# Ferritin Task Card Task id: `ferritin` Evidence status: grounded Evidence reviewed: 2026-07-29 Benchmark status: complete ## Target Ferritin as a continuous participant-visit scalar. Implementation facts: - bodily system: hematologic - target route: `blood_tests.bt__ferritin` - unit: ng/mL - acquisition/source: pinned Weizmann blood-test source (assay/device not documented in reachable HPP dataset docs); records selected by exact research stage, then participant-stage mean - default stages: `00_00_visit` and `02_00_visit` - collapse: participant-stage mean - range/QC: 1-5000 ng/mL broad plausibility guard - paper fidelity: HPP hematopoietic extension ## Why This Matters Ferritin is used to assess iron stores but is also an acute-phase reactant. Prediction does not by itself establish iron deficiency or overload. ## Related HPP Measurements Hemoglobin, MCV, iron/transferrin markers, CRP/inflammation, liver tests, sex, and menstrual factors. ## Benchmark-Track Evidence Population NMR/MS profiling finds sex-specific ferritin associations with fatty-acid species, branched-chain amino-acid catabolites, heme catabolites, and lipoprotein particles. This supports a Nightingale track for total ferritin. Because ferritin is an acute-phase reactant, a fixed panel containing GlycA may capture inflammation and cannot be interpreted as prediction of iron stores. ## Clinically Meaningful Variants Continuous prediction is primary. Iron-deficiency thresholds are guideline- and context-dependent: WHO defines iron deficiency as serum ferritin below 15 ng/mL in adults without inflammation and below 70 ng/mL in the presence of infection or inflammation, and the AGA uses below 45 ng/mL when evaluating anemic patients for iron-deficiency anemia. Ferritin is an acute-phase reactant, so any cutoff is assay- and inflammation-dependent. Because the HPP Weizmann ferritin assay and device are not documented here, these thresholds are external context and do not define labels for this Task without assay-compatible validation; this card does not authorize diagnostic bins. ## Baselines And Ceilings For V1→V2 evaluation, compare an age, sex, and BMI floor, prior-value LOCF, OLS on prior value plus allowed demographics, and GBDT on the same inputs. Use one cohort and split for all comparisons. ## Caveats - Mask same-window iron studies, ferritin aliases, and deterministic iron-status labels. - Inflammation, liver disease, infection, supplementation, sex, and menstrual status alter ferritin independently of iron stores. - A good score supports prediction of this measurement only; it does not establish diagnosis, causality, or intervention response. ## Metrics Primary R2. Also report RMSE, MAE, Pearson r, Spearman rho, split Ns, target quantiles, and same-cohort deltas versus LOCF. ## References - Lutsker et al. (2026), HealthFormer, . - Reicher et al. (2024), HPP longitudinal aging phenotypes, . - World Health Organization (2020), WHO guideline on use of ferritin concentrations to assess iron status, . - Garcia-Casal et al. (2021), ferritin as an index of iron deficiency and overload, . - Kaul et al. (2018), population metabolomic fingerprints of ferritin and hemoglobin, . - Ko et al. (2020), AGA guideline on iron-deficiency-anemia evaluation (ferritin <45 ng/mL), . - Cullis et al. (2018), BSH investigation of a raised serum ferritin, . - HPP documentation `blood_tests` dataset documentation and field/QC inventory. - PhenoBench issue-15 blood-source audit applies to every `blood_tests` Task in this card set. ## V0 Benchmark Execution The two same-visit V1 tracks completed on 2026-07-29; both Ridge/GBDT rows per track are locally successful and remotely `FINISHED` in operational source. Results predict total ferritin, not iron stores. - Nightingale: n=3,351; test=512; best R2=0.217 (GBDT), with paired delta R2 +0.063 [-0.001, 0.104] over the GBDT age/sex/BMI floor. - DXA, exploratory: n=2,178; test=334; best R2=0.209 (GBDT), with paired delta R2 +0.071 [-0.016, 0.107]. - Full rows and provenance: deep dive and aggregate review. ## Execution Evidence - protocol: paired HPP V1→V2; every baseline uses the same eligible participants - common cohort: n=394; train=264, validation=70, test=60 | Baseline | Strategy | Test R2 | RMSE | Pearson r | | --- | --- | ---: | ---: | ---: | | Demographic | `demographic_floor` | -0.009 | 71.037 | 0.258 | | Linear | `linear` | 0.735 | 36.422 | 0.867 | | Trees | `gbdt_delta_r2` | 0.487 | 50.631 | 0.774 | | LOCF | `last_observation_carried_forward` | 0.685 | 39.656 | 0.868 | ## Open Questions - Domain-expert review remains outstanding; evidence grounding and real-run gates are complete.