So we run a keyword search on Twitter in real time. People tweet when they or someone in their house is sick. I take these tweets and look at how the numbers are changing. I know what patterns in the change mean because machine learning techniques have been used and have been shown to be very accurate.
I check how well the algorithm works by comparing the predictions with the numbers of real results coming out of labs testing sick and poo samples (I know that I would rather be at my desk looking at the numbers). The maths algorithm does very well. Next winter look out for school communication about sick bugs – that will have come from my research!
Comments