{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:Q6KJXRDWAAR3Q5RJVT2RX4BQ4R","short_pith_number":"pith:Q6KJXRDW","canonical_record":{"source":{"id":"2602.03730","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2026-02-03T16:49:44Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"e755dfb0d3a97edb93ad217775793e95eb0fb5b271d2e8e0774af1710b62ebe1","abstract_canon_sha256":"ea8a8c7e4fe2f8919a03fd8f83949d2b74cfc52bc9412c2b76edaf2f47a11bcb"},"schema_version":"1.0"},"canonical_sha256":"87949bc4760023b87629acf51bf030e442ff8d160aea0a5f716d8636aaa90be2","source":{"kind":"arxiv","id":"2602.03730","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2602.03730","created_at":"2026-05-18T03:10:11Z"},{"alias_kind":"arxiv_version","alias_value":"2602.03730v2","created_at":"2026-05-18T03:10:11Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.03730","created_at":"2026-05-18T03:10:11Z"},{"alias_kind":"pith_short_12","alias_value":"Q6KJXRDWAAR3","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"Q6KJXRDWAAR3Q5RJ","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"Q6KJXRDW","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:Q6KJXRDWAAR3Q5RJVT2RX4BQ4R","target":"record","payload":{"canonical_record":{"source":{"id":"2602.03730","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2026-02-03T16:49:44Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"e755dfb0d3a97edb93ad217775793e95eb0fb5b271d2e8e0774af1710b62ebe1","abstract_canon_sha256":"ea8a8c7e4fe2f8919a03fd8f83949d2b74cfc52bc9412c2b76edaf2f47a11bcb"},"schema_version":"1.0"},"canonical_sha256":"87949bc4760023b87629acf51bf030e442ff8d160aea0a5f716d8636aaa90be2","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:10:11.255499Z","signature_b64":"A3O5Jc70aS0gGveMW41GhgbiebZFsKdStZt8ilr8JTpMv9bxKBIVXRTuI2Vs9XuVGqhI2F6YhRg83nQOqzvhDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"87949bc4760023b87629acf51bf030e442ff8d160aea0a5f716d8636aaa90be2","last_reissued_at":"2026-05-18T03:10:11.254812Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:10:11.254812Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2602.03730","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T03:10:11Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"BxRKNbTxmAReJEOzo0q6vIXTOAPdxoB6R+zQnvO+IWmLZN5g43mC976Ca0m16dmQpGMPh16b0hQ18jFUI0vVCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-29T15:54:21.441739Z"},"content_sha256":"a29b95f964c831c777183b0be86a2b0c029e2bc3b6392e0fe22ac4f470a7e348","schema_version":"1.0","event_id":"sha256:a29b95f964c831c777183b0be86a2b0c029e2bc3b6392e0fe22ac4f470a7e348"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:Q6KJXRDWAAR3Q5RJVT2RX4BQ4R","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Efficient Generative Prediction for EHR Foundation Models: The SCOPE and REACH Estimators","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"SCOPE and REACH estimators enable unbiased clinical outcome prediction from generative EHR models with far fewer tokens than Monte Carlo sampling.","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Bashar Ramadan, Brett K. Beaulieu-Jones, Luke Solo, Matthew B.A. McDermott, Michael C. Burkhart, William F. Parker","submitted_at":"2026-02-03T16:49:44Z","abstract_excerpt":"Generative foundation models trained on tokenized electronic health record (EHR) timelines show promise for clinical outcome prediction via Monte Carlo sampling of simulated future trajectories. However, this approach suffers from three coupled limitations: sparse estimate distributions that poorly differentiate patient risk levels, extreme computational cost, and high sampling variance. We propose two new estimators that leverage next-token probability distributions underutilized by standard Monte Carlo: the Sum of Conditional Outcome Probability Estimator (SCOPE) and Risk Estimation from Ant"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We prove both are unbiased, that REACH guarantees variance reduction over Monte Carlo for any model and outcome, and that REACH is a Rao-Blackwellization of any naive importance sampling scheme that preserves the non-outcome token distribution.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The generative model's next-token probability distributions accurately reflect the underlying data distribution and can be directly leveraged for conditional outcome probability calculations without further approximation or model-specific adjustments.