My Authors
Read all threads
Millions of smart thermometers are picking up higher than expected rates of fever readings in #Florida. How many more canaries in the coal mine do we need?
healthweather.us

#CoronavirusOutbreak
A “smart” thermometer is showing some unexpectedly high rates of fever measurements in Florida the week leading into 3/20 during the #CoronavirusOutbreak (healthweather.us). Is this a signal or just noise? Does it matter? A thread from an ER doc/epidemiology professor:
2/ First, what is a smart thermometer? It’s a thermometer that connects to a smartphone app. Temperature readings upload into the app. Users can *voluntarily* allow those readings, their geolocation, and self-reported symptoms to automatically upload to the company’s cloud.
3/ How are thermometer readings used to generate the map? The company has developed a validated forecasting model to predict flu like illness based on thermometer readings in a given area. The model considers fevers as >100.0 F per the CDC definition of "influenza like illness".
4/ How accurate is the model? It’s very accurate. Many health apps do no validate their devices. This is one did. The readings are *highly* correlated with observed flu like illness ER visits (r>0.95) at the regional levels and by age
academic.oup.com/cid/article/67…
5/ So what's shown on the map? They are taking the model and predicting flu like illness rates for a given region based on preceding trends. Higher than expected rates of fever readings than predicted by the model -> flags that region as having elevated “atypical illness level”
6/ For example, see the anomaly detected in Brooklyn. A spike in febrile illnesses were measured when the model expected a downtrend with tail end of flu season. This correlates with the spike in #COVID19 cases seen in Brooklyn at the same time.
content.kinsahealth.com/covid-detectio…
8/ So is what's going on in Florida a signal for #COVID19 or is it misleading noise? No doubt #COVID19 anxiety is causing people to buy more smart thermometers and check their temps more often. But would this disproportionately be the case in Florida? Not a great explanation
9/ Why do some areas with lots of known #COVID19 look grey? It just means that the number of cases of fever are not higher than what they predicted. See LA below, they predicted moderate illness levels and they observed moderate illness, this will likely change with time
10/ The more plausible explanation for Florida for having higher than expected febrile illness is that this is an early warning sign for #COVID19. Feb/March is peak season for tourists visiting from all over the country and mixing in large crowds
foxnews.com/health/califor…
11/ As an ER doc treating a potentially sick patient I don’t just look at one abnormal lab test, I consider the all patient’s vital signs, story, exam, and lab/imaging data. In the same way, the thermometer data should be used in conjunction with other data to drive policy
12) Other #CoronavirusOutbreak indicators we could look at to determine when to deploy critical policy interventions include: 1) ED visits for flu like illness; 2) confirmed cases of COVID19; and 3) deaths for #COVID19. All three indicators are also very concerning in Florida:
13/ *Observed* ED visits for flu like illness have spiked unexpectedly (correlating closely with the spike in fevers as measured by the smart thermometers). The downside with this indicator is that it could be biased by anxious people seeking treatment floridahealth.gov/diseases-and-c…
14/ *Observed* cases of #COVID19. These have risen exponentially in Florida. The downside with this indicator is that it’s completely dependent on the availability of testing in addition to treatment seeking behavior.
thecapitolist.com/florida-corona…
15/ Observed #COVID19 deaths are also rising exponentially. The downside with using deaths as indicator to drive policy is there could be a several week lag between diagnosis and the date of death. Furthermore, death rates are subject to the quality/ capacity of hospitals.
16) As an ER doctor, we seek symptoms/tests that are predictive of impending doom to activate time-sensitive life saving treatment (e.g. antibiotics for a bloodstream infection). Ignoring a single warning sing can be catastrophic and lead to the loss of a life and a hefty lawsuit
17/ Ignoring multiple warning signs of public health pandemic and not delivering time-sensitive public health interventions can lead to the loss of life for significant populations of people not just in Florida but also to places visitors return to:
tmz.com/2020/03/19/34-…
18/ So @GovRonDeSantis, don't ignore the warning signs in the data being put out by @kinsa. Step up the policies to halt the #CoronavirusOutbreak. If not for the people of Florida and the US, do it for the health care workers. Here's my team - I love them
19/ Kudos to the lead scientist behind the modeling of smart thermometer data at @OregonState Benjamin Dalziel ib.oregonstate.edu/faculty/benjam…. For news coverage see:

miamiherald.com/news/coronavir…

nytimes.com/2020/03/18/hea…

MSNBC-
Missing some Tweet in this thread? You can try to force a refresh.

Enjoying this thread?

Keep Current with Kit Delgado

Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

Twitter may remove this content at anytime, convert it as a PDF, save and print for later use!

Try unrolling a thread yourself!

how to unroll video

1) Follow Thread Reader App on Twitter so you can easily mention us!

2) Go to a Twitter thread (series of Tweets by the same owner) and mention us with a keyword "unroll" @threadreaderapp unroll

You can practice here first or read more on our help page!

Follow Us on Twitter!

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just three indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3.00/month or $30.00/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!