Similarity Analysis of Blood Count Reference Intervals Across Continents Reveals No Reproducible Population or Geography-Linked Structure and Supports Personalised Values
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Genomics Radio. Generated commentary on the latest computational biology and genomics papers.
Ines: I'm Ines, and with me are Marcus and Yuki, guest researcher.
Marcus: Today's paper: "Similarity Analysis of Blood Count Reference Intervals Across Continents Reveals No Reproducible Population or Geography-Linked Structure and Supports Personalised Values".
Ines: The gist: CBC reference intervals do not exhibit reproducible continent-level or geography-linked structure in the compiled global datasets, suggesting they reflect properties of reference-interval systems rather than population averages,
Marcus: First, who's behind it and why it matters.
Paper summary: Ines: To wrap up this discussion on the paper "Similarity Analysis of Blood Count Reference Intervals Across Continents Reveals No Reproducible Population or Geography-Linked Structure and Supports Personalised Values", the core finding is that CBC reference intervals do not show any reproducible continent-level or geography-linked structure when analyzed across different methods.
Marcus: They found that the analysis showed uniformly high cohesion scores for CBC, averaging around one point two seven to one point three zero according to page two of this work > This lack of structure is what they used to argue against the idea that these intervals represent universal biological standards based on population averages.
Yuki: So, what does this mean practically? It suggests that the current systems are reflecting methodological and historical variability rather than actual underlying physiological population structure across those different countries.
Ines: Right. The authors are using this lack of structure to support a shift toward personalized reference frameworks, meaning we should be looking at longitudinal individual baselines instead of just applying broad, inherited standards >
Marcus: They also pointed out the gaps they see in current practice, like regional bias in data sources from places such as NHANES and Japan with limited representation from Africa, Middle East, South Asia, and Latin America according to page three of this work > <ref:2511.09584#pg1>
Yuki: So it’s a call for greater equity concerns when developing these standards because the current data sources are not representative of the global population.
Ines: That’s the connection they are making between their analysis of these intervals and broader diagnostic equity issues across different regions.
Marcus: The paper itself notes a limitation, which is that CBC data are only available as reference intervals, not raw individual-level distributions, so future work needs to use personalized CBC datasets with analyser metadata and longitudinal measurements to truly distinguish biological homogeneity from structure masked by measurement heterogeneity >
Yuki: So the next step in this research isn't just analyzing these published intervals but getting the actual raw data from individuals over time.
Ines: That’s right. The paper concludes that their analysis of CBC reference intervals doesn't encode coherent global structure and it provides limited support for universal population-based diagnostic thresholds >
Conclusion: Ines: So we're wrapping up on this paper about CBC reference intervals across different countries and what that actually tells us about biology.
Marcus: It’s basically saying that when you look at all those reference interval sets from twenty-eight countries, they don’t have any real pattern based on where the people live or which continent they're on.
Yuki: The authors used a bunch of statistical methods like variability mapping and clustering to test this idea.
Ines: And what they found is that the CBC data showed no reproducible geography-linked structure across all those different methods they tested.
Marcus: They even contrasted it with body mass index, which *did* show clear continent-level clustering, so the comparison really hammers home the point for CBC.
Yuki: So the authors are arguing that these reference intervals aren't reflecting some kind of universal biological standard for everyone globally.
Ines: Exactly. The implication they pull from this is that we shouldn't just rely on those broad, inherited standards anymore.
Marcus: They suggest moving toward recalibrated frameworks based on individual longitudinal baselines instead of one fixed population average for diagnostics.
Yuki: That makes sense because the way populations evolve and how their physiology changes over time means a single global interval might not fit anyone well.
Ines: And the authors are pointing out that we need to look at those individual baselines, not just the reference intervals themselves.
Marcus: It’s a big shift in how we think about setting diagnostic thresholds for blood counts across different regions.
Yuki: If you want to get a sense of what this means for the wider field, we could look at how this lack of structure impacts things like disease surveillance or clinical trials that rely on standardized measurements.
