, putting their results on the record. There’s then a further delay before the reports are sent to the Centers for Disease Control and Prevention. (With the flu, this takes a week.) That means estimates of an outbreak’s size lag behind reality.
But with a smart thermometer, this essentially happens in real time. The devices operate via a Bluetooth-connected app, giving the company geographically accurate data about the onset, duration, and severity of flu-like symptoms that it shares anonymously with some health authorities and businesses, including pharmacies and manufacturers of cold and flu drugs. (For example, its early warning system shows that the risk of flu-like illness in New York’s Hudson Valley and Albany regions are slightly lower than in the rest of the state.) If a child has a fever and their parents use the thermometer too, it is relatively easy to determine how quickly a virus is spreading within a household.
Singh says a smart thermometer reading is particularly useful for people who have only mild symptoms and can be spared a visit to the doctor, as well as those who can’t access or afford medical care. That means the thermometers can catch people that health departments miss, and provide a more accurate picture of how a flu wave is building.
University researchers collaborating with Kinsa employees have compared the company’s data with that of health departments to see how well they match, and to determine to what extent users’ temperature readings and symptom reports can help model and predict outbreaks. On the national and state levels, researchers found that the Kinsa data correlated closely with official surveillance measures but detected flu-like cases up to three weeks before they were reported by health officials—likely because of the lag between someone’s fever starting and their illness being officially reported.
Of course, smart thermometers aren’t the only form of digital surveillance. Analyzing Google search data can also provide a head start on predicting an outbreak—people often google their symptoms before going to the doctor. But their searches are usually based only on perceived symptoms, which can be subjective. “The advantage of Kinsa is that it relies on objective temperature readings, as opposed to solely self-reported symptoms,” says Sarah Ackley, a postdoctoral fellow in epidemiology and biostatistics at the University of California, San Francisco, who has examined the accuracy of Kinsa’s regional predictions.
Kinsa thermometers are sold at major pharmacies, and the company also distributes them for free to families and staff at US public schools that apply for them. So far, the New York City Department of Health and Mental Hygiene is the only government agency that has partnered with the company on a citywide basis, distributing more than 100,000 thermometers free of charge to 500 public schools since September 2021.
New York City is reporting very high levels of flu-like illnesses at the moment. In the week ending December 3, 13 percent of patient visits to health care facilities were due to respiratory illness involving a fever combined with cough or sore throat. This is above the national baseline of 2.5 percent for this winter season. According to Kinsa’s forecast, the current wave of cases will peak slightly later in New York City than nationally—meaning the risk of infection is predicted to remain high into the new year. “We believe it’s going to be the most severe season, and we’ve been projecting that for a while,” says Singh.
However, the extent to which New York City’s health department is using Kinsa’s real-time data this flu season is unclear: “The work with Kinsa is a pilot, and we’re still exploring how best to use the data,” a department spokesperson wrote to WIRED by email.
Health agencies tend to be relatively conservative when it comes to new data systems, says Jay Varma, a professor of population health sciences at Cornell University who was involved in the New York pilot project. “When new systems arise, it can take time to understand how best to use them for decision-making,” he says. Moreover, new systems require additional staff and resources, he says, and there is always the question of how to maintain those resources in the future.
Drug manufacturers, pharmacies, and insurers are typically a little more willing to experiment with new tools to stay ahead of the curve—so it’s here that thermometer data is making inroads. For insurers, the data lets them know where to deploy additional case managers. And a number of companies are using Kinsa’s seasonal forecast to coordinate their inventories or target their advertising to areas where more people will be seeking drugs or disinfecting wipes. The maker of cold and flu drugs Tylenol and Motrin has been working with Kinsa since 2021 to predict seasonal demand and has so far averted a widespread shortage in the US, according to a Johnson & Johnson Consumer Health spokesperson.
There are, however, limits to what Kinsa’s information can do. It’s particularly difficult to translate symptom-based data into predictions of hospital admissions or a disease’s contagiousness when multiple respiratory illnesses are at play, especially when there are vaccines for some of those diseases but not others, says epidemiologist Ackley. Also, without a lab-confirmed test, it is often unclear whether someone has the flu, Covid-19, or another respiratory virus. “The symptoms overlap considerably, with the exception of the loss of taste and smell that we see with Covid-19,” says Ackley. In her opinion, smart thermometers can offer useful real-time data, “but I would tend to view it as a complement to, and not a replacement for, public health surveillance,” she says.
Singh, who worked for global health organizations for years before founding Kinsa in 2012, is candid about the limitations of his company’s data. If a new Covid variant emerges this winter and Americans change their behavior, for example by returning to masking or isolating, Kinsa’s seasonal forecast could prove wrong, he says, “but we’ll be able to adjust it once the real-time data on incidence comes in.”
In the meantime, he is excited to be working with pharmacy chains, insurers, and other companies this flu season and aspires to make smart thermometers a commonly used tool for disease surveillance. “My hope is that there is a way to build out a national system,” he says.
