I am a scientist/astronomer working at the Space Telescope Science Institute on a couple NASA astrophysics missions. I deal with high contrast imaging, the direct detection of exoplanets, coronagraphy, image processing. I am interested in big data but I am not an expert.

The idea I have would be to have large social medias like Facebook/IG, twitter ask each person everyday to answer a few basic questions about symptoms of the absence there of. This way you would have a day to day “picture” (and map) of the general health status of millions and millions. Applying machine learning algorithms on several days of such data you could perhaps help tp spot who might have covid-19 symtoms and should be tested or monitored more closely, where to plan and redirect resources to such or such county / zone affected or about to receive a lot of critical patients. A control sample of of people who have been tested would be used to train the algorithms.

So far, I have seen people and companies use social medias to get a feel of how people are doing and what they want to buy (and influence that). I would ask specific yes/no questions like they do when there is a major earthquake for instance. Facebook asks to report yourself. Here the daily questionnaire would be something like (with boxes to tick). It would not take more than 2 minutes to answer:

** Mandatory questions:

  • Fever Today? Yes/No
  • Dry cough Today? Yes/No
  • Fatigue Today? Yes/No
  • Sputum (mucus) production Today? Yes/No
  • Shortness of breath Today? Yes/No
  • Sore throat Today? Yes/No
  • Muscle pain Today? Yes/No
  • Headaches Today? Yes/No
  • Loss of smell Today? Yes/No
  • Globally to you feel better than Yesterday (or before)? Yes/No
  • Are you social distancing Today? Yes/No

** Optional questions (or once a week / for all):

  • How many day have you had significant fever: number
  • What was your highest fever temperature: ºC / ºF (multiple choice)
  • Have you been tested screened for COVID-19? Yes/No
  • Have you been tested for COVID-19? Yes/No
  • Have you been tested positive for COVID-19? Yes/No

** Data like age, sex, etc. would be provided or asked once for all.

It has to do with telemedicine too, evaluation and prevention of outbreaks. It may be too late for this one but not for the next one. Another potential outcome would be a better understanding based on millions of entries of the symptoms and their evolution for such or such disease. People could we warned if algorithms converge towards other trends (other pathologies).  The basic data would be public (interfaced with github) and some parts would be anonymous (precise geolocation, names) or locked to super users with ethic standards (TBD). Let me know if you are interested. Currently I don't know how to develop this, I am willing to meet with Social Media companies reps. and machine learning specialists to discuss it.

What it does

How I built it

Challenges I ran into

Accomplishments that I'm proud of

What I learned

What's next for Assess pandemic & optimize response with social medias




Devpost Software Identifier