{"id":"bf36f430636c50240c91bf35b143c11b7b80cdad5e35284639cb6b1c003126ec","title":"The authors’ VirtualTrialHarness coordinates decisions across virtual-trial workflows","argument":"The authors introduce VirtualTrialHarness, an agent framework for turning virtual-trial protocols into executable analyses across complex clinical schemas. They argue that this takes more than retrieving procedural guidance: teams must decide which data sources to use, how to identify and link patients and encounters, and how to interpret time. VirtualTrialHarness selects reusable procedural skills for each workflow stage and uses explicit contracts to keep study decisions consistent across stages.\n\nThe authors also introduce VTBenchmark to evaluate source selection and cohort construction. Across three complex synthetic environments, required cohort outputs matched references in 95.1% of workflows after requirement clarification, compared with 60.4% using retrieval from the same procedural corpus. The abstract also reports improved source recovery across three language models, increased source overlap with analyst code in 13 Mayo Clinic cases, and reference-matched cohort execution across eight MIMIC-IV tasks after site-specific adaptation. Within the Mayo Clinic Virtual Trials Program, the framework supported investigator review of cohort construction and endpoint analysis. The authors’ central idea is to coordinate procedural guidance with decisions across the workflow, rather than rely on retrieval alone.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The reported workflow-match result is from three complex synthetic environments and is stated to follow requirement clarification. The abstract describes 13 Mayo Clinic cases and eight MIMIC-IV tasks; it does not establish results across a broader range of clinical sites or workflows.","speaker":{"name":"Li, Pengze, Feng, Jingna, Yu, Yue, Wu, Yining, Hu, Xinyue, Lin, Jie, Bhandarkar, Avanti, Tao, Cui","roleAtSource":"Article author"},"source":{"title":"VirtualTrialHarness: Coordinating Agentic Operationalization of Virtual Trial Protocols Across Complex Clinical Schemas","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2611.05551","publishedAt":"2026-11-03","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors introduce VirtualTrialHarness, which selects reusable procedural skills for workflow stages and uses explicit contracts to keep decisions consistent across those stages.","excerpt":"selecting reusable procedural skills for each workflow stage and uses explicit contracts to maintain consistent study decisions across stages","url":"https://arxiv.org/abs/2611.05551","relationship":"Source-stated"},{"text":"The authors report that, after requirement clarification, required cohort outputs matched references in 95.1% of workflows across three synthetic environments, versus 60.4% with retrieval from the same procedural corpus.","excerpt":"Across three complex synthetic environments, VTHarness successfully produced all required cohort outputs matching the references in 95.1% of workflows after requirement clarification, compared with 60.4% using retrieval from the same procedural corpus.","url":"https://arxiv.org/abs/2611.05551","relationship":"Source-stated"},{"text":"The authors report improved source recovery across three language models and increased source overlap with analyst code in 13 Mayo Clinic cases.","excerpt":"Procedural skills improved source recovery across three language models and increased source overlap with analyst code in 13 Mayo Clinic cases.","url":"https://arxiv.org/abs/2611.05551","relationship":"Source-stated"},{"text":"The abstract reports reference-matched cohort execution across eight MIMIC-IV tasks with site-specific adaptation, and investigator review support within the Mayo Clinic Virtual Trials Program.","excerpt":"Site-specific adaptation enabled reference-matched cohort execution across eight MIMIC-IV tasks. Within the Mayo Clinic Virtual Trials Program, the framework supported investigator review of cohort construction and endpoint analysis.","url":"https://arxiv.org/abs/2611.05551","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.05551","checkedAt":"2026-11-09T08:54:10.239Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How would VirtualTrialHarness perform on virtual-trial workflows beyond the synthetic environments, Mayo Clinic cases, and MIMIC-IV tasks described in the abstract?