BloodCounts!

Artificial Intelligence for the Complete Blood Count Test

The complete blood count (CBC)* is the most commonly used medical test, with 3.6 billion CBCs done per year worldwide. The test is used frequently because it can screen for a wide range of conditions, from anaemia to infection.

The instruments that run CBC tests generate far richer data than the parameters routinely reported to doctors, nurses and midwives; however, this additional data is often discarded, despite containing clinically valuable signals for detecting signs of impending ill-health before conditions present themselves.

BloodCounts! will develop a DeepCBC foundational model (FM) of blood using the full parameters of the raw CBC measurement data and supplementary blood smear images from our research partners in Singapore to amplify the medical diagnostic value of the CBC test. AI analysis will be used to enhance the FM’s predictive and inferential capabilities.

The large-scale FM to be developed by the BloodCounts! team in Singapore will be the first to be trained on CBC measurement data under clinical conditions in an agnostic manner, meaning it is not optimised for any single disease, making it adaptable to a wide range of ill-health conditions.

*Complete Blood Count (CBC) is also known in the UK and Singapore as Full Blood Count (FBC)

CBC measurement data captured at our partner hospitals and curated
Use of curated anonymised data for FM development
Return of the FM to the hospital’s trusted compute research environment
Analysis of the FM’s performance by using electronic health record data
FM development and application to specific clinical use cases, focusing on stroke and lung cancer
Connecting Singapore to the federated learning network of the BloodCounts! consortium in other countries
Deployment of the FM at our partner hospitals to fine-tune the model
Prepare the path for regulatory approval in Singapore and the UK

The project will converge science from maths, computer science, bioinformatics, and clinical medicine to leverage AI for healthcare.

Prior research since 2021 in the UK has built a federated learning network and generated a first FM with the collaboration of academic research centres via the international BloodCounts! Consortium, which includes partners in the UK, the Netherlands, Belgium, Gambia, Ghana and India.

The Singapore-specific research will access health data from a new population to develp a local FM with the aim of clinical use in Singapore. The FM’s deployment will be refined through collaborative global development, with the Singapore data eventually connecting to the existing federated learning network of the BloodCounts! consortium.

BloodCounts! will contribute to Singapore’s hub for cutting-edge AI method development, infrastructure, and local training, while also becoming a pioneer in secure data governance and large-scale AI clinical modelling for the public good. For healthcare, improved screening capabilities and diagnostics will aid earlier intervention for several diseases in Singapore and worldwide.

This research is supported by the National Research Foundation, Prime Minister’s Office, Singapore under its Campus for Research Excellence and Technological Enterprise (CREATE) programme. This is one of eight projects supported by NRF’s new SGD$120m AI-for-Science (AI4S) initiative announced in June 2026.