This Key Blood Measurement Could Signal Your Future Disease Risk

If you’ve ever had a blood test ordered by a doctor, chances are it included a complete blood count (CBC). As one of the most common medical tests worldwide, CBCs are performed billions of times each year to diagnose conditions and monitor overall health.

Despite their widespread use, the way clinicians interpret CBC results can often be imprecise. Currently, these tests rely on standardized reference intervals that don’t account for individual differences, which may limit their accuracy.

At the University of Washington School of Medicine, my team and I are working to improve clinical blood testing by applying computational tools. In collaboration with the Higgins Lab at Harvard Medical School, we analyzed 20 years of blood count data from tens of thousands of patients across the U.S. Using machine learning, we developed methods to identify personalized blood count ranges and predict future disease risks.

How CBCs and Clinical Tests Work

Unlike purely diagnostic tests, such as those for pregnancy or COVID-19, which give a straightforward positive or negative result, most clinical tests measure biological traits that fluctuate within a regulated range.

A CBC test provides a detailed profile of your blood, including red and white blood cell counts and platelet levels. These markers are crucial across nearly all areas of medicine. For instance:

• Hemoglobin, an iron-containing protein in red blood cells, indicates oxygen-carrying capacity. Low levels may suggest iron deficiency.
• Platelets, which help form blood clots, can indicate internal bleeding if their levels are too low.
• White blood cells, essential to the immune system, often increase in response to infections.

Redefining “Normal”

Currently, clinicians use reference intervals based on the middle 95% of values from healthy individuals to define “normal” ranges. However, these population-based intervals don’t account for individual variability, which is largely influenced by genetics and environment.

For example, while the standard normal platelet range is 150 to 400 billion cells per liter, your personal set point—the value your body naturally regulates—might be closer to 200, with a narrower healthy range of 150 to 250.

This mismatch between individual set points and generalized reference intervals can lead to misdiagnoses. Doctors may overlook disease symptoms if your personal set point differs significantly from population averages or conduct unnecessary tests if your results hover near a cutoff.

Using Machine Learning to Personalize Lab Results

Many patients undergo routine CBCs during annual checkups, creating a rich data history. By applying machine learning to this data, we estimated blood count set points for over 50,000 patients and discovered that individual normal ranges are about three times narrower than population-based ranges. For instance, while the standard white blood cell range is 4.0 to 11.0 billion cells per liter, most individuals’ true ranges fall between 4.5 to 7.0 or 7.5 to 10.0.

Interpreting test results based on personalized set points improved the detection of diseases like iron deficiency, chronic kidney disease, and hypothyroidism. It also allowed us to identify subtle changes that might go unnoticed when using broader reference ranges.

Predicting Future Disease Risk

Interestingly, individual set points also proved to be strong predictors of future health risks. For example, patients with higher white blood cell set points were more likely to develop Type 2 diabetes and faced nearly twice the risk of death from any cause compared to those with lower counts. Other blood count markers also correlated with disease and mortality risks.

The Future of Personalized Medicine

Incorporating personalized set points into clinical practice could revolutionize how diseases are screened and diagnosed. By leveraging your medical history to define what “healthy” truly means for you, doctors could provide more accurate and tailored care. This approach represents a promising step forward in the field of personalized medicine.

Read Original Article: Science Alert

Read More: Scitke