Post

New research · Hematology
Current research in translational medicine · 23h
Cohort studyCurrent research in translational medicine · 2026

Non-invasive differentiation between aplastic anemia and myelodysplastic syndromes based on an 8-feature lightGBM model: A multicenter external validation study.

Sumei Wang, Zihan Cai, Wei Wei … Yurong Zhang
Read paper
HematologyCohort study

A new AI model accurately differentiates aplastic anemia from myelodysplastic syndromes.

Non-invasive differentiation between aplastic anemia and myelodysplastic syndromes based on an 8-feature lightGBM model: A multicenter external validation study.

Sumei Wang et al. · Current research in translational medicine · 2026
Background

Given the highly overlapping pancytopenic phenotypes, non-invasive differentiation between aplastic anemia (AA) and myelodysplastic syndromes (MDS) remains a formidable clinical challenge.

Methods

We retrospectively enrolled patients with histopathologically confirmed AA or MDS, partitioning them into a training set (n = 310) and an internal validation set (n = 131).

n = 310
97.30%
Results
The AI model correctly identified aplastic anemia or myelodysplastic syndromes 97.30% of the time.
n = 310
More results

Algorithmic intersection distilled the high-dimensional data into a parsimonious 8-feature panel (CHOL, hsCRP, IL-6, SAA, AGE, LDL-C, HRF, and IL-10).

The optimized LightGBM model demonstrated superior discriminative accuracy.

More results

SHAP analysis bridged mathematical predictions with pathophysiology, revealing that advanced age and an "inflammaging" cytokine profile (IL-6, hsCRP, SAA) are the strongest predictive features associated with MDS.

Conversely, distinct lipid remodeling (CHOL, LDL-C) and erythropoietic alterations (HRF) shifted the diagnostic probability toward AA.

“
Conclusion · 1 of 2

In conclusion, we established and externally validated a highly accurate, non-invasive 8-feature LightGBM diagnostic model.

Conclusion · 2 of 2

Translated into a practical clinical nomogram, this interpretable artificial-intelligence tool serves as an efficient "triage gatekeeper" to minimize unnecessary invasive biopsies, thereby facilitating precision triage in hematology.

Read paper
0 comments

No comments yet. Be the first.

Related papers

LatestFoundational
Study
92.3%
Share of treated patients whose lymphoma shrank or cleared after treatment
Randomized Trial
60%
Cumulative mortality was 60% with beta-glucan, reduced from 100% in untreated fish.
Cohort Study
HR 2.59
sudden kidney damage meant over twice the risk of death for CAR-T recipients
Study
30%
About 30% of peptides from multiple myeloma cell lines stimulated CD8+ T-cell responses.
AI / Informatics
C-index = 96.0%
The model accurately predicted when initial leukemia treatment would not work
Cohort Study
55.9% vs. 68.1%
Diabetic patients had lower major molecular response at 12 months (55.9% vs 68.1%).
Cohort Study
OR 3.8
Each additional low lab value associated with 3.8 times higher odds of Multisystem Inflammatory Syndrome in Children.
Cohort Study
30% higher risk
increased risk of new low blood cell counts per 0.1 g/L C3 decrease