Abstract

Research since 2006 has shown that internet data, including search engine queries, social media, and ads, can be used to assess the incidence of different diseases. Thus, when COVID19 began, it was natural to try and apply the tools that were developed to track influenza-like illnesses from internet data to this novel pathogen. However, COVID19 presented several unforeseen challenges, ranging from the novelty of symptoms to the reaction of people and governments to its spread. Two years into the pandemic I will try to summarize some of the benefits that internet data and its analysis provided to health authorities and to people, the challenges faced by practitioners (especially at the beginning of the pandemic) and some of the lessons that might be applied in case of a future pandemic.

I will exemplify these lessons from three of our projects. The first provided the UK’s Health Security Agency with weekly reports on indications for COVID19 anomalies at regional levels in England. The second project used advertising in North America during the very early stages of the pandemic to help people decide if they should seek medical attention for COVID19 . Data collected during this effort has possibly uncovered an early wave of the pandemic in the USA . Finally, I will discuss a project where advertising, together with the Microsoft Health Bot, helped people in Israel decide if they should seek emergency care for COVID19 and provided data which allowed us to predict hospitalizations with a lead time of 9 days.