{"paper":{"title":"Interpretable Machine Learning for Antepartum Prediction of Pregnancy-Associated Thrombotic Microangiopathy Using Routine Longitudinal Laboratory Data","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"Gradient boosting on routine longitudinal lab tests predicts pregnancy-associated thrombotic microangiopathy risk with AUROC 0.872.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chuanchuan Sun, Feng Yu, Qin Fan, Qingchao Chen, Zhen Yu","submitted_at":"2026-05-13T17:07:00Z","abstract_excerpt":"Background: Pregnancy-associated thrombotic microangiopathy (P-TMA) is rare but life-threatening. Early risk prediction before overt clinical presentation remains challenging, as the associated laboratory abnormalities are subtle, multidimensional, and frequently masked by common physiological changes such as gestational thrombocytopenia and pregnancy-related proteinuria, thus overlapping heavily with benign obstetric and renal conditions. This complexity is poorly captured by univariate or rule-based approaches; however, it is addressable by machine learning, which can extract latent, time-de"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Gradient boosting was selected by cross-validation; the model achieved an AUROC of 0.872 (95% CI: 0.769-0.952) and AUPRC of 0.883 (95% CI: 0.780-0.959) in the held-out test cohort, with sensitivity 0.750 and specificity 0.812.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the single-center retrospective cohort of 300 pregnancies is representative of future patients and that the held-out test performance will generalize without external validation or prospective testing.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Gradient boosting on 146 longitudinal lab predictors from routine prenatal care achieves AUROC 0.872 on held-out test data for antepartum P-TMA risk prediction.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Gradient boosting on routine longitudinal lab tests predicts pregnancy-associated thrombotic microangiopathy risk with AUROC 0.872.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"ae7b240330a0afb2e2926edd5cadcdbadff79e5b20c0fb01041448b3972185bc"},"source":{"id":"2605.13786","kind":"arxiv","version":1},"verdict":{"id":"191b6187-8245-4691-8ae1-04c5a350eced","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:12:24.161586Z","strongest_claim":"Gradient boosting was selected by cross-validation; the model achieved an AUROC of 0.872 (95% CI: 0.769-0.952) and AUPRC of 0.883 (95% CI: 0.780-0.959) in the held-out test cohort, with sensitivity 0.750 and specificity 0.812.","one_line_summary":"Gradient boosting on 146 longitudinal lab predictors from routine prenatal care achieves AUROC 0.872 on held-out test data for antepartum P-TMA risk prediction.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the single-center retrospective cohort of 300 pregnancies is representative of future patients and that the held-out test performance will generalize without external validation or prospective testing.","pith_extraction_headline":"Gradient boosting on routine longitudinal lab tests predicts pregnancy-associated thrombotic microangiopathy risk with AUROC 0.872."},"references":{"count":3,"sample":[{"doi":"","year":2024,"title":"Thrombotic microangiopathy in pregnancy: Current understanding and management strategies[J]","work_id":"5aadc0ac-fe44-47b0-b637-27f01dc71df0","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Machine learning models for predicting preeclampsia: A systematic review[J]","work_id":"96e87f3c-216b-458b-9c08-9eb09b94acde","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2013,"title":"Cystatin C versus creatinine in determining risk based on kidney function[J]","work_id":"7331c53e-2ce1-4f7e-8e64-52fa83e8cdc8","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":3,"snapshot_sha256":"ada3a8ccfb90f3a2c958653f52bcf11dcb37fec794acd74f94191576a64e38e9","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}