Algorithmic Dynamics Lab, King’s Institute for Artificial Intelligence, King’s College London · Oxford Immune Algorithmics, Oxford University Innovation & London Institute for Healthcare Engineering · Departments of Biomedical Computing and Digital Twins, School of Biomedical Engineering and Medical Sciences, King’s Faculty of Life Sciences and Medicine, King’s Health Partners AHSC (NHS–KCL), King’s College London
q-bio.OT
Submitted: 2025-11-12
Updated: 2026-05-04
Comments: 25 pages
Journal ref: npj Digit. Public Health 1, 32 (2026)
DOI: 10.1038/s44482-026-00035-5
Code: https://github.com/kunlin889797/Proximity-and-Similarity-Analysis-of-Complete-Blood-CountAuthor
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
Importance score: 69/100
The gist: The gist: CBC reference intervals do not exhibit reproducible continent-level or geography-linked structure in the compiled global datasets, suggesting they reflect properties of reference-interval
Key concepts
- Reference Intervals (RIs)
- These are the specific ranges of lab results considered normal for a population. The study examined how these ranges vary globally to see if they follow consistent patterns across different countries and continents, suggesting they might be based on local practices rather than shared biological truths.
- Variability Mapping
- This is a statistical method used to visualize and analyze how similar or different data points are across multiple datasets. The researchers used it to map the similarity between CBC reference intervals from various countries, helping them identify patterns in how these ranges relate to each other.
- Cohesion Benchmarking
- This technique measures how tightly grouped or cohesive a set of data is based on different mathematical distances. It was used to test if the clustering observed in BMI was robust across different distance metrics, providing a way to quantify the strength of geographic groupings.
- Nonlinear Manifold Visualisation
- This is a complex visualization tool that helps map high-dimensional data into lower dimensions while preserving important relationships. It was used to visually inspect the structure of CBC reference intervals and confirm whether any hidden, continent-level organizational patterns existed.
Terminology
Summary
The gist: CBC reference intervals do not exhibit reproducible continent-level or geography-linked structure in the compiled global datasets, suggesting they reflect properties of reference-interval systems rather than population averages, while BMI does.
How it works
The study examined the structure of published Complete Blood Count (CBC) reference intervals across 28 countries using a multi-stage framework combining variability mapping, hierarchical clustering, informationtheoretic distances, cohesion benchmarking, and nonlinear manifold visualization The analysis aimed to evaluate whether currently reported RIs encode reproducible cross-population organisation or instead reflect heterogeneous clinical derivation practices The researchers compiled CBC RI data from 28 countries and analyzed their similarity using this multi-stage framework
Data Collection and Methodology
The data collection involved sourcing CBC RIs from various origins, including official national websites, peer-reviewed studies, university hospitals, and local health authorities The dataset comprised RIs for WBC, Platelets, MCHC, MCV, MCH, RBC, HGB, and HT across these 28 countries Body mass index (BMI) served as a methodological positive benchmark and exhibited clear continent-level clustering with mean cohesion scores around 0.78–0.81 The researchers applied multiple clustering strategies to both CBC and BMI, benchmarking their geographical sensitivity using the Two-Level Cohesion Score
Key Findings on CBC Structure
The analysis revealed that CBC reference intervals showed no reproducible geography-linked clustering across analytical methods, with uniformly high cohesion scores (mean ≈ 1.27–1.30) Weak signals in red-cell indices like MCV and HGB were unstable across sexes and distance metrics The absence of structure should not be interpreted as evidence that current CBC reference intervals represent universal biological standards, but is more consistent with the fragmented and historically inherited nature of the global RI landscape Cross-national variability in WBC and MCV RIs displayed modest overall dispersion with no reproducible continent-level structure, confirming that between-population variability is low relative to within-population variance
Contrast with BMI Structure
In contrast to CBC, BMI clustering formed continent-coherent groupings under the same modelling framework, supporting the validity of the analytical pipeline The contrast between CBC and BMI directly supports the interpretation that CBC lacks stable geo-physiological boundaries, whereas BMI retains them The analysis showed that BMI exhibited substantially lower cohesion scores, particularly under information-theoretic and correlation-based distances, demonstrating that it encodes genuine geographic structure Furthermore, the mutual-information feature ranking in the BMI trees showed that geography was driven by mid- to late-adulthood BMI intervals
Robustness and Conclusion
