{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:EE2VZFG7OP62HUQZJRHHKAVLZ4","short_pith_number":"pith:EE2VZFG7","canonical_record":{"source":{"id":"2605.13081","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T06:54:21Z","cross_cats_sorted":[],"title_canon_sha256":"c3966583a74b1dffb38fc2ddb936c7d4973a17538cfda3e5c5913d43704ccc02","abstract_canon_sha256":"10fadce2751fcb9d2f82987b202d232bf7a2e5890e2b37ca55ae0e15b0174e9b"},"schema_version":"1.0"},"canonical_sha256":"21355c94df73fda3d2194c4e7502abcf00f2d71d4a87590b29918b92ca0f4e98","source":{"kind":"arxiv","id":"2605.13081","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.13081","created_at":"2026-05-18T03:08:58Z"},{"alias_kind":"arxiv_version","alias_value":"2605.13081v1","created_at":"2026-05-18T03:08:58Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13081","created_at":"2026-05-18T03:08:58Z"},{"alias_kind":"pith_short_12","alias_value":"EE2VZFG7OP62","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"EE2VZFG7OP62HUQZ","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"EE2VZFG7","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:EE2VZFG7OP62HUQZJRHHKAVLZ4","target":"record","payload":{"canonical_record":{"source":{"id":"2605.13081","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T06:54:21Z","cross_cats_sorted":[],"title_canon_sha256":"c3966583a74b1dffb38fc2ddb936c7d4973a17538cfda3e5c5913d43704ccc02","abstract_canon_sha256":"10fadce2751fcb9d2f82987b202d232bf7a2e5890e2b37ca55ae0e15b0174e9b"},"schema_version":"1.0"},"canonical_sha256":"21355c94df73fda3d2194c4e7502abcf00f2d71d4a87590b29918b92ca0f4e98","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:08:58.684853Z","signature_b64":"tAzuTo6ufXiQGHsfVgy+omYZmHvwhvrQBOMWclFINAfi1TQRBs7P0JGdLE0TzDVWZILRt+D1GagrIUODmltvBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"21355c94df73fda3d2194c4e7502abcf00f2d71d4a87590b29918b92ca0f4e98","last_reissued_at":"2026-05-18T03:08:58.684228Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:08:58.684228Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.13081","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T03:08:58Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"khXQ7kgFoa5DivDTZqJ6JBQoQ48e8XLDlDg91MMERR1EYcHb/UQtll6/swtFgElKAiqzVe43myABbxSm4D4cBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T18:06:48.623618Z"},"content_sha256":"37755ee5b5034f76bee618341dc18dd6fd12a59ed12b825d531695eed679503f","schema_version":"1.0","event_id":"sha256:37755ee5b5034f76bee618341dc18dd6fd12a59ed12b825d531695eed679503f"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:EE2VZFG7OP62HUQZJRHHKAVLZ4","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"PRA-PoE: Robust Alzheimer's Diagnosis with Arbitrary Missing Modalities","license":"http://creativecommons.org/licenses/by/4.0/","headline":"PRA-PoE aligns representations across missing modality subsets and fuses uncertain Gaussian experts to improve Alzheimer's diagnosis on incomplete data.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Guangqian Yang, Qian Niu, Shujun Wang, Wenlong Hou, Ye Du","submitted_at":"2026-05-13T06:54:21Z","abstract_excerpt":"Missing modalities are prevalent in real-world Alzheimer's disease (AD) assessment and pose a significant challenge to multimodal learning, particularly when the distribution of observed modality subsets differs between training and deployment. Such missingness pattern mismatch induces a conditional representation shift across modality subsets. Existing approaches that rely on implicit imputation or modality synthesis often fail to explicitly model modality availability and uncertainty, leading to overconfident dependence on synthesized features, reduced robustness, and miscalibrated uncertain"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"PRA-PoE consistently outperforms the state-of-the-art across datasets, achieving a 5.4% relative improvement in average accuracy on ADNI and a 10.9% relative gain in average F1 on OASIS-3 over the strongest baseline across all non-empty modality subsets.