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"SCOPE and REACH are unbiased estimators that deliver Monte Carlo-level accuracy for EHR outcome prediction using 2.5-80x fewer tokens via direct use of conditional probabilities and variance reduction guarantees.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"SCOPE and REACH estimators enable unbiased clinical outcome prediction from generative EHR models with far fewer tokens than Monte Carlo sampling.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"e90aec0e2a89f5dcf7c563e93c228a2255347f33ac2cc724534e815a59263f5f"},"source":{"id":"2602.03730","kind":"arxiv","version":2},"verdict":{"id":"f7652816-1e14-4070-8604-0a45b6f78a9a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T07:18:35.675829Z","strongest_claim":"We prove both are unbiased, that REACH guarantees variance reduction over Monte Carlo for any model and outcome, and that REACH is a Rao-Blackwellization of any naive importance sampling scheme that preserves the non-outcome token distribution.","one_line_summary":"SCOPE and REACH are unbiased estimators that deliver Monte Carlo-level accuracy for EHR outcome prediction using 2.5-80x fewer tokens via direct use of conditional probabilities and variance reduction guarantees.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The generative model's next-token probability distributions accurately reflect the underlying data distribution and can be directly leveraged for conditional outcome probability calculations without further approximation or model-specific adjustments.","pith_extraction_headline":"SCOPE and REACH estimators enable unbiased clinical outcome prediction from generative EHR models with far fewer tokens than Monte Carlo sampling."},"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":"f7652816-1e14-4070-8604-0a45b6f78a9a"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T03:10:11Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"rM8x5uaYiDc6nkoCk8uAZcrWuGZ15m8DGKnKfUAJ34Xun5HjkRw4XE9adbNt7PfjpCnVRxlV+V0+RsjJZdLpCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-29T15:54:21.442612Z"},"content_sha256":"b4d54c47b00d9c4341d822d587e1a004762d440de3e823b903f5e6ca14bede95","schema_version":"1.0","event_id":"sha256:b4d54c47b00d9c4341d822d587e1a004762d440de3e823b903f5e6ca14bede95"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/Q6KJXRDWAAR3Q5RJVT2RX4BQ4R/bundle.json","state_url":"https://pith.science/pith/Q6KJXRDWAAR3Q5RJVT2RX4BQ4R/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/Q6KJXRDWAAR3Q5RJVT2RX4BQ4R/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":"2026-05-29T15:54:21Z","links":{"resolver":"https://pith.science/pith/Q6KJXRDWAAR3Q5RJVT2RX4BQ4R","bundle":"https://pith.science/pith/Q6KJXRDWAAR3Q5RJVT2RX4BQ4R/bundle.json","state":"https://pith.science/pith/Q6KJXRDWAAR3Q5RJVT2RX4BQ4R/state.json","well_known_bundle":"https://pith.science/.well-known/pith/Q6KJXRDWAAR3Q5RJVT2RX4BQ4R/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:Q6KJXRDWAAR3Q5RJVT2RX4BQ4R","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":"ea8a8c7e4fe2f8919a03fd8f83949d2b74cfc52bc9412c2b76edaf2f47a11bcb","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2026-02-03T16:49:44Z","title_canon_sha256":"e755dfb0d3a97edb93ad217775793e95eb0fb5b271d2e8e0774af1710b62ebe1"},"schema_version":"1.0","source":{"id":"2602.03730","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2602.03730","created_at":"2026-05-18T03:10:11Z"},{"alias_kind":"arxiv_version","alias_value":"2602.03730v2","created_at":"2026-05-18T03:10:11Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.03730","created_at":"2026-05-18T03:10:11Z"},{"alias_kind":"pith_short_12","alias_value":"Q6KJXRDWAAR3","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"Q6KJXRDWAAR3Q5RJ","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"Q6KJXRDW","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:b4d54c47b00d9c4341d822d587e1a004762d440de3e823b903f5e6ca14bede95","target":"graph","created_at":"2026-05-18T03:10:11Z","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":"We prove both are unbiased, that REACH guarantees variance reduction over Monte Carlo for any model and outcome, and that REACH is a Rao-Blackwellization of any naive importance sampling scheme that preserves the non-outcome token distribution."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The generative model's next-token probability distributions accurately reflect the underlying data distribution and can be directly leveraged for conditional outcome probability calculations without further approximation or model-specific adjustments."