BlueDot — Spotting disease outbreaks with human expertise and AI
BlueDot’s mission is to protect people around the world from infectious diseases with human and artificial intelligence. Their system became aware of the pneumonia cases (now called COVID19) in Wuhan on December 31–2019. They publicly released their findings in the Journal of Travel Medicine on January 14–2020. An article in the WIRED magazine on January 25–2020 featured BlueDot and their now famous findings. Let’s take a closer look into how BlueDot’s solution works.
Why do we need to keep an eye on disease outbreaks?
This question perhaps does not need an answer now, given that the entire world is living with the COVID19 pandemic and knows what that means. Speed is key in spotting outbreaks as every day lost has severe consequences for human life.
History reveals that infectious disease outbreaks like those of SARS-1 or MERS or now COVID19 are not reported fast enough from the epicenters. Though there are worldwide protocols for disease communication and world bodies like the WHO governing these, they are not enforced enough. Governments of countries where outbreaks have occurred take time to assess the impact and weigh the severe economic consequences of reporting the same. This why we need alternate systems to keep track.
How has such a system been built by BlueDot?
BlueDot’s system comprises of 4 key components: Surveillance, Dispersion, Impact and Communication.
Human experts — Epidemiologists and Clinicians frame what is relevant in the context of the most important infectious disease outbreaks. An AI system then trains to pick these out from a sea of online articles spread across multiple languages. When deployed, techniques such as NLP are used to crunch the incoming data from articles on the web into a small number of alerts that may then be analysed by human experts who deem it to be a real red flag or a false alarm.
Once the outbreak is confirmed as a genuine red flag, the next step lies in finding out where this outbreak will spread to. In this age of globalization, people travel all around the world and the pathogen is carried on these very human carriers throughout the earth. To understand this dispersion, the flow of people outward from the outbreak epicenter by air, sea and land is tracked. While air travel is assessed via monitoring flight ticket sales, anonymized location data from ~400 million mobile phones yields information on where people move via land and sea. This helps ascertain which other towns or cities that get most of these travelers are going to become the next hot-spots for the disease. An understanding of the local weather in those towns and cities as well as an idea of the public health infrastructure helps assess overall impact.
The Governments of those towns and cities as well as the relevant public health officials and airport authorities as well as national defense authorities now need to be notified as soon as possible so that appropriate preparations may be made to contain or mitigate the outbreak impact. Post identifying the COVID19 outbreak in Wuhan, BlueDot’s system published it’s findings that stated that the next set of cities to be affected would be Tokyo, Bangkok, Hong Kong, Seoul, and Taipei.
- Human Expertise + Artificial Intelligence is a winning combo
Humans frame the problem, AI crunches the data and looks for the key terms of interest and then human experts identify the real red flags from the false alarms. Humans use their power of intuition and machines use their power of computation — that results in an incredibly useful solution for the public.
2. A possible pandemic alert service
Though this system notifies specific organizations and government bodies, the question to ask is whether this service may be extended to the public at large to act like a Pandemic alert service. This would allow the public to prepare fast and also put pressure on their local, state and nationwide public health and Government officials to take prompt action.