{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:VQ43UB3RG4BKYMZRD773M6TI6O","short_pith_number":"pith:VQ43UB3R","schema_version":"1.0","canonical_sha256":"ac39ba07713702ac33311fffb67a68f3b11af11f8cb1c5faf6d6dce0c5a237b7","source":{"kind":"arxiv","id":"2601.15161","version":2},"attestation_state":"computed","paper":{"title":"Automated Rubrics for Reliable Evaluation of Medical Dialogue Systems","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A retrieval-augmented multi-agent system automatically generates instance-specific rubrics that ground medical dialogue evaluation in verifiable clinical facts.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Abdine Maiga, Emine Yilmaz, Hossein A. Rahmani, Yinzhu Chen","submitted_at":"2026-01-21T16:40:41Z","abstract_excerpt":"Large Language Models (LLMs) are increasingly used for clinical decision support, where hallucinations and unsafe suggestions may pose direct risks to patient safety. These risks are hard to assess: subtle clinical errors are often missed by generic metrics and LLM judges using general criteria, while expert-authored fine-grained rubrics are expensive and difficult to scale. In this paper, we propose a retrieval-augmented multi-agent framework designed to automate the generation of instance-specific evaluation rubrics.\n  Our approach grounds evaluation in authoritative medical evidence by deco"},"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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2601.15161","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-01-21T16:40:41Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"71ca9f7437f9cd29e19f2b77486d1eea22c0b7d054bc77417e736b2b2cba09dc","abstract_canon_sha256":"e33b3174929e4238e0e315c59977f1cfa46ec8ec91de621c06cfb178827ae9f4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:45:06.028457Z","signature_b64":"wKjpfwr6y6m8nEZ/jIIa1uHXW8C/RFz9Si4I0NTsJuamZLlkz9kopwB0uiumRy7nVTYSjTdHPuEM9t/W9FX/DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ac39ba07713702ac33311fffb67a68f3b11af11f8cb1c5faf6d6dce0c5a237b7","last_reissued_at":"2026-05-18T02:45:06.027939Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:45:06.027939Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Automated Rubrics for Reliable Evaluation of Medical Dialogue Systems","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A retrieval-augmented multi-agent system automatically generates instance-specific rubrics that ground medical dialogue evaluation in verifiable clinical facts.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Abdine Maiga, Emine Yilmaz, Hossein A. Rahmani, Yinzhu Chen","submitted_at":"2026-01-21T16:40:41Z","abstract_excerpt":"Large Language Models (LLMs) are increasingly used for clinical decision support, where hallucinations and unsafe suggestions may pose direct risks to patient safety. These risks are hard to assess: subtle clinical errors are often missed by generic metrics and LLM judges using general criteria, while expert-authored fine-grained rubrics are expensive and difficult to scale. In this paper, we propose a retrieval-augmented multi-agent framework designed to automate the generation of instance-specific evaluation rubrics.\n  Our approach grounds evaluation in authoritative medical evidence by deco"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"our framework achieves Clinical Intent Alignment (CIA) scores of 50.20% and 31.90%, significantly outperforming the GPT-4o baseline and demonstrating robust cross-lingual generalization. In discriminative tests on HealthBench, our rubrics yield a 7.8% higher win rate than GPT-4o baseline with nearly double score Δ.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"that retrieved authoritative medical content can be reliably decomposed into atomic facts and synthesized with interaction constraints to produce verifiable, fine-grained criteria without introducing new errors or hallucinations.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A retrieval-augmented multi-agent system creates evidence-based, fine-grained rubrics for medical LLM evaluation, achieving 50.20% and 31.90% CIA scores on HealthBench and LLMEval-Med while outperforming GPT-4o baselines.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A retrieval-augmented multi-agent system automatically generates instance-specific rubrics that ground medical dialogue evaluation in verifiable clinical facts.