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That learnable global prototypes plus availability-conditioned tokens can reliably encode modality presence, re-synthesize missing features, and align latent spaces across subsets without introducing new biases or overfitting to the specific missingness patterns in the training data.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"PRA-PoE combines learnable prototypes for representation alignment across modality subsets with uncertainty-weighted product-of-experts fusion to improve robustness and calibration, reporting 5.4% accuracy gain on ADNI and 10.9% F1 gain on OASIS-3 over baselines across all non-empty modality subsets","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"PRA-PoE aligns representations across missing modality subsets and fuses uncertain Gaussian experts to improve Alzheimer's diagnosis on incomplete data.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"69c488f01c7d14257a7d8b7754ad98648bfffc92b31a3ae632a4c1c4251ad2da"},"source":{"id":"2605.13081","kind":"arxiv","version":1},"verdict":{"id":"f483e9e2-595f-42a5-a5f0-4c6c39ef95fc","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T20:04:14.109747Z","strongest_claim":"PRA-PoE consistently outperforms the state-of-the-art across datasets, achieving a 5.4% relative improvement in average accuracy on ADNI and a 10.9% relative gain in average F1 on OASIS-3 over the strongest baseline across all non-empty modality subsets.","one_line_summary":"PRA-PoE combines learnable prototypes for representation alignment across modality subsets with uncertainty-weighted product-of-experts fusion to improve robustness and calibration, reporting 5.4% accuracy gain on ADNI and 10.9% F1 gain on OASIS-3 over baselines across all non-empty modality subsets","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That learnable global prototypes plus availability-conditioned tokens can reliably encode modality presence, re-synthesize missing features, and align latent spaces across subsets without introducing new biases or overfitting to the specific missingness patterns in the training data.","pith_extraction_headline":"PRA-PoE aligns representations across missing modality subsets and fuses uncertain Gaussian experts to improve Alzheimer's diagnosis on incomplete data."},"references":{"count":26,"sample":[{"doi":"","year":2024,"title":"IEEE Open Journal of Engineering in Medicine and Biology6, 183–192 (2024)","work_id":"90a758e3-49f4-439e-baa0-9959168a81bc","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Alzheimer’s & Dementia18(4), 561–571 (2022)","work_id":"6e944187-6369-4c30-9093-e06aba2b82f3","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"IEEE Transactions on Medical Imaging42(12), 3566–3578 (2023)","work_id":"ebe5868c-23f3-4e9c-831a-c9f6a74be443","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"In: International Con- ference on Medical Image Computing and Computer-Assisted Intervention","work_id":"b59e6fff-1128-473e-81b5-78c22bad3456","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Advances in Neural Information Processing Sys- tems37, 67850–67900 (2025)","work_id":"ec4f771d-f087-419d-ae9f-7f749b7ca907","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":26,"snapshot_sha256":"46f93f5610fb11916dd9585edc228bd58d9e951896e4fc11a53b35ed86a5312a","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":"f483e9e2-595f-42a5-a5f0-4c6c39ef95fc"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T03:08:58Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"utrqKlf0lho7muyWt1/MuDP0Q18qlDZf7ufidxomK5VYmlDfAqk2Sh4PH59RQMWGEm7owSIQ4jdxN3/eJKi9CQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T18:06:48.624307Z"},"content_sha256":"f14a0e40667c91e0d7ef4167480076ab2f139b3fb831108f2d6e13b6501e342c","schema_version":"1.0","event_id":"sha256:f14a0e40667c91e0d7ef4167480076ab2f139b3fb831108f2d6e13b6501e342c"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/EE2VZFG7OP62HUQZJRHHKAVLZ4/bundle.json","state