{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:Q7QSPR2VKXRRWWBDMQACH3NPHZ","short_pith_number":"pith:Q7QSPR2V","canonical_record":{"source":{"id":"2603.20698","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-03-21T07:47:37Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"e1d9a9ae168014743478785384a7601d1694dc45c7e8e871b76fc7091d4473c0","abstract_canon_sha256":"95c755db2c86826bd0ad61aa11ab9b6ed27911fc21af613131e3a45d48d433db"},"schema_version":"1.0"},"canonical_sha256":"87e127c75555e31b5823640023edaf3e774e87a48f586232011513bf4800dcc8","source":{"kind":"arxiv","id":"2603.20698","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2603.20698","created_at":"2026-07-07T01:16:10Z"},{"alias_kind":"arxiv_version","alias_value":"2603.20698v3","created_at":"2026-07-07T01:16:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.20698","created_at":"2026-07-07T01:16:10Z"},{"alias_kind":"pith_short_12","alias_value":"Q7QSPR2VKXRR","created_at":"2026-07-07T01:16:10Z"},{"alias_kind":"pith_short_16","alias_value":"Q7QSPR2VKXRRWWBD","created_at":"2026-07-07T01:16:10Z"},{"alias_kind":"pith_short_8","alias_value":"Q7QSPR2V","created_at":"2026-07-07T01:16:10Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:Q7QSPR2VKXRRWWBDMQACH3NPHZ","target":"record","payload":{"canonical_record":{"source":{"id":"2603.20698","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-03-21T07:47:37Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"e1d9a9ae168014743478785384a7601d1694dc45c7e8e871b76fc7091d4473c0","abstract_canon_sha256":"95c755db2c86826bd0ad61aa11ab9b6ed27911fc21af613131e3a45d48d433db"},"schema_version":"1.0"},"canonical_sha256":"87e127c75555e31b5823640023edaf3e774e87a48f586232011513bf4800dcc8","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T01:16:10.111464Z","signature_b64":"aA2jl/atn72qaJtKJYx1L9aTIcsoSKU8ZsXyoxeuElq3gC5QfHfV1pe/jzK85lnnQpxnwCJiQ5O18M6lmNk1Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"87e127c75555e31b5823640023edaf3e774e87a48f586232011513bf4800dcc8","last_reissued_at":"2026-07-07T01:16:10.105887Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T01:16:10.105887Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2603.20698","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-07T01:16:10Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"y0u1fIE577sOkYcjqP+stT50JE1m9Z2qhy94D9CSC39Zi+p8jH6OfFGNppHOEjjXWUKfXlMXyWT8XZDJbofYCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T15:10:13.619268Z"},"content_sha256":"2889971a954f286f43da002e1af6f048e76b1bc4080f0f962a13c1f083b1a26e","schema_version":"1.0","event_id":"sha256:2889971a954f286f43da002e1af6f048e76b1bc4080f0f962a13c1f083b1a26e"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:Q7QSPR2VKXRRWWBDMQACH3NPHZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Clinical Cognition Alignment for Gastrointestinal Diagnosis with Multimodal LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"The CogAlign framework aligns multimodal LLMs to hierarchical clinical cognition and enforces causal lesion features via counterfactual reinforcement learning for gastrointestinal endoscopy diagnosis.","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Dubing Chen, Hongbo Lu, Huan Zheng, Jianbing Shen, Pai Peng, Tao He, Tianyi Yan, Wenlong Liao, Yucheng Zhou","submitted_at":"2026-03-21T07:47:37Z","abstract_excerpt":"Multimodal Large Language Models (MLLMs) have demonstrated remarkable potential in medical image analysis. However, their application in gastrointestinal endoscopy is currently hindered by two critical limitations: the misalignment between general model reasoning and standardized clinical cognitive pathways, and the lack of causal association between visual features and diagnostic outcomes. In this paper, we propose a novel Clinical-Cognitive-Aligned (CogAlign) framework to address these challenges. First, we endow the model with rigorous clinical analytical capabilities by constructing the hi"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Extensive experiments demonstrate that our approach achieves State-of-the-Art (SoTA) performance across multiple benchmarks, significantly enhancing diagnostic accuracy in complex clinical scenarios.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That lesion-masked counterfactual samples plus clinical-cognition-centric rewards will strictly constrain the model to causal lesion features without introducing new biases or failing to generalize beyond the training distribution.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"CogAlign uses hierarchical supervised fine-tuning on clinical cognition data plus counterfactual RL to align MLLMs with expert diagnostic pathways and enforce causal lesion grounding for GI endoscopy diagnosis.