The study employed several robust checks to confirm the null result, including phylogenetic tree clustering across diverse linkage–distance combinations and Robinson–Foulds tree-distance analysis The Robinson–Foulds (RF) analysis showed high topological instability across all six linkage–distance combinations, confirming that CBC reference-interval topology is highly metric-dependent and does not reflect a stable biological signal UMAP embeddings and feature importance analyses also showed that no latent spatial separation exists in CBC values, regardless of whether reference intervals were represented by midpoints or alternative representations like lower limits or interval widths Collectively, the lack of separation across all four analytical layers constitutes convergent evidence that CBC reference intervals do not encode continent-level structure The findings suggest a transition from population-based to individual-based reference systems, with CBC values interpreted against personal baselines rather than presumed universal norms
Limitations and Future Directions
The principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work will require large-scale individual-level CBC datasets with diverse cohorts and longitudinal measurements to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study concludes that the absence of geography-linked CBC structure reflects properties of published reference-interval systems rather than an analytic artefact The findings support a transition toward recalibrated and ultimately personalised reference frameworks based on longitudinal individual baselines and harmonised derivation standards
Competing Interests
One author holds editorial board positions; however, all editorial roles were fully recused from the handling, review, and decision process for this manuscript, which was managed independently in accordance with journal policy The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference intervals, not raw individual-level distributions Future work should address these gaps by using personalised CBC datasets with analyser metadata, multiomics profiling, environmental covariates, and longitudinal sampling to distinguish true biological homogeneity from structure masked by measurement heterogeneity or RI summarisation The study’s principal limitation is that CBC data are available only as reference
Improvements for AI systems
-
Bold Header: Develop a geography-adaptive reference interval (RI) system for Complete Blood Count (CBC). This improved AI system can dynamically adjust diagnostic thresholds based on continent or region, moving away from
universal population-based diagnostic thresholds
as suggested by the paper's conclusion. -
Bold Header: Implement robust structural validation across multiple analytical layers. The improved AI can employ a combination of
variability mapping, hierarchical clustering, informationtheoretic distances, cohesion benchmarking, and nonlinear manifold visualisation
to ensure that any detected structure isbiologically anchored rather than a methodological limitation.
-
Bold Header: Integrate gender-specific and age-stratified modeling for enhanced precision. The system can leverage the finding that
BMI values exhibited clear continent-level clustering under the same modelling framework,
allowing for contextually relevant risk assessment by factoring in life stages, as indicated by theage-anchored concentration of MI signal.
-
Bold Header: Utilize metric-independent topological analysis for RI comparison. The AI can use
Robinson–Foulds tree-distance analysis
to compare linkage–distance combinations, ensuring that clustering results are not merely anartifact of modelling or metric choice,
as evidenced by the finding thatmean RF scores were 0.77, confirming that each method produces a largely unique dendrogram topology.
-
Bold Header: Transition diagnostic interpretation toward individual longitudinal baselines. The system can be designed to prioritize the paper's recommendation:
personalised longitudinal stability, rather than population averages, anchors clinical interpretability,
by trackinglongitudinal individual baselines
instead of relying on static, aggregated reference intervals.
Abstract
Blood reference intervals (RIs) underpin diagnostic interpretation and therapeutic monitoring worldwide. However, many widely used RI systems originate from limited historical cohorts and have been propagated across health systems without harmonised derivation protocols, shared metadata, or cross-population validation. Consequently, the global RI landscape reflects a heterogeneous mixture of legacy standards and local laboratory practices rather than a biologically grounded framework. Here we examine published Complete Blood Count (CBC) reference intervals, one of the most commonly used laboratory panels worldwide. We compiled CBC RI data from 28 countries and analysed their similarity using variability mapping, hierarchical clustering, information-theoretic distances, cohesion benchmarking, and nonlinear manifold visualisation. Body mass index (BMI) served as a methodological positive benchmark and exhibited clear continent-level clustering (mean cohesion approximately 0.78-0.81). In contrast, CBC reference intervals showed no reproducible geography-linked clustering across methods, with uniformly high cohesion scores (mean approximately 1.27-1.30). Weak signals in red-cell indices (MCV, HGB) were unstable across sexes and distance metrics. This absence of structure should not be interpreted as evidence that current CBC reference intervals represent universal biological standards. Rather, it is more consistent with the fragmented and historically inherited nature of the global RI landscape. These findings indicate that published CBC reference intervals do not encode coherent global structure and provide limited support for universal population-based diagnostic thresholds. Instead, they support a transition toward recalibrated and personalised reference frameworks based on longitudinal individual baselines and harmonised derivation standards.