_url":"https://pith.science/pith/EE2VZFG7OP62HUQZJRHHKAVLZ4/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/EE2VZFG7OP62HUQZJRHHKAVLZ4/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":"2026-05-27T18:06:48Z","links":{"resolver":"https://pith.science/pith/EE2VZFG7OP62HUQZJRHHKAVLZ4","bundle":"https://pith.science/pith/EE2VZFG7OP62HUQZJRHHKAVLZ4/bundle.json","state":"https://pith.science/pith/EE2VZFG7OP62HUQZJRHHKAVLZ4/state.json","well_known_bundle":"https://pith.science/.well-known/pith/EE2VZFG7OP62HUQZJRHHKAVLZ4/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:EE2VZFG7OP62HUQZJRHHKAVLZ4","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":"10fadce2751fcb9d2f82987b202d232bf7a2e5890e2b37ca55ae0e15b0174e9b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T06:54:21Z","title_canon_sha256":"c3966583a74b1dffb38fc2ddb936c7d4973a17538cfda3e5c5913d43704ccc02"},"schema_version":"1.0","source":{"id":"2605.13081","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.13081","created_at":"2026-05-18T03:08:58Z"},{"alias_kind":"arxiv_version","alias_value":"2605.13081v1","created_at":"2026-05-18T03:08:58Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13081","created_at":"2026-05-18T03:08:58Z"},{"alias_kind":"pith_short_12","alias_value":"EE2VZFG7OP62","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"EE2VZFG7OP62HUQZ","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"EE2VZFG7","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:f14a0e40667c91e0d7ef4167480076ab2f139b3fb831108f2d6e13b6501e342c","target":"graph","created_at":"2026-05-18T03:08:58Z","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":"PRA-PoE consistently outperforms the state-of-the-art across datasets, achieving a 5.4% relative improvement in average accuracy on ADNI and a 10.9% relative gain in average F1 on OASIS-3 over the strongest baseline across all non-empty modality subsets."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That learnable global prototypes plus availability-conditioned tokens can reliably encode modality presence, re-synthesize missing features, and align latent spaces across subsets without introducing new biases or overfitting to the specific missingness patterns in the training data."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"PRA-PoE combines learnable prototypes for representation alignment across modality subsets with uncertainty-weighted product-of-experts fusion to improve robustness and calibration, reporting 5.4% accuracy gain on ADNI and 10.9% F1 gain on OASIS-3 over baselines across all non-empty modality subsets"},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"PRA-PoE aligns representations across missing modality subsets and fuses uncertain Gaussian experts to improve Alzheimer's diagnosis on incomplete data."}],"snapshot_sha256":"69c488f01c7d14257a7d8b7754ad98648bfffc92b31a3ae632a4c1c4251ad2da"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Missing modalities are prevalent in real-world Alzheimer's disease (AD) assessment and pose a significant challenge to multimodal learning, particularly when the distribution of observed modality subsets differs between training and deployment. Such missingness pattern mismatch induces a conditional representation shift across modality subsets. Existing approaches that rely on implicit imputation or modality synthesis often fail to explicitly model modality availability and uncertainty, leading to overconfident dependence on synthesized features, reduced robustness, and miscalibrated uncertain","authors_text":"Guangqian Yang, Qian Niu, Shujun Wang, Wenlong Hou, Ye Du","cross_cats":[],"headline":"PRA-PoE aligns representations across missing modality subsets and fuses uncertain Gaussian experts to improve Alzheimer's diagnosis on incomplete data.