{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:GBDOXIASUMMFZCXAJB5JAIW7CA","short_pith_number":"pith:GBDOXIAS","schema_version":"1.0","canonical_sha256":"3046eba012a3185c8ae0487a9022df100229608426c91e84e861eccdef331d88","source":{"kind":"arxiv","id":"2602.02560","version":2},"attestation_state":"computed","paper":{"title":"Auditing Sybil: Explaining Deep Lung Cancer Risk Prediction Through Generative Interventional Attributions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Sybil lung cancer risk model differentiates malignant nodules like radiologists but shows sensitivity to artifacts and radial bias.","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Bartlomiej Sobieski, Jakub Grzywaczewski, Karol Dobiczek, Mateusz W\\'ojcik, Matthew Tivnan, Patryk Szatkowski, Przemyslaw Biecek, Przemys{\\l}aw Bombi\\'nski, Tomasz Bartczak","submitted_at":"2026-01-30T15:21:52Z","abstract_excerpt":"Lung cancer remains the leading cause of cancer mortality, driving the development of automated screening tools to alleviate radiologist workload. Standing at the frontier of this effort is Sybil, a deep learning model capable of predicting future risk solely from computed tomography (CT) with high precision. However, despite extensive clinical validation, current assessments rely purely on observational metrics. This correlation-based approach overlooks the model's actual reasoning mechanism, necessitating a shift to causal verification to ensure robust decision-making before clinical deploym"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":true},"canonical_record":{"source":{"id":"2602.02560","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-01-30T15:21:52Z","cross_cats_sorted":["cs.AI","cs.CV"],"title_canon_sha256":"18a985f5271c018504f55e857e5f30a718381a6295cd8c304aaa0fb2c9509ecd","abstract_canon_sha256":"14ad4cf085388ad838427c38dd45571f13481b8d9775c6232204702826344905"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:45:05.588475Z","signature_b64":"zbW4U3yYElxzuxU8v1LKHBfZDgut0dYdSaHxVVNsm6AZyU4j4/I32Dr+8EALQnLhCwFkXom6TBTlKysLiPZnBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3046eba012a3185c8ae0487a9022df100229608426c91e84e861eccdef331d88","last_reissued_at":"2026-05-18T02:45:05.587573Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:45:05.587573Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Auditing Sybil: Explaining Deep Lung Cancer Risk Prediction Through Generative Interventional Attributions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Sybil lung cancer risk model differentiates malignant nodules like radiologists but shows sensitivity to artifacts and radial bias.","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Bartlomiej Sobieski, Jakub Grzywaczewski, Karol Dobiczek, Mateusz W\\'ojcik, Matthew Tivnan, Patryk Szatkowski, Przemyslaw Biecek, Przemys{\\l}aw Bombi\\'nski, Tomasz Bartczak","submitted_at":"2026-01-30T15:21:52Z","abstract_excerpt":"Lung cancer remains the leading cause of cancer mortality, driving the development of automated screening tools to alleviate radiologist workload. Standing at the frontier of this effort is Sybil, a deep learning model capable of predicting future risk solely from computed tomography (CT) with high precision. However, despite extensive clinical validation, current assessments rely purely on observational metrics. This correlation-based approach overlooks the model's actual reasoning mechanism, necessitating a shift to causal verification to ensure robust decision-making before clinical deploym"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Providing the first interventional audit of Sybil, we demonstrate that while the model often exhibits behavior akin to an expert radiologist, differentiating malignant pulmonary nodules from benign ones, it suffers from critical failure modes, including dangerous sensitivity to clinically unjustified artifacts and a distinct radial bias.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the 3D diffusion bridge produces modifications that isolate true causal contributions without introducing new confounding artifacts that the model could latch onto.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"S(H)NAP audits Sybil via generative interventions and finds it generally distinguishes malignant from benign nodules like experts but shows dangerous sensitivity to unjustified artifacts and radial bias.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Sybil lung cancer risk model differentiates malignant nodules like radiologists but shows sensitivity to artifacts and radial bias.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"ea8912ea83eb9ccc62a64e41d7ec8a56073f47d1d57e27f4af2c340dfee71aee"},"source":{"id":"2602.02560","kind":"arxiv","version":2},"verdict":{"id":"7c6442eb-e1ba-4eed-8fc7-4dc2384c9307","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T09:21:52.223195Z","strongest_claim":"Providing the first interventional audit of Sybil, we demonstrate that while the model often exhibits behavior akin to an expert radiologist, differentiating malignant pulmonary nodules from benign ones, it suffers from critical failure modes, including dangerous sensitivity to clinically unjustified artifacts and a distinct radial bias.","one_line_summary":"S(H)NAP audits Sybil via generative interventions and finds it generally distinguishes malignant from benign nodules like experts but shows dangerous sensitivity to unjustified artifacts and radial bias.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the 3D diffusion bridge produces modifications that isolate true causal contributions without introducing new confounding artifacts that the model could latch onto.","pith_extraction_headline":"Sybil lung cancer risk model differentiates malignant nodules like radiologists but shows sensitivity to artifacts and radial bias."},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"f7a9b535252d44f270a3867fccd9abe5090d64fa8c98807db41e9241ed07b971"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2602.02560","created_at":"2026-05-18T02:45:05.587680+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.02560v2","created_at":"2026-05-18T02:45:05.587680+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.02560","created_at":"2026-05-18T02:45:05.587680+00:00"},{"alias_kind":"pith_short_12","alias_value":"GBDOXIASUMMF","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"GBDOXIASUMMFZCXA","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"GBDOXIAS","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GBDOXIASUMMFZCXAJB5JAIW7CA","json":"https://pith.science/pith/GBDOXIASUMMFZCXAJB5JAIW7CA.json","graph_json":"https://pith.science/api/pith-number/GBDOXIASUMMFZCXAJB5JAIW7CA/graph.json","events_json":"https://pith.science/api/pith-number/GBDOXIASUMMFZCXAJB5JAIW7CA/events.json","paper":"https://pith.science/paper/GBDOXIAS"},"agent_actions":{"view_html":"https://pith.science/pith/GBDOXIASUMMFZCXAJB5JAIW7CA","download_json":"https://pith.science/pith/GBDOXIASUMMFZCXAJB5JAIW7CA.json","view_paper":"https://pith.science/paper/GBDOXIAS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.02560&json=true","fetch_graph":"https://pith.science/api/pith-number/GBDOXIASUMMFZCXAJB5JAIW7CA/graph.json","fetch_events":"https://pith.science/api/pith-number/GBDOXIASUMMFZCXAJB5JAIW7CA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GBDOXIASUMMFZCXAJB5JAIW7CA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GBDOXIASUMMFZCXAJB5JAIW7CA/action/storage_attestation","attest_author":"https://pith.science/pith/GBDOXIASUMMFZCXAJB5JAIW7CA/action/author_attestation","sign_citation":"https://pith.science/pith/GBDOXIASUMMFZCXAJB5JAIW7CA/action/citation_signature","submit_replication":"https://pith.science/pith/GBDOXIASUMMFZCXAJB5JAIW7CA/action/replication_record"}},"created_at":"2026-05-18T02:45:05.587680+00:00","updated_at":"2026-05-18T02:45:05.587680+00:00"}