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"SCOPE and REACH are unbiased estimators that deliver Monte Carlo-level accuracy for EHR outcome prediction using 2.5-80x fewer tokens via direct use of conditional probabilities and variance reduction guarantees."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"SCOPE and REACH estimators enable unbiased clinical outcome prediction from generative EHR models with far fewer tokens than Monte Carlo sampling."}],"snapshot_sha256":"e90aec0e2a89f5dcf7c563e93c228a2255347f33ac2cc724534e815a59263f5f"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Generative foundation models trained on tokenized electronic health record (EHR) timelines show promise for clinical outcome prediction via Monte Carlo sampling of simulated future trajectories. However, this approach suffers from three coupled limitations: sparse estimate distributions that poorly differentiate patient risk levels, extreme computational cost, and high sampling variance. We propose two new estimators that leverage next-token probability distributions underutilized by standard Monte Carlo: the Sum of Conditional Outcome Probability Estimator (SCOPE) and Risk Estimation from Ant","authors_text":"Bashar Ramadan, Brett K. Beaulieu-Jones, Luke Solo, Matthew B.A. McDermott, Michael C. Burkhart, William F. Parker","cross_cats":["cs.LG"],"headline":"SCOPE and REACH estimators enable unbiased clinical outcome prediction from generative EHR models with far fewer tokens than Monte Carlo sampling.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2026-02-03T16:49:44Z","title":"Efficient Generative Prediction for EHR Foundation Models: The SCOPE and REACH Estimators"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.03730","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-16T07:18:35.675829Z","id":"f7652816-1e14-4070-8604-0a45b6f78a9a","model_set":{"reader":"grok-4.3"},"one_line_summary":"SCOPE and REACH are unbiased estimators that deliver Monte Carlo-level accuracy for EHR outcome prediction using 2.5-80x fewer tokens via direct use of conditional probabilities and variance reduction guarantees.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"SCOPE and REACH estimators enable unbiased clinical outcome prediction from generative EHR models with far fewer tokens than Monte Carlo sampling.","strongest_claim":"We prove both are unbiased, that REACH guarantees variance reduction over Monte Carlo for any model and outcome, and that REACH is a Rao-Blackwellization of any naive importance sampling scheme that preserves the non-outcome token distribution.","weakest_assumption":"The generative model's next-token probability distributions accurately reflect the underlying data distribution and can be directly leveraged for conditional outcome probability calculations without further approximation or model-specific adjustments."}},"verdict_id":"f7652816-1e14-4070-8604-0a45b6f78a9a"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:a29b95f964c831c777183b0be86a2b0c029e2bc3b6392e0fe22ac4f470a7e348","target":"record","created_at":"2026-05-18T03:10:11Z","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":"ea8a8c7e4fe2f8919a03fd8f83949d2b74cfc52bc9412c2b76edaf2f47a11bcb","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2026-02-03T16:49:44Z","title_canon_sha256":"e755dfb0d3a97edb93ad217775793e95eb0fb5b271d2e8e0774af1710b62ebe1"},"schema_version":"1.0","source":{"id":"2602.03730","kind":"arxiv","version":2}},"canonical_sha256":"87949bc4760023b87629acf51bf030e442ff8d160aea0a5f716d8636aaa90be2","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"87949bc4760023b87629acf51bf030e442ff8d160aea0a5f716d8636aaa90be2","first_computed_at":"2026-05-18T03:10:11.254812Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:10:11.254812Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"A3O5Jc70aS0gGveMW41GhgbiebZFsKdStZt8ilr8JTpMv9bxKBIVXRTuI2Vs9XuVGqhI2F6YhRg83nQOqzvhDA==","signature_status":"signed_v1","signed_at":"2026-05-18T03:10:11.255499Z","signed_message":"canonical_sha256_bytes"},"source_id":"2602.03730","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:a29b95f964c831c777183b0be86a2b0c029e2bc3b6392e0fe22ac4f470a7e348","sha256:b4d54c47b00d9c4341d822d587e1a004762d440de3e823b903f5e6ca14bede95"],"state_sha256":"770ede356098cf699be2efc824aca4d2b13c5834ab8a65ef532a8f62f9d858de"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"voxYEK2yT9wmyJha744QYmHxXBKf4bkqvTXwHfq6FfxBDla/41WNCju0MZk37LziGt8pnTbL+jFaRszR3YmpAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-29T15:54:21.446984Z","bundle_sha256":"1a1e77abf7a69616707efc127e7c7f13b07beb645db65f8f96776a769000d306"}}