Machine learning-based prediction of inflammation adjusted iron deficiency anaemia using blood cell indices
An algorithm using only routine blood counts diagnoses iron deficiency anemia accurately without ferritin testing, crucial for low-resource settings.
A random forest ML model using only CBC blood indices predicted inflammation-adjusted iron deficiency anemia with ~95% sensitivity and specificity in both US (NHANES) and Indian validation cohorts, without requiring serum ferritin measurement. The model also predicted hemoglobin response to iron therapy, providing a clinically useful CBC-only IDA diagnostic that is particularly valuable in low-resource settings where ferritin testing is unavailable.
What the study was
- Study design
- ML model development and external validation
- Population
- Women of reproductive age, NHANES 2017-2023 (n=3604) + Indian validation cohort (n=381)
- Sample size
- 3985
- Category
- Diagnostics
- Maturity
- Validated
- Journal
- Indian J Med Res
Why it surfaced
High sensitivity/specificity for IDA without ferritin testing; validated across US + Indian populations; highly relevant for global health and underserved settings.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.