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"The CogAlign framework aligns multimodal LLMs to hierarchical clinical cognition and enforces causal lesion features via counterfactual reinforcement learning for gastrointestinal endoscopy diagnosis.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"24fed3333a47fe749188d2b7fd641652e4c7191b0627c2625d42b8f8f8ccdee8"},"source":{"id":"2603.20698","kind":"arxiv","version":3},"verdict":{"id":"1603b762-c4cf-45b0-9910-d0f3aa3f865a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T06:31:40.783597Z","strongest_claim":"Extensive experiments demonstrate that our approach achieves State-of-the-Art (SoTA) performance across multiple benchmarks, significantly enhancing diagnostic accuracy in complex clinical scenarios.","one_line_summary":"CogAlign uses hierarchical supervised fine-tuning on clinical cognition data plus counterfactual RL to align MLLMs with expert diagnostic pathways and enforce causal lesion grounding for GI endoscopy diagnosis.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That lesion-masked counterfactual samples plus clinical-cognition-centric rewards will strictly constrain the model to causal lesion features without introducing new biases or failing to generalize beyond the training distribution.","pith_extraction_headline":"The CogAlign framework aligns multimodal LLMs to hierarchical clinical cognition and enforces causal lesion features via counterfactual reinforcement learning for gastrointestinal endoscopy diagnosis."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2603.20698/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","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"},"verdict_id":"1603b762-c4cf-45b0-9910-d0f3aa3f865a"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-07T01:16:10Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0NbgWqaPKXf6IThnbKcSMmsMNcTjz1g1/6u06LBI8e64fiF9AQCpL6bAh8Gtj7eH1ptuiQeI5u/VULsStLo/BQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T15:10:13.619737Z"},"content_sha256":"6d25f8a0e5d584d24393f06626cd1d0742947ccb5fb5118d86cc3c9cad386588","schema_version":"1.0","event_id":"sha256:6d25f8a0e5d584d24393f06626cd1d0742947ccb5fb5118d86cc3c9cad386588"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/Q7QSPR2VKXRRWWBDMQACH3NPHZ/bundle.json","state_url":"https://pith.science/pith/Q7QSPR2VKXRRWWBDMQACH3NPHZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/Q7QSPR2VKXRRWWBDMQACH3NPHZ/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-07-07T15:10:13Z","links":{"resolver":"https://pith.science/pith/Q7QSPR2VKXRRWWBDMQACH3NPHZ","bundle":"https://pith.science/pith/Q7QSPR2VKXRRWWBDMQACH3NPHZ/bundle.json","state":"https://pith.science/pith/Q7QSPR2VKXRRWWBDMQACH3NPHZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/Q7QSPR2VKXRRWWBDMQACH3NPHZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:Q7QSPR2VKXRRWWBDMQACH3NPHZ","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":"95c755db2c86826bd0ad61aa11ab9b6ed27911fc21af613131e3a45d48d433db","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-03-21T07:47:37Z","title_canon_sha256":"e1d9a9ae168014743478785384a7601d1694dc45c7e8e871b76fc7091d4473c0"},"schema_version":"1.0","source":{"id":"2603.20698","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2603.20698","created_at":"2026-07-07T01:16:10Z"},{"alias_kind":"arxiv_version","alias_value":"2603.20698v3","created_at":"2026-07-07T01:16:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.20698","created_at":"2026-07-07T01:16:10Z"},{"alias_kind":"pith_short_12","alias_value":"Q7QSPR2VKXRR","created_at":"2026-07-07T01:16:10Z"},{"alias_kind":"pith_short_16","alias_value":"Q7QSPR2VKXRRWWBD","created_at":"2026-07-07T01:16:10Z"},{"alias_kind":"pith_short_8","alias_value":"Q7QSPR2V","created_at":"2026-07-07T01:16:10Z"}],"graph_snapshots":[{"event_id":"sha256:6d25f8a0e5d584d24393f06626cd1d0742947ccb5fb5118d86cc3c9cad386588","target":"graph","created_at":"2026-07-07T01:16:10Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Extensive experiments demonstrate that our approach achieves State-of-the-Art (SoTA) performance across multiple benchmarks, significantly enhancing diagnostic accuracy in complex clinical scenarios."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That lesion-masked counterfactual samples plus clinical-cognition-centric rewards will strictly constrain the model to causal lesion features without introducing new biases or failing to generalize beyond the training distribution."