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"The authors’ VirtualTrialHarness coordinates decisions across virtual-trial workflows","summary":"The authors introduce VirtualTrialHarness, an agent framework for turning virtual-trial protocols into executable analyses across complex clinical schemas. They argue that this takes more than retrieving procedural guidance: teams must decide which data sources to use, how to identify and link patients and encounters, and how to interpret time. VirtualTrialHarness selects reusable procedural skills for each workflow stage and uses explicit contracts to keep study decisions consistent across stages.\n\nThe authors also introduce VTBenchmark to evaluate source selection and cohort construction. Across three complex synthetic environments, required cohort outputs matched references in 95.1% of workflows after requirement clarification, compared with 60.4% using retrieval from the same procedural corpus. The abstract also reports improved source recovery across three language models, increased source overlap with analyst code in 13 Mayo Clinic cases, and reference-matched cohort execution across eight MIMIC-IV tasks after site-specific adaptation. Within the Mayo Clinic Virtual Trials Program, the framework supported investigator review of cohort construction and endpoint analysis. The authors’ central idea is to coordinate procedural guidance with decisions across the workflow, rather than rely on retrieval alone.","question":"How would VirtualTrialHarness perform on virtual-trial workflows beyond the synthetic environments, Mayo Clinic cases, and MIMIC-IV tasks described in the abstract?","claims":[{"text":"The authors introduce VirtualTrialHarness, which selects reusable procedural skills for workflow stages and uses explicit contracts to keep decisions consistent across those stages.","excerpt":"selecting reusable procedural skills for each workflow stage and uses explicit contracts to maintain consistent study decisions across stages","support":"source-stated"},{"text":"The authors report that, after requirement clarification, required cohort outputs matched references in 95.1% of workflows across three synthetic environments, versus 60.4% with retrieval from the same procedural corpus.","excerpt":"Across three complex synthetic environments, VTHarness successfully produced all required cohort outputs matching the references in 95.1% of workflows after requirement clarification, compared with 60.4% using retrieval from the same procedural corpus.","support":"source-stated"},{"text":"The authors report improved source recovery across three language models and increased source overlap with analyst code in 13 Mayo Clinic cases.","excerpt":"Procedural skills improved source recovery across three language models and increased source overlap with analyst code in 13 Mayo Clinic cases.","support":"source-stated"},{"text":"The abstract reports reference-matched cohort execution across eight MIMIC-IV tasks with site-specific adaptation, and investigator review support within the Mayo Clinic Virtual Trials Program.","excerpt":"Site-specific adaptation enabled reference-matched cohort execution across eight MIMIC-IV tasks. Within the Mayo Clinic Virtual Trials Program, the framework supported investigator review of cohort construction and endpoint analysis.","support":"source-stated"}],"limitations":["The reported workflow-match result is from three complex synthetic environments and is stated to follow requirement clarification.","The abstract describes 13 Mayo Clinic cases and eight MIMIC-IV tasks; it does not establish results across a broader range of clinical sites or workflows."],"mediaId":"9ec66997e4c67d1609559ded4a0913e6570dce6a69a9a44977ad37bd41f94581","correctionReason":null,"id":"bf36f430636c50240c91bf35b143c11b7b80cdad5e35284639cb6b1c003126ec","publicationId":"d409c862-bdb3-441c-8416-5b1e2c59a5c0","version":1,"sourceUrl":"https://arxiv.org/abs/2611.05551","sourceTitle":"VirtualTrialHarness: Coordinating Agentic Operationalization of Virtual Trial Protocols Across Complex Clinical Schemas","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-11-03","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-05T05:06:29.537Z","createdAt":"2026-11-09T08:38:28.338Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"9ec66997e4c67d1609559ded4a0913e6570dce6a69a9a44977ad37bd41f94581","kind":"article","url":"https://arxiv.org/abs/2611.05551","sourceUrl":"https://arxiv.org/abs/2611.05551","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-09T08:37:41.130Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.05551","checkedAt":"2026-11-09T08:54:10.239Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"9ec66997e4c67d1609559ded4a0913e6570dce6a69a9a44977ad37bd41f94581","confidence":0.84,"provider":"typesafe","model":"jev-1.13.0","inputHash":"7101