{"id":"0a01eb10baec3cbfd34a62fbf950daaf0a9678964bbf462b49d69a3c73c1768c","title":"arXiv cs.AI: Can General-Purpose Generative AI Supply Training Data for Medical Vision Models?","argument":"The paper’s authors investigate whether Nano Banana, a general-purpose image generation and editing model, can supply synthetic medical images for training and testing medical vision models. They report results across chest radiography, fundus photography, CT, and carotid ultrasound, including tests on real-world data. Models trained only on synthetic images achieved AUROCs comparable to models trained on real data for pneumonia, retinal detachment, and renal calculus. But the authors say calibration and net benefit were not consistently more favorable, cautioning against judging clinical value by discrimination alone. In carotid ultrasound, mixed synthetic-real training produced an ordinal vascular age index that stratified external subjects into groups differing in BMI and glucose. The authors also report that editing a pattern into fundus photographs reduced a model’s true-class probability for central serous chorioretinopathy, and clinician-guided editing improved the plausibility ratings of low-quality CT images. They interpret these results as evidence that synthetic data can complement real data in training, subgroup representation, and model testing. Their conclusion is qualified: assessment should include calibration, decision curve analysis, and clinician review, not discrimination alone.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The evidence is a retrospective study; the abstract does not establish how these results would transfer to other models, datasets, or clinical use. The abstract reports that calibration and net benefit were not consistently more favorable, so comparable AUROCs alone do not establish comparable clinical value.","speaker":{"name":"Peng, Jiachuan, Lam, Kyle, Qu, Jierui, Li, Lin, Zheng, Haining, Zheng, Bang, Agarwal, Aniruddha, Almallah, Zaki, Chen, Min, Grau, Vicente, Rossman, Hagai, Segal, Eran, Qiu, Jianing","roleAtSource":"Article author"},"source":{"title":"Can General-Purpose Generative AI Supply Training Data for Medical Vision Models? A Multimodal Retrospective Study","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2611.10484","publishedAt":"2026-11-07","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors retrospectively evaluated Nano Banana-generated medical images for model training, subgroup augmentation, and auditing across imaging modalities.","excerpt":"We evaluated Nano Banana, a general-purpose image generation and editing model, as a source of synthetic medical images for model training, subgroup augmentation, and model auditing.","url":"https://arxiv.org/abs/2611.10484","relationship":"Source-stated"},{"text":"For three conditions, synthetic-only models achieved AUROCs comparable to real-data-trained models, while calibration and net benefit were not consistently more favorable.","excerpt":"Synthetic-only models achieved AUROCs of 0.877 (95% CI 0.843-0.910) for pneumonia (n=14863), 0.963 (0.948-0.975) for retinal detachment (n=1149), and 0.711 (0.600-0.823) for renal calculus (n=4122), comparable to real-data-trained models, but their calibration and net benefit were not consistently","url":"https://arxiv.org/abs/2611.10484","relationship":"Source-stated"},{"text":"With mixed synthetic-real training, the carotid ultrasound model's vascular age index separated external subjects into subgroups with differing BMI and glucose.","excerpt":"a carotid ultrasound model produced an ordinal vascular age index that stratified 633 external subjects into subgroups differing in BMI and glucose (both p","url":"https://arxiv.org/abs/2611.10484","relationship":"Source-stated"},{"text":"Editing a pattern into fundus photographs reduced the model’s true-class probability for central serous chorioretinopathy.","excerpt":"editing a moderate tessellated pattern into fundus photographs reduced the true-class probability for central serous chorioretinopathy from 0.996 to 0.822.","url":"https://arxiv.org/abs/2611.10484","relationship":"Source-stated"},{"text":"The authors conclude synthetic data can complement real data, but recommend assessing clinical value with measures beyond discrimination.","excerpt":"Synthetic data can complement real data in model training, subgroup representation, and model testing via controlled editing, but its clinical value should be assessed with calibration, decision curve analysis, and clinician review, not discrimination alone.","url":"https://arxiv.org/abs/2611.10484","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.10484","checkedAt":"2026-11-12T09:34:11.624Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"Across further evaluations, would synthetic-only models show calibration and net benefit as favorable as their discrimination?