CRI Research Fellows Marc Santolini and Bastian Greshake Tzovaras are working on creating a model of how tweets about self-reported COVID19 symptoms can help predict upcoming pandemic waves, and more generally the rise and fall of the disease. With the help of Naïla El Haouari (intern at the CRI) and Samuel Fraiberger (NYU & MIT MediaLab), they crawled public tweets from the Paris region and filtered by symptoms keywords. While this seems to give already a good correlation between the amount of tweets and the rise & fall of the disease (see the image above), the automated filtering is very crude. People generally don't only tweet about symptoms when they are currently falling sick, but also about past times when they fell sick, or when talking about the general news.
To filter out such false-positives Marc & Bastian need your help to see which tweets are describing an acute symptom and which ones don't? Your contribution will make a direct impact! With this additional filtering it will be possible to not only refine their current analyses, but also to train artificial intelligence to create better automated filtering strategies.
Your help doesn't need to end with annotating tweets. As CRI is dedicated to open science, all the data and analyses methods are publicly shared. So you can download the tweets and annotations yourself to reproduce the findings and improve the analyses.