bed420a0d1f152db9e32c9a70c7c0c39a21246fa51177a5a61d22acb64c6","decidedAt":"2026-11-09T08:37:41.817Z"},"provenance":{"textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observationId":"8ff06bcad296ecad5ecaecdcb94613ab90798b4ae9dfa3a06f70936bff34cf8b","retrievedAt":"2026-11-09T08:37:41.011Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"19d23d6c-6ad2-4225-8ebb-296b0d99e98b"},"reviewer":{"model":"gpt-6-luna","runId":"fe0f32a4-28a1-447c-87f5-f0137bbdf989"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-09T08:38:28.338Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Li, Pengze","url":null,"kind":"unknown"},{"name":"Feng, Jingna","url":null,"kind":"unknown"},{"name":"Yu, Yue","url":null,"kind":"unknown"},{"name":"Wu, Yining","url":null,"kind":"unknown"},{"name":"Hu, Xinyue","url":null,"kind":"unknown"},{"name":"Lin, Jie","url":null,"kind":"unknown"},{"name":"Bhandarkar, Avanti","url":null,"kind":"unknown"},{"name":"Tao, Cui","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Li, Pengze","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Li, Pengze","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Feng, Jingna","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Feng, Jingna","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Yu, Yue","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yu, Yue","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Wu, Yining","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Wu, Yining","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Hu, Xinyue","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Hu, Xinyue","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Lin, Jie","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lin, Jie","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Bhandarkar, Avanti","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Bhandarkar, Avanti","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:6","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Tao, Cui","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Tao, Cui","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:7","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}}],"participationCoverage":{"total":8,"returned":8,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.05551","snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z","basis":"publisher-page-metadata","sourceLinks":43,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/bf36f430636c50240c91bf35b143c11b7b80cdad5e35284639cb6b1c003126ec/source"}},"canonicalUrl":"https://escapemeridian.com/readings/the-authors-virtualtrialharness-coordinates-decisions-across-bf36f430","formats":{"html":"/readings/the-authors-virtualtrialharness-coordinates-decisions-across-bf36f430","markdown":"/readings/the-authors-virtualtrialharness-coordinates-decisions-across-bf36f430.md","json":"/readings/the-authors-virtualtrialharness-coordinates-decisions-across-bf36f430.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"},{"id":"time","name":"TIME","kind":"organization","context":"TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.","referenceUrls":["https://time.com/"],"path":"/entities/time","json":"/entities/time.json","markdown":"/entities/time.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"bf36f430636c50240c91bf35b143c11b7b80cdad5e35284639cb6b1c003126ec","revision":"d409c862-bdb3-441c-8416-5b1e2c59a5c0","revisionKind":"publication-id","title":"The authors’ VirtualTrialHarness coordinates decisions across virtual-trial workflows","url":"https://escapemeridian.com/readings/the-authors-virtualtrialharness-coordinates-decisions-across-bf36f430"},"source":{"type":"source","id":"https://arxiv.org/abs/2611.05551","revision":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","revisionKind":"sha256-source-text","title":"VirtualTrialHarness: Coordinating Agentic Operationalization of Virtual Trial Protocols Across Complex Clinical Schemas","url":"https://arxiv.org/abs/2611.05551"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"bf36f430636c50240c91bf35b143c11b7b80cdad5e35284639cb6b1c003126ec","title":"The authors’ VirtualTrialHarness coordinates decisions across virtual-trial workflows","argument":"The authors introduce VirtualTrialHarness, an agent framework for turning virtual-trial protocols into executable analyses across complex clinical schemas. They argue that this takes more than retrieving procedural guidance: teams must decide which data sources to use, how to identify and link patients and encounters, and how to interpret time. VirtualTrialHarness selects reusable procedural skills for each workflow stage and uses explicit contracts to keep study decisions consistent across stages.