{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:QAJH2MII72FDAZMHP3X7EQUYF3","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"c51441026965d32fd624f5028d2393bbaf0f1693ec02f0e5256c46d8ac55543a","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-15T21:56:19Z","title_canon_sha256":"792da533680eabc6ce9103b0dbc5f5f54a5b3a5af5bfc4bf6ee8d79acfb6002e"},"schema_version":"1.0","source":{"id":"2607.14385","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2607.14385","created_at":"2026-07-17T00:21:10Z"},{"alias_kind":"arxiv_version","alias_value":"2607.14385v1","created_at":"2026-07-17T00:21:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.14385","created_at":"2026-07-17T00:21:10Z"},{"alias_kind":"pith_short_12","alias_value":"QAJH2MII72FD","created_at":"2026-07-17T00:21:10Z"},{"alias_kind":"pith_short_16","alias_value":"QAJH2MII72FDAZMH","created_at":"2026-07-17T00:21:10Z"},{"alias_kind":"pith_short_8","alias_value":"QAJH2MII","created_at":"2026-07-17T00:21:10Z"}],"graph_snapshots":[{"event_id":"sha256:796e9edd31b36d54c3197b9804b72b04eb5c4986a88e61c43342d2704a23bc56","target":"graph","created_at":"2026-07-17T00:21:10Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2607.14385/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models achieve strong scores on medical benchmarks, yet these benchmarks evaluate each question in isolation, providing no measure of whether a system can distinguish clinically similar presentations requiring different interventions. We introduce MamaBench, the first counterfactual benchmark for maternal and paediatric AI: 434 expert-authored clinical narratives in 217 pairs across 371 pathologies, evaluated via the Bias Trap Rate (BTR), the conditional probability that a model fails the counterfactual given success on the base case. We propose Evidence-Anchored RAG (EA-RAG), a","authors_text":"Abiodun Adereni, Angel Ezendu, Anuoluwa Sotome, Ifeoma Nwabueze, Oluwadarasimi Oguntuase, Oluwafunke Akinbuwa, Oluwasegun Oguntuase, Oluwaseun Odunsi, Samuel Okoko, Thanni Adewuyi","cross_cats":["cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-15T21:56:19Z","title":"MamaBench: Benchmarking LLM Robustness in Maternal and Child Health Diagnosis through Counterfactual Clinical Perturbation"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.14385","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:91c2ac89d3ebea60667e468eae1eab6d86ac16480610ecce474222a4cccbf661","target":"record","created_at":"2026-07-17T00:21:10Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"c51441026965d32fd624f5028d2393bbaf0f1693ec02f0e5256c46d8ac55543a","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-15T21:56:19Z","title_canon_sha256":"792da533680eabc6ce9103b0dbc5f5f54a5b3a5af5bfc4bf6ee8d79acfb6002e"},"schema_version":"1.0","source":{"id":"2607.14385","kind":"arxiv","version":1}},"canonical_sha256":"80127d3108fe8a3065877eeff242982ed59fdbbf6572700f6554db64adcb0aeb","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"80127d3108fe8a3065877eeff242982ed59fdbbf6572700f6554db64adcb0aeb","first_computed_at":"2026-07-17T00:21:10.031899Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-17T00:21:10.031899Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"LNsLz7sCUjuaGMBVVcZewqRKBE50zzFVCNGRlCNi+nJniHcbIrljvl5aBxkVeBRCJ2+1U8sTEIX8EULRBeTYBQ==","signature_status":"signed_v1","signed_at":"2026-07-17T00:21:10.032763Z","signed_message":"canonical_sha256_bytes"},"source_id":"2607.14385","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:91c2ac89d3ebea60667e468eae1eab6d86ac16480610ecce474222a4cccbf661","sha256:796e9edd31b36d54c3197b9804b72b04eb5c4986a88e61c43342d2704a23bc56"],"state_sha256":"2d82bb3522e45bf943714774df71aff450edb44f62f9a8524334ca0040c623c5"}