{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:4RPZQURXRAPT4JF5PGTMQMBRR6","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":"f914fa4f22247427031bbe0098db6018905160aec9f8f4e84ef38bcb5aa42064","cross_cats_sorted":["cs.CV","cs.SI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-12-02T16:39:32Z","title_canon_sha256":"48e20ea2869f79b6da0e57ced18dd97fdd49c5eb62627d377ba898ef8245f765"},"schema_version":"1.0","source":{"id":"2512.02920","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2512.02920","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"arxiv_version","alias_value":"2512.02920v3","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.02920","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"pith_short_12","alias_value":"4RPZQURXRAPT","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"4RPZQURXRAPT4JF5","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"4RPZQURX","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:15364f8ffb15cbca5e9be421675c1e0925a8ea35754bc53800899870bdf2e5c9","target":"graph","created_at":"2026-05-17T23:39:16Z","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":"integrating both data modalities improves prediction accuracy, achieving an average AUROC of 90.1%, a 3.7% gain over graph neural network models that use only graph structures. With the improved embeddings, we conduct a causal analysis using a matching estimator to identify the key factors influencing traffic accidents. We find that accident rates rise by 24% under higher precipitation, by 22% on higher-speed roads such as motorways, and by 29% due to seasonal patterns, after adjusting for other confounding factors."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"Satellite imagery supplies predictive information about road surface and surroundings that is not already captured by the provided weather statistics, road type labels, and traffic volume features; the matching estimator fully balances all relevant confounders between high- and low-precipitation locations."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Multimodal embeddings from satellite images and road graphs raise accident prediction AUROC to 90.1 percent and attribute 24 percent higher rates to increased precipitation after confounder adjustment."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Combining satellite images with road network graphs predicts traffic accidents at 90.1% AUROC and identifies causal factors."}],"snapshot_sha256":"cdecd87adc1e89266e843eef60112b467164ba355a0aefece54b7bab33356fa3"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"ffa341249c20b48b2d0a2366543d47fadb714361ea72f4808295c070881152f4"},"paper":{"abstract_excerpt":"We consider analyzing traffic accident patterns using both road network data and satellite images aligned to road graph nodes. Previous work for predicting accident occurrences relies primarily on road network structural features while overlooking physical and environmental information from the road surface and its surroundings. In this work, we construct a large multimodal dataset spanning six U.S. states, containing nine million traffic accident records from official sources, and one million high-resolution satellite images for each node of the road network. Additionally, every node is annot","authors_text":"Haris N. Koutsopoulos, Hongyang R. Zhang, Minxuan Duan, Ziniu Zhang","cross_cats":["cs.CV","cs.SI"],"headline":"Combining satellite images with road network graphs predicts traffic accidents at 90.1% AUROC and identifies causal factors.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-12-02T16:39:32Z","title":"Learning Multimodal Embeddings for Traffic Accident Prediction and Causal Estimation"},"references":{"count":45,"internal_anchors":0,"resolved_work":45,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Roadtracer: Automatic extraction of road networks from aerial images","work_id":"9551f7fa-bb11-4229-b506-566dc5dccd13","year":2018},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"L. J. Blincoe, T. R. Miller, E. Zaloshnja, and B. Lawrence.The economic and societal impact of motor vehicle crashes, 2010 (Revised). Tech. rep. United States. Department of Transportation. National H","work_id":"4b1e5910-3bbf-485b-a01b-846703d0fe97","year":2010},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"TEMPO: Prompt- based Generative Pre-trained Transformer for Time Series Forecasting","work_id":"5cb483dc-e6b9-42ac-8ffb-6f8abd7a9dab","year":2024},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"FT-AED: Benchmark dataset for early freeway traffic anomalous event detection","work_id":"542c3376-3dee-4084-a0e3-e41745234509","year":2024},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Multimodal learning with graphs","work_id":"f87a8e23-167f-4f9c-a9e6-508ee73f8363","year":2023}],"snapshot_sha256":"35d8ecdbb4fb33dddff9e3417089473cce63f7a400e176535e6c84ee20eed379"},"source":{"id":"2512.02920","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-17T02:15:54.609744Z","id":"4140a9e4-ed74-47b9-8f2a-722891a06036","model_set":{"reader":"grok-4.3"},"one_line_summary":"Multimodal embeddings from satellite images and road graphs raise accident prediction AUROC to 90.1 percent and attribute 24 percent higher rates to increased precipitation after confounder adjustment.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Combining satellite images with road network graphs predicts traffic accidents at 90.1% AUROC and identifies causal factors.","strongest_claim":"integrating both data modalities improves prediction accuracy, achieving an average AUROC of 90.1%, a 3.7% gain over graph neural network models that use only graph structures. With the improved embeddings, we conduct a causal analysis using a matching estimator to identify the key factors influencing traffic accidents. We find that accident rates rise by 24% under higher precipitation, by 22% on higher-speed roads such as motorways, and by 29% due to seasonal patterns, after adjusting for other confounding factors.","weakest_assumption":"Satellite imagery supplies predictive information about road surface and surroundings that is not already captured by the provided weather statistics, road type labels, and traffic volume features; the matching estimator fully balances all relevant confounders between high- and low-precipitation locations."}},"verdict_id":"4140a9e4-ed74-47b9-8f2a-722891a06036"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c76ffc1dc2d1c93652ee73f0bb39e83fc77da9ef47467c7bc5b59bb91eab2532","target":"record","created_at":"2026-05-17T23:39:16Z","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":"f914fa4f22247427031bbe0098db6018905160aec9f8f4e84ef38bcb5aa42064","cross_cats_sorted":["cs.CV","cs.SI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-12-02T16:39:32Z","title_canon_sha256":"48e20ea2869f79b6da0e57ced18dd97fdd49c5eb62627d377ba898ef8245f765"},"schema_version":"1.0","source":{"id":"2512.02920","kind":"arxiv","version":3}},"canonical_sha256":"e45f985237881f3e24bd79a6c830318f9c61270ccb646606ef1065938bc8adce","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e45f985237881f3e24bd79a6c830318f9c61270ccb646606ef1065938bc8adce","first_computed_at":"2026-05-17T23:39:16.941121Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:16.941121Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"PriZ0JBsLqopXy2yhY7NqI9Nuw3CydVPdJboFEmfkVdGGA9ZUijgH3pfr75NQpFKT43CKmu9630k5Fk8BxU4AQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:16.941810Z","signed_message":"canonical_sha256_bytes"},"source_id":"2512.02920","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c76ffc1dc2d1c93652ee73f0bb39e83fc77da9ef47467c7bc5b59bb91eab2532","sha256:15364f8ffb15cbca5e9be421675c1e0925a8ea35754bc53800899870bdf2e5c9"],"state_sha256":"7d01031bfa39bb6ec781aaa49d3888b56969dc18a1009f5ee3e02455476dc4bb"}