New emerging outbreaks

The early detection of pandemics has always been a significant challenge for public health authorities around the world. With novel infectious diseases, such as the recent COVID-19 pandemic, the threat to global health is even greater. Therefore, we are exploring new pandemic detection systems.

We predict that we can use an autoencoder to provide an early warning for pandemics in a disease-agnostic manner. When trained on healthy full blood counts, an outbreak will present through a rise of anomalous full blood counts in a specific location, indicated by an increase in reconstruction error. We have shown that this is effective for detecting COVID-19 in Cambridge. Error signals aligned with COVID-19 waves when trained on rich full blood counts from before the pandemic.

Since the COVID-19 pandemic, we have had a smaller-scale Monkeypox outbreak and recently a new Ebola outbreak. In addition, an increase in Polio cases has been found in London through sewage screening. The autoencoder-based system has the potential to be an additional layer of information for public health surveillance, allowing for early detection and more rapid response.