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv cs.AI: Can General-Purpose Generative AI Supply Training Data for Medical Vision Models?","summary":"The paper’s authors investigate whether Nano Banana, a general-purpose image generation and editing model, can supply synthetic medical images for training and testing medical vision models. They report results across chest radiography, fundus photography, CT, and carotid ultrasound, including tests on real-world data. Models trained only on synthetic images achieved AUROCs comparable to models trained on real data for pneumonia, retinal detachment, and renal calculus. But the authors say calibration and net benefit were not consistently more favorable, cautioning against judging clinical value by discrimination alone. In carotid ultrasound, mixed synthetic-real training produced an ordinal vascular age index that stratified external subjects into groups differing in BMI and glucose. The authors also report that editing a pattern into fundus photographs reduced a model’s true-class probability for central serous chorioretinopathy, and clinician-guided editing improved the plausibility ratings of low-quality CT images. They interpret these results as evidence that synthetic data can complement real data in training, subgroup representation, and model testing. Their conclusion is qualified: assessment should include calibration, decision curve analysis, and clinician review, not discrimination alone.","question":"Across further evaluations, would synthetic-only models show calibration and net benefit as favorable as their discrimination?","claims":[{"text":"The authors retrospectively evaluated Nano Banana-generated medical images for model training, subgroup augmentation, and auditing across imaging modalities.","excerpt":"We evaluated Nano Banana, a general-purpose image generation and editing model, as a source of synthetic medical images for model training, subgroup augmentation, and model auditing.","support":"source-stated"},{"text":"For three conditions, synthetic-only models achieved AUROCs comparable to real-data-trained models, while calibration and net benefit were not consistently more favorable.","excerpt":"Synthetic-only models achieved AUROCs of 0.877 (95% CI 0.843-0.910) for pneumonia (n=14863), 0.963 (0.948-0.975) for retinal detachment (n=1149), and 0.711 (0.600-0.823) for renal calculus (n=4122), comparable to real-data-trained models, but their calibration and net benefit were not consistently","support":"source-stated"},{"text":"With mixed synthetic-real training, the carotid ultrasound model's vascular age index separated external subjects into subgroups with differing BMI and glucose.","excerpt":"a carotid ultrasound model produced an ordinal vascular age index that stratified 633 external subjects into subgroups differing in BMI and glucose (both p","support":"source-stated"},{"text":"Editing a pattern into fundus photographs reduced the model’s true-class probability for central serous chorioretinopathy.","excerpt":"editing a moderate tessellated pattern into fundus photographs reduced the true-class probability for central serous chorioretinopathy from 0.996 to 0.822.","support":"source-stated"},{"text":"The authors conclude synthetic data can complement real data, but recommend assessing clinical value with measures beyond discrimination.","excerpt":"Synthetic data can complement real data in model training, subgroup representation, and model testing via controlled editing, but its clinical value should be assessed with calibration, decision curve analysis, and clinician review, not discrimination alone.","support":"source-stated"}],"limitations":["The evidence is a retrospective study; the abstract does not establish how these results would transfer to other models, datasets, or clinical use.","The abstract reports that calibration and net benefit were not consistently more favorable, so comparable AUROCs alone do not establish comparable clinical value."],"mediaId":"67fbe957d652e39b266d21fe8a4221afccef5c3f9da7a64b2e5ff596dcbab29d","correctionReason":null,"id":"0a01eb10baec3cbfd34a62fbf950daaf0a9678964bbf462b49d69a3c73c1768c","publicationId":"2a0f11a1-0480-4534-8082-229a97ef3f34","version":1,"sourceUrl":"https://arxiv.org/abs/2611.10484","sourceTitle":"Can General-Purpose Generative AI Supply Training Data for Medical Vision Models? A Multimodal Retrospective Study","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-11-07","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-10T08:42:32.173Z","createdAt":"2026-11-12T09:03:44.716Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"67fbe957d652e39b266d21fe8a4221afccef5c3f9da7a64b2e5ff596dcbab29d","kind":"article","url":"https://arxiv.org/abs/2611.10484","sourceUrl":"https://arxiv.org/abs/2611.10484","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-12T09:02:51.141Z","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/2611.10484","checkedAt":"2026-11-12T09:34:11.624Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v6","mediaId":"67fbe957d652e39b266d21fe8a4221afccef5c3f9da7a64b2e5ff596dcbab29d","confidence":0.9