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"5b6c8309cb188ddb9838e956216ccdaf78527ea825b65c71e59208e343f9efbd"},"source":{"id":"2601.15161","kind":"arxiv","version":2},"verdict":{"id":"87ebfcba-e330-4510-9282-46efac87bbf5","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T12:12:54.298066Z","strongest_claim":"our framework achieves Clinical Intent Alignment (CIA) scores of 50.20% and 31.90%, significantly outperforming the GPT-4o baseline and demonstrating robust cross-lingual generalization. In discriminative tests on HealthBench, our rubrics yield a 7.8% higher win rate than GPT-4o baseline with nearly double score Δ.","one_line_summary":"A retrieval-augmented multi-agent system creates evidence-based, fine-grained rubrics for medical LLM evaluation, achieving 50.20% and 31.90% CIA scores on HealthBench and LLMEval-Med while outperforming GPT-4o baselines.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"that retrieved authoritative medical content can be reliably decomposed into atomic facts and synthesized with interaction constraints to produce verifiable, fine-grained criteria without introducing new errors or hallucinations.","pith_extraction_headline":"A retrieval-augmented multi-agent system automatically generates instance-specific rubrics that ground medical dialogue evaluation in verifiable clinical facts."},"references":{"count":27,"sample":[{"doi":"","year":null,"title":"Guidelines: CDC (site:cdc.gov), WHO (site:who.int), NICE (site:nice.org.uk), Merck Manuals (site:merckmanuals.com)","work_id":"d6729640-33b1-42fb-9d1e-0a13d7e3df75","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Drugs: Drugs.com (site:drugs.com), BNF (site:bnf.nice.org.uk)","work_id":"decc3106-a4f4-4932-8faa-719c4ef31df9","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Patient Ed: Mayo Clinic (site:mayoclinic.org), Cleveland Clinic (site:clevelandclinic.org), NHS (site:nhs.uk)","work_id":"4743015b-4b41-4480-95cd-4207b0db246c","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Research: PubMed (site:ncbi.nlm.nih.gov) Task:","work_id":"e21bff1f-5723-463c-8c17-f3b86df9fd2f","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"intent”: “string","work_id":"ab47405d-0aa5-4db4-b1a3-019335d06cc3","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":27,"snapshot_sha256":"239225e7e02ef859bcf16fe01ae7c88af2bcb71a02ef7f170e757acbf97d49af","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"e429c75217d7c4f852099acbd85437a03b1238c3eadbcc0ed6ffce3385f80988"},"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":"2601.15161","created_at":"2026-05-18T02:45:06.028015+00:00"},{"alias_kind":"arxiv_version","alias_value":"2601.15161v2","created_at":"2026-05-18T02:45:06.028015+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.15161","created_at":"2026-05-18T02:45:06.028015+00:00"},{"alias_kind":"pith_short_12","alias_value":"VQ43UB3RG4BK","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"VQ43UB3RG4BKYMZR","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"VQ43UB3R","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/VQ43UB3RG4BKYMZRD773M6TI6O","json":"https://pith.science/pith/VQ43UB3RG4BKYMZRD773M6TI6O.json","graph_json":"https://pith.science/api/pith-number/VQ43UB3RG4BKYMZRD773M6TI6O/graph.json","events_json":"https://pith.science/api/pith-number/VQ43UB3RG4BKYMZRD773M6TI6O/events.json","paper":"https://pith.science/paper/VQ43UB3R"},"agent_actions":{"view_html":"https://pith.science/pith/VQ43UB3RG4BKYMZRD773M6TI6O","download_json":"https://pith.science/pith/VQ43UB3RG4BKYMZRD773M6TI6O.json","view_paper":"https://pith.science/paper/VQ43UB3R","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2601.15161&json=true","fetch_graph":"https://pith.science/api/pith-number/VQ43UB3RG4BKYMZRD773M6TI6O/graph.json","fetch_events":"https://pith.science/api/pith-number/VQ43UB3RG4BKYMZRD773M6TI6O/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VQ43UB3RG4BKYMZRD773M6TI6O/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VQ43UB3RG4BKYMZRD773M6TI6O/action/storage_attestation","attest_author":"https://pith.science/pith/VQ43UB3RG4BKYMZRD773M6TI6O/action/author_attestation","sign_citation":"https://pith.science/pith/VQ43UB3RG4BKYMZRD773M6TI6O/action/citation_signature","submit_replication":"https://pith.science/pith/VQ43UB3RG4BKYMZRD773M6TI6O/action/replication_record"}},"created_at":"2026-05-18T02:45:06.028015+00:00","updated_at":"2026-05-18T02:45:06.028015+00:00"}