How BlueDot is using LLMs to detect outbreaks faster

Next generation surveillance of infectious disease intelligence

Lauren Charbon, Director of Product Marketing

When protecting your people, operations, and economy from the disruption of infectious diseases, speed and accuracy is paramount. From capturing the earliest signal of an outbreak to swiftly and comprehensively assessing the risk at the location of outbreak, or around the world, trust is placed in the intelligence used to inform high-stakes, high-consequence decisions. That is why BlueDot is a purpose-built organization focused on curating and creating the world’s most accurate and up-to-date infectious disease intelligence platform.

In the midst of a new era of epidemics our mission is more important than ever – outbreaks are happening more often, and at a higher scale than ever before. It requires more actors across public and private organizations to be engaged and equipped to fight. However, that is easier said than done. Monitoring outbreaks and the factors surrounding their disruption, spread, and burden is a complex topic. The nuances of answering questions such as “will an outbreak of Marburg in Guinea cause disruption to people in Toronto, Canada?” requires multiple data assets, detailed analysis, and expert judgement. As a result, the actors who can perform this crucial analysis – and the speed and scope at which they can perform it – is limited.

BlueDot has always leveraged elements of AI to move the yardstick of innovation forward to solve these problems, but large language models (LLMs) are enabling us to move faster and further than ever before. By purposefully integrating LLMs across our workflows – from disease detection, to risk assessments, and mitigative action – we get closer to the first signal of an outbreak and ensure that anyone, regardless of epidemiology, scientific, or data science background, has the intelligence needed to make time sensitive, tough decisions that save lives. Diseases may move fast, but BlueDot is moving faster.

Finding disease outbreaks faster with LLMs and next-generation surveillance

It is impossible to manually monitor every local publication required to catch disease outbreaks at the first possible signal. Missing important signals, or catching them reactively, is an accepted reality in most surveillance teams. We refuse to accept that reality. Our disease surveillance engine already uses natural language processing to scan 300,000+ articles across 130+ languages in near real-time, every day, reviewed and vetted by industry experts. With LLMs the results of our engine will be available faster than ever before – in real time – for proactive alerting, investigative searches, or instant incorporation into critical workflows that identify emerging outbreaks. And after that, the possibility of moving from intelligence extraction to intelligence production becomes a reality. Our epidemiologists will continue to be an essential part of our quality assurance process to ensure trust in our global surveillance.

Democratizing intel that informs time sensitive, tough decisions that saves lives

The diverse range of factors which indicate the disruption of a disease event lives across thousands of different databases. Even in a consolidated system, accessing them requires tedious one-by-one extraction and manual data consolidation – and the education to know what analysis to perform. With even “rapid risk assessments” taking days to complete, stripping away tedious data collection and analysis to focus instead on action is critical to meaningful progression in the fight against infectious diseases. LLMs will enable BlueDot to operate at the speed and simplicity of a text chat, while maintaining the trust and credibility of a fully trained epidemiology team. We believe we can turn days of analysis into hours, then minutes, and eventually seconds. Getting not only faster, but also lowering the bar of technical knowledge needed to access it. Everyone is empowered to act.

Disease outbreaks will continue to happen, but with BlueDot everyone can get ahead of the curve to reduce disruption. Together, with trusted intelligence, we can build a safer and more resilient world.

Recommended Reading