New AI tool could improve screening for hemoglobin disorders

Hemoglobin disorders such as HbS and HbC affect millions of people worldwide. Countries around the world spend substantial resources each year identifying carriers, often relying on detailed questionnaires, HPLC tests and, in many cases, costly genetic analyses.

Now, research from Aalborg University in Denmark indicates that artificial intelligence applied to routine blood parameters offer a more efficient and more precise approach.

“Our method makes it possible to identify potential carriers of hemoglobin disorders quickly and with very high accuracy – providing the foundation for a better targeted screening effort,” says Professor Izabela Ewa Nielsen of Aalborg University.

The findings have been published in peer reviewed journals including “Computer Methods and Programs in Biomedicine” and “Computers in Biology and Medicine”.

Data from UK participants

The new approach, known as AI HEMO, uses machine learning models trained on routine blood tests to identify carriers of hemoglobin disorders such as HbS and HbC.

A crucial part of the research draws on the UK Biobank dataset, where over 500,000 participants have contributed blood samples and genetic information. These data have allowed the team to develop and validate the method with accuracies of up to 95% in the study population.

If applied within the screening systems such as the UK’s, the method could potentially reduce the need for unnecessary follow up tests. Instead of relying on broad questionnaires or immediate genetic testing, health authorities could apply a more targeted approach.

“With AI HEMO, screening can become more precise and better targeted – which will in the future potentially reduce the costs of screen related procedures, while also reducing the administrative burden on pregnant women and families in high-risk areas,” says Izabela Ewa Nielsen.

Treatment for hemoglobin disorders include lifelong hospital monitoring for complications and blood transfusions. Potential curative options are bone marrow transplantation and gene therapy, but they remain available to few.

Fact box: AI HEMO

Method: Applies machine learning to routine blood tests to predict whether a person carries hemoglobin disorders, including HbS and HbC.

High accuracy: Catching up to 95 % of carriers (in the study populations), which is significantly higher than traditional screening methods based on questionnaires and basic blood indices.

Potential cost benefits: A better targeted screening programme potentially could reduce costs for NHS over time.

Strong validation: Tested on more than 500,000 UK Biobank participants and published in peer reviewed journals.

Fact box: Hemoglobin disorders (including HbS and HbC)

What are hemoglobin disorders?

Inherited conditions affecting the structure or production of hemoglobin. HbS can lead to sickle cell disease in its severe form, while HbC only causes substantial disease, if co-inherited with HbS.

How common are they?

Approx. 7 % of the global population carries a hemoglobin disorder. Prevalence varies widely across ethnic groups.

Why is this relevant for UK and other European countries?

Migration from regions with high prevalence – including parts of Africa, the Caribbean, the Middle East and Asia – has made screening for hemoglobin disorders increasingly important. If two carriers have a child, there is a 25 % risk the child will have the disorder.

Current screening practice:

The World Health Organization recommends screening for hemoglobinopathies. Often programs focus screening pregnant women, which requires additional tests that many individuals ultimately do not need.