Early detection of infectious disease outbreaks is paramount for mounting effective public health measures. Current approaches used for infectious disease surveillance relies on prior knowledge of the pathogen and, as illustrated by the current pandemic, they are inadequate for early detection of novel threats.

There is an urgent need for simple, affordable, and scalable methods that can be applied in High-, Middle-, and Low-Income Countries (HIC, LMIC) which can detect infectious disease outbreaks in real-time and are pathogen-agnostic.

In the BloodCounts! team we believe that data from the Full Blood Count test can be used to predict infectious disease outbreaks.

What is a Full Blood Count?

The Full Blood Count (FBC) is an essential test used to inform medical decision making. It is the world’s most common medical laboratory test - being performed and estimated 3.6 billion times per year.

About Global FBC testing. Numbers represent millions of tests, coloured bars represent the proportion of tests performed in primary (green) and secondary (orange) healthcare settings.

During the 30-second FBC test, an automated machine called a haematology analyser is used to measure the characteristics of the seven main blood cell types.

In short, the cells in the blood sample are dye-stained and then laser-illuminated one-by-one from different angles - the scatter pattern of each laser is then used to identify the type of cell, for example a Red Blood Cell, and count it!

Despite the wide-scale use of the test, results are generally interpreted on a case-by-case basis instead of being compared to all other samples in a population. Additionally, once high-level results such as Red Cell Count are calculated, data from the many thousands of laser measurements performed during the test are generally discarded!

This is where the BloodCounts! team step in.

Machine Learning and FBC data

As you can see in the figure below, different illnesses and pathogens stimulate unique responses in the human immune system - and this is observable in FBC data.

About FBC test laser scatter plots of immune cells in a healthy person, a person infected with influenza, and a person infected with SARS-CoV-2.

We have developed sophisticated machine learning models which are trained using all the historical FBC data from a population. These models can then be used to detect anomalies in current FBC tests from the same population - a rise in the number of anomalous samples from unrelated people could be a warning sign of that an outbreak of infectious disease has occurred.

You can think of the BloodCounts! solution as a tsunami-like warning system for infectious disease outbreaks which uses the human immune response measured by the FBC test.

About The BloodCounts! workflow.

A major advantage is that our method requires zero knowledge of the causative pathogen to work!