\n\nThe authors also introduce VTBenchmark to evaluate source selection and cohort construction. Across three complex synthetic environments, required cohort outputs matched references in 95.1% of workflows after requirement clarification, compared with 60.4% using retrieval from the same procedural corpus. The abstract also reports improved source recovery across three language models, increased source overlap with analyst code in 13 Mayo Clinic cases, and reference-matched cohort execution across eight MIMIC-IV tasks after site-specific adaptation. Within the Mayo Clinic Virtual Trials Program, the framework supported investigator review of cohort construction and endpoint analysis. The authors’ central idea is to coordinate procedural guidance with decisions across the workflow, rather than rely on retrieval alone.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The reported workflow-match result is from three complex synthetic environments and is stated to follow requirement clarification. The abstract describes 13 Mayo Clinic cases and eight MIMIC-IV tasks; it does not establish results across a broader range of clinical sites or workflows.","speaker":{"name":"Li, Pengze, Feng, Jingna, Yu, Yue, Wu, Yining, Hu, Xinyue, Lin, Jie, Bhandarkar, Avanti, Tao, Cui","roleAtSource":"Article author"},"source":{"title":"VirtualTrialHarness: Coordinating Agentic Operationalization of Virtual Trial Protocols Across Complex Clinical Schemas","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2611.05551","publishedAt":"2026-11-03","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors introduce VirtualTrialHarness, which selects reusable procedural skills for workflow stages and uses explicit contracts to keep decisions consistent across those stages.","excerpt":"selecting reusable procedural skills for each workflow stage and uses explicit contracts to maintain consistent study decisions across stages","url":"https://arxiv.org/abs/2611.05551","relationship":"Source-stated"},{"text":"The authors report that, after requirement clarification, required cohort outputs matched references in 95.1% of workflows across three synthetic environments, versus 60.4% with retrieval from the same procedural corpus.","excerpt":"Across three complex synthetic environments, VTHarness successfully produced all required cohort outputs matching the references in 95.1% of workflows after requirement clarification, compared with 60.4% using retrieval from the same procedural corpus.","url":"https://arxiv.org/abs/2611.05551","relationship":"Source-stated"},{"text":"The authors report improved source recovery across three language models and increased source overlap with analyst code in 13 Mayo Clinic cases.","excerpt":"Procedural skills improved source recovery across three language models and increased source overlap with analyst code in 13 Mayo Clinic cases.","url":"https://arxiv.org/abs/2611.05551","relationship":"Source-stated"},{"text":"The abstract reports reference-matched cohort execution across eight MIMIC-IV tasks with site-specific adaptation, and investigator review support within the Mayo Clinic Virtual Trials Program.","excerpt":"Site-specific adaptation enabled reference-matched cohort execution across eight MIMIC-IV tasks. Within the Mayo Clinic Virtual Trials Program, the framework supported investigator review of cohort construction and endpoint analysis.","url":"https://arxiv.org/abs/2611.05551","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.05551","checkedAt":"2026-11-09T08:54:10.239Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How would VirtualTrialHarness perform on virtual-trial workflows beyond the synthetic environments, Mayo Clinic cases, and MIMIC-IV tasks described in the abstract?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"The authors’ VirtualTrialHarness coordinates decisions across virtual-trial workflows","summary":"The authors introduce VirtualTrialHarness, an agent framework for turning virtual-trial protocols into executable analyses across complex clinical schemas. They argue that this takes more than retrieving procedural guidance: teams must decide which data sources to use, how to identify and link patients and encounters, and how to interpret time. VirtualTrialHarness selects reusable procedural skills for each workflow stage and uses explicit contracts to keep study decisions consistent across stages.