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T06:54:21Z","title":"PRA-PoE: Robust Alzheimer's Diagnosis with Arbitrary Missing Modalities"},"references":{"count":26,"internal_anchors":0,"resolved_work":26,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"IEEE Open Journal of Engineering in Medicine and Biology6, 183–192 (2024)","work_id":"90a758e3-49f4-439e-baa0-9959168a81bc","year":2024},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Alzheimer’s & Dementia18(4), 561–571 (2022)","work_id":"6e944187-6369-4c30-9093-e06aba2b82f3","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"IEEE Transactions on Medical Imaging42(12), 3566–3578 (2023)","work_id":"ebe5868c-23f3-4e9c-831a-c9f6a74be443","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"In: International Con- ference on Medical Image Computing and Computer-Assisted Intervention","work_id":"b59e6fff-1128-473e-81b5-78c22bad3456","year":2024},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Advances in Neural Information Processing Sys- tems37, 67850–67900 (2025)","work_id":"ec4f771d-f087-419d-ae9f-7f749b7ca907","year":2025}],"snapshot_sha256":"46f93f5610fb11916dd9585edc228bd58d9e951896e4fc11a53b35ed86a5312a"},"source":{"id":"2605.13081","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T20:04:14.109747Z","id":"f483e9e2-595f-42a5-a5f0-4c6c39ef95fc","model_set":{"reader":"grok-4.3"},"one_line_summary":"PRA-PoE combines learnable prototypes for representation alignment across modality subsets with uncertainty-weighted product-of-experts fusion to improve robustness and calibration, reporting 5.4% accuracy gain on ADNI and 10.9% F1 gain on OASIS-3 over baselines across all non-empty modality subsets","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"PRA-PoE aligns representations across missing modality subsets and fuses uncertain Gaussian experts to improve Alzheimer's diagnosis on incomplete data.","strongest_claim":"PRA-PoE consistently outperforms the state-of-the-art across datasets, achieving a 5.4% relative improvement in average accuracy on ADNI and a 10.9% relative gain in average F1 on OASIS-3 over the strongest baseline across all non-empty modality subsets.","weakest_assumption":"That learnable global prototypes plus availability-conditioned tokens can reliably encode modality presence, re-synthesize missing features, and align latent spaces across subsets without introducing new biases or overfitting to the specific missingness patterns in the training data."}},"verdict_id":"f483e9e2-595f-42a5-a5f0-4c6c39ef95fc"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:37755ee5b5034f76bee618341dc18dd6fd12a59ed12b825d531695eed679503f","target":"record","created_at":"2026-05-18T03:08:58Z","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":"10fadce2751fcb9d2f82987b202d232bf7a2e5890e2b37ca55ae0e15b0174e9b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T06:54:21Z","title_canon_sha256":"c3966583a74b1dffb38fc2ddb936c7d4973a17538cfda3e5c5913d43704ccc02"},"schema_version":"1.0","source":{"id":"2605.13081","kind":"arxiv","version":1}},"canonical_sha256":"21355c94df73fda3d2194c4e7502abcf00f2d71d4a87590b29918b92ca0f4e98","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"21355c94df73fda3d2194c4e7502abcf00f2d71d4a87590b29918b92ca0f4e98","first_computed_at":"2026-05-18T03:08:58.684228Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:08:58.684228Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"tAzuTo6ufXiQGHsfVgy+omYZmHvwhvrQBOMWclFINAfi1TQRBs7P0JGdLE0TzDVWZILRt+D1GagrIUODmltvBA==","signature_status":"signed_v1","signed_at":"2026-05-18T03:08:58.684853Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.13081","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:37755ee5b5034f76bee618341dc18dd6fd12a59ed12b825d531695eed679503f","sha256:f14a0e40667c91e0d7ef4167480076ab2f139b3fb831108f2d6e13b6501e342c"],"state_sha256":"0ea9ac0b8a9f9678a46c2383e13dfef61cba721725e1df99154863b6c3f91924"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"v8ffsGmTYiI/LUEtu3ZwA+Up3/lV2odLebTPnYojxdFku24D/pQMJwfImRboghAbheBuYOnsq8U1Da//zKo3Dw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-27T18:06:48.627337Z","bundle_sha256":"8b435573ae844375d7670688c2124aa88eb30ff87c8a463a6110c6043dc03994"}}