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"CogAlign uses hierarchical supervised fine-tuning on clinical cognition data plus counterfactual RL to align MLLMs with expert diagnostic pathways and enforce causal lesion grounding for GI endoscopy diagnosis."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"The CogAlign framework aligns multimodal LLMs to hierarchical clinical cognition and enforces causal lesion features via counterfactual reinforcement learning for gastrointestinal endoscopy diagnosis."}],"snapshot_sha256":"24fed3333a47fe749188d2b7fd641652e4c7191b0627c2625d42b8f8f8ccdee8"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2603.20698/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Multimodal Large Language Models (MLLMs) have demonstrated remarkable potential in medical image analysis. However, their application in gastrointestinal endoscopy is currently hindered by two critical limitations: the misalignment between general model reasoning and standardized clinical cognitive pathways, and the lack of causal association between visual features and diagnostic outcomes. In this paper, we propose a novel Clinical-Cognitive-Aligned (CogAlign) framework to address these challenges. First, we endow the model with rigorous clinical analytical capabilities by constructing the hi","authors_text":"Dubing Chen, Hongbo Lu, Huan Zheng, Jianbing Shen, Pai Peng, Tao He, Tianyi Yan, Wenlong Liao, Yucheng Zhou","cross_cats":["cs.CL"],"headline":"The CogAlign framework aligns multimodal LLMs to hierarchical clinical cognition and enforces causal lesion features via counterfactual reinforcement learning for gastrointestinal endoscopy diagnosis.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-03-21T07:47:37Z","title":"Clinical Cognition Alignment for Gastrointestinal Diagnosis with Multimodal LLMs"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.20698","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-15T06:31:40.783597Z","id":"1603b762-c4cf-45b0-9910-d0f3aa3f865a","model_set":{"reader":"grok-4.3"},"one_line_summary":"CogAlign uses hierarchical supervised fine-tuning on clinical cognition data plus counterfactual RL to align MLLMs with expert diagnostic pathways and enforce causal lesion grounding for GI endoscopy diagnosis.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"The CogAlign framework aligns multimodal LLMs to hierarchical clinical cognition and enforces causal lesion features via counterfactual reinforcement learning for gastrointestinal endoscopy diagnosis.","strongest_claim":"Extensive experiments demonstrate that our approach achieves State-of-the-Art (SoTA) performance across multiple benchmarks, significantly enhancing diagnostic accuracy in complex clinical scenarios.","weakest_assumption":"That lesion-masked counterfactual samples plus clinical-cognition-centric rewards will strictly constrain the model to causal lesion features without introducing new biases or failing to generalize beyond the training distribution."}},"verdict_id":"1603b762-c4cf-45b0-9910-d0f3aa3f865a"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:2889971a954f286f43da002e1af6f048e76b1bc4080f0f962a13c1f083b1a26e","target":"record","created_at":"2026-07-07T01:16:10Z","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":"95c755db2c86826bd0ad61aa11ab9b6ed27911fc21af613131e3a45d48d433db","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-03-21T07:47:37Z","title_canon_sha256":"e1d9a9ae168014743478785384a7601d1694dc45c7e8e871b76fc7091d4473c0"},"schema_version":"1.0","source":{"id":"2603.20698","kind":"arxiv","version":3}},"canonical_sha256":"87e127c75555e31b5823640023edaf3e774e87a48f586232011513bf4800dcc8","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"87e127c75555e31b5823640023edaf3e774e87a48f586232011513bf4800dcc8","first_computed_at":"2026-07-07T01:16:10.105887Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-07T01:16:10.105887Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"aA2jl/atn72qaJtKJYx1L9aTIcsoSKU8ZsXyoxeuElq3gC5QfHfV1pe/jzK85lnnQpxnwCJiQ5O18M6lmNk1Cg==","signature_status":"signed_v1","signed_at":"2026-07-07T01:16:10.111464Z","signed_message":"canonical_sha256_bytes"},"source_id":"2603.20698","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:2889971a954f286f43da002e1af6f048e76b1bc4080f0f962a13c1f083b1a26e","sha256:6d25f8a0e5d584d24393f06626cd1d0742947ccb5fb5118d86cc3c9cad386588"],"state_sha256":"2b782042150c68d9366b148314c270ff11945701a65b6edb72178014240f8f8f"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"hzHFibe0f2+J3QyGHyYmCsftITeXL0CGFq7PMn8Xy+EdYBh7bjS6abfGEJENNs+XTlTllDO3ikkT4RGFmRlvBQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-07T15:10:13.621840Z","bundle_sha256":"60d86bb1714589714eaa38aaf108cdcdde6518c851ff4a7e14dde7ccbfc8d0bb"}}