\n\nThe authors also introduce VTBenchmark to evaluate source selection and cohort construction. Across three complex synthetic environments, required cohort outputs matched references in 95.1% of workflows after requirement clarification, compared with 60.4% using retrieval from the same procedural corpus. The abstract also reports improved source recovery across three language models, increased source overlap with analyst code in 13 Mayo Clinic cases, and reference-matched cohort execution across eight MIMIC-IV tasks after site-specific adaptation. Within the Mayo Clinic Virtual Trials Program, the framework supported investigator review of cohort construction and endpoint analysis. The authors’ central idea is to coordinate procedural guidance with decisions across the workflow, rather than rely on retrieval alone.","question":"How would VirtualTrialHarness perform on virtual-trial workflows beyond the synthetic environments, Mayo Clinic cases, and MIMIC-IV tasks described in the abstract?","claims":[{"text":"The authors introduce VirtualTrialHarness, which selects reusable procedural skills for workflow stages and uses explicit contracts to keep decisions consistent across those stages.","excerpt":"selecting reusable procedural skills for each workflow stage and uses explicit contracts to maintain consistent study decisions across stages","support":"source-stated"},{"text":"The authors report that, after requirement clarification, required cohort outputs matched references in 95.1% of workflows across three synthetic environments, versus 60.4% with retrieval from the same procedural corpus.","excerpt":"Across three complex synthetic environments, VTHarness successfully produced all required cohort outputs matching the references in 95.1% of workflows after requirement clarification, compared with 60.4% using retrieval from the same procedural corpus.","support":"source-stated"},{"text":"The authors report improved source recovery across three language models and increased source overlap with analyst code in 13 Mayo Clinic cases.","excerpt":"Procedural skills improved source recovery across three language models and increased source overlap with analyst code in 13 Mayo Clinic cases.","support":"source-stated"},{"text":"The abstract reports reference-matched cohort execution across eight MIMIC-IV tasks with site-specific adaptation, and investigator review support within the Mayo Clinic Virtual Trials Program.","excerpt":"Site-specific adaptation enabled reference-matched cohort execution across eight MIMIC-IV tasks. Within the Mayo Clinic Virtual Trials Program, the framework supported investigator review of cohort construction and endpoint analysis.","support":"source-stated"}],"limitations":["The reported workflow-match result is from three complex synthetic environments and is stated to follow requirement clarification.","The abstract describes 13 Mayo Clinic cases and eight MIMIC-IV tasks; it does not establish results across a broader range of clinical sites or workflows."],"mediaId":"9ec66997e4c67d1609559ded4a0913e6570dce6a69a9a44977ad37bd41f94581","correctionReason":null,"id":"bf36f430636c50240c91bf35b143c11b7b80cdad5e35284639cb6b1c003126ec","publicationId":"d409c862-bdb3-441c-8416-5b1e2c59a5c0","version":1,"sourceUrl":"https://arxiv.org/abs/2611.05551","sourceTitle":"VirtualTrialHarness: Coordinating Agentic Operationalization of Virtual Trial Protocols Across Complex Clinical Schemas","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-11-03","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-05T05:06:29.537Z","createdAt":"2026-11-09T08:38:28.338Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"9ec66997e4c67d1609559ded4a0913e6570dce6a69a9a44977ad37bd41f94581","kind":"article","url":"https://arxiv.org/abs/2611.05551","sourceUrl":"https://arxiv.org/abs/2611.05551","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-09T08:37:41.130Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.05551","checkedAt":"2026-11-09T08:54:10.239Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"9ec66997e4c67d1609559ded4a0913e6570dce6a69a9a44977ad37bd41f94581","confidence":0.84,"provider":"typesafe","model":"jev-1.13.0","inputHash":"7101bed420a0d1f152db9e32c9a70c7c0c39a21246fa51177a5a61d22acb64c6","decidedAt":"2026-11-09T08:37:41.817Z"},"provenance":{"textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observationId":"8ff06bcad296ecad5ecaecdcb94613ab90798b4ae9dfa3a06f70936bff34cf8b","retrievedAt":"2026-11-09T08:37:41.011Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"19d23d6c-6ad2-4225-8ebb-296b0d99e98b"},"reviewer":{"model":"gpt-6-luna","runId":"fe0f32a4-28a1-447c-87f5-f0137bbdf989"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-09T08:38:28.338Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Li, Pengze","url":null,"kind":"unknown