Blood
Iron deficiency anemia
Living starter kit
Not yet expert-checkedUpdated Nov 13, 2026Start building on iron deficiency anemia
Iron-deficiency anemia means both too little hemoglobin and depleted iron stores, so a model that only estimates hemoglobin, from a photo of the eye for example, isn't diagnosing it. This kit starts with U.S. survey lab data that can tell the two apart and an eye-photo dataset for anemia screening; a dataset linking photos to iron tests is still missing.
Know iron deficiency anemia well? No expert has checked this kit yet.Check our work →Open datasets2
- NHANES 2017–2018 ferritin data ↗
Ferritin (iron store) measurements from the U.S. National Health and Nutrition Examination Survey, which can be joined to blood counts to study iron deficiency with and without anemia.
ClinicalOpen downloadLicense not stated
- Ghana eyelid photos for anemia screening ↗
Cropped photos of the inner eyelid from Ghana, labeled anemic or not by hemoglobin test rather than iron tests, from the University of Energy and Natural Resources.
710 original images, expanded to 4,260 by augmentation
ImagingOpen downloadCC BY 4.0
Benchmarks and tasks1
- Blood-count classifier for iron-deficiency anemia (2024) ↗Tell iron-deficiency anemia apart from other anemias using routine blood-count results, against a published model trained on U.S. survey data. Clinical
Open questions4
- 01
Can routine blood counts tell iron-deficiency anemia apart from other causes, without the model seeing the answer?
The published model's labels are partly defined by hemoglobin, which is also one of its inputs, so its accuracy may be overstated.
ClinicalSource: Classification and Explanation of Iron Deficiency Anemia ↗
- 02
Can a phone photo of the eye detect iron deficiency, not just low hemoglobin?
Photo screening is cheap, but the available photo datasets are labeled by hemoglobin only, not by iron tests.
ImagingClinicalSource: Physiologically based ferritin thresholds ↗
- 03
How should iron labels be corrected for inflammation in different populations?
Inflammation raises ferritin and can hide iron deficiency; the BRINDA project provides correction methods.
- 04
Do the newest ferritin cutoffs for iron deficiency hold outside the United States?
They were derived from U.S. survey data, and their authors call for testing elsewhere.
ClinicalEpidemiologySource: Physiologically based ferritin thresholds ↗
Who's working on this4
Pullakhandam and McRoy, University of Wisconsin–Milwaukee
Two researchers published a model that detects iron-deficiency anemia from blood counts, with their code.
A concrete baseline to reproduce and improve, with code you can inspect.
No public invitation foundEvidence, Mar 1, 2024 ↗
BRINDA research group
The group develops methods for correcting iron and other nutrient measurements for inflammation.
Its code and tutorials help correct iron labels for inflammation.
No public invitation foundEvidence, date unknown ↗
Ferritin-threshold researchers, CDC and Columbia
Researchers who derived ferritin cutoffs for iron deficiency from U.S. survey data.
Their definitions separate iron deficiency with and without anemia, which matters for labels.
No public invitation foundEvidence, date unknown ↗
Eye-photo researchers, University of Energy and Natural Resources
The researchers released consented eye photos for hemoglobin-based anemia screening.
An imaging starting point, though its labels are anemia, not iron deficiency.
No public invitation foundEvidence, Jun 27, 2022 ↗
Not in the kit yet
- No dataset yet pairs eye or nail photos with iron tests, which is what photo-based iron screening needs.
- Only one benchmark so far: a published model whose labels overlap with its inputs, so read its method before comparing.
- The U.S. survey data states no license; check terms, and use the survey weights for population estimates.
- Iron deficiency without anemia and anemia caused by inflammation are separate conditions, and no dataset here labels them in photos.
This kit keeps growing as new data, benchmarks and teams appear. Know something we missed? Tell us →
Curated by the Unsolved project team. Every link was checked on the date shown; licenses are as stated by each source.
Healthy years lost (DALYs), 2023
39.3M
95% range 24.8M to 63.5M
Deaths, 2023
Not modeled
People living with it, 2023
1.4B
GBD cause: Dietary iron deficiency (partial match)
Research attention
What there is to build on
Open Targets has no entry for this disease, so there is no target or drug data yet.
Identifiers
- MeSH
- D018798
Read these numbers with care
- Burden figure covers only part of this disease: GBD cause is dietary iron deficiency only.
- No Open Targets entry, so readiness is not scored.
- Only a phenotype term in Open Targets.
Trial and paper counts are rough measures of attention, not of progress. This page is for research planning, not medical advice.