"},{"name":"Feng, Jingna","url":null,"kind":"unknown"},{"name":"Yu, Yue","url":null,"kind":"unknown"},{"name":"Wu, Yining","url":null,"kind":"unknown"},{"name":"Hu, Xinyue","url":null,"kind":"unknown"},{"name":"Lin, Jie","url":null,"kind":"unknown"},{"name":"Bhandarkar, Avanti","url":null,"kind":"unknown"},{"name":"Tao, Cui","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Li, Pengze","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Li, Pengze","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Feng, Jingna","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Feng, Jingna","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Yu, Yue","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yu, Yue","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Wu, Yining","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Wu, Yining","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Hu, Xinyue","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Hu, Xinyue","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Lin, Jie","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lin, Jie","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Bhandarkar, Avanti","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Bhandarkar, Avanti","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:6","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}},{"actor":{"name":"Tao, Cui","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Tao, Cui","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05551","locator":"head/meta[name=author|citation_author]"},"assertionId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138:participation:7","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T08:54:10.239Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","sourceUrl":"https://arxiv.org/abs/2611.05551","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z"}}],"participationCoverage":{"total":8,"returned":8,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.05551","snapshotId":"d5e2d9ce686314fd3b7fc179d4144f069004920e48fa355d23da5d8515bf0138","contentSha256":"a4dec1fa129b4b5200401a76d2d1cf8b87cb8987f09fd4e832a4cf7de00d80b4","textHash":"dbe86a2fbaf3a79e2db26df804e5d063276d2dd3bb682f3a9d37a35f6bf2e6ee","observedAt":"2026-11-09T08:54:10.239Z","basis":"publisher-page-metadata","sourceLinks":43,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/bf36f430636c50240c91bf35b143c11b7b80cdad5e35284639cb6b1c003126ec/source"}}}],"audits":[{"publicationId":"d409c862-bdb3-441c-8416-5b1e2c59a5c0","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading credits the framework, benchmark and results to the paper's authors (Pengze Li et al.) as their claims. It keeps reported figures as findings the authors or abstract report, not as established facts.","excerpt":"Authors:Pengze Li, Jingna Feng, Yue Yu, Yining Wu, Xinyue Hu, Jie Lin, Avanti Bhandarkar, Cui Tao"},{"dimension":"summary","status":"uncertain","explanation":"The summary is faithful, but it never says what a virtual trial is. The source explains that it uses routinely collected clinical data to evaluate treatment strategies. Without that, a non-specialist cannot see the practical purpose or stakes. Cohort, clinical schemas and MIMIC-IV are also unexplained central terms. Smallest repair: add the source's one-line definition of virtual trials, then a brief plain gloss of cohort (the patient group selected for a study). MIMIC-IV can be named as a dataset, or a tooltip can be suggested.","excerpt":"Virtual trials use routinely collected clinical data to evaluate treatment strategies"},{"dimension":"date","status":"pass","explanation":"originalPublished 2026-11-03 matches the v1 submission date. The later feed date of Nov 5 is correctly not used.","excerpt":"[Submitted on 3 Nov 2026]"},{"dimension":"support","status":"pass","explanation":"All claims match the abstract verbatim in substance: 95.1% vs 60.4% after requirement clarification, three language models, 13 Mayo Clinic cases, eight MIMIC-IV tasks with site-specific adaptation, and investigator review support. The framing of the central idea against retrieval alone tracks the authors' comparison and conclusion. The limitations and the question stay within the evidence.","excerpt":"These findings support coordinating procedural guidance with workflow decisions to improve virtual-trial implementation across complex clinical schemas."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct arXiv abstract page for this paper, which is the most direct access to the source.","excerpt":null}],"createdAt":"2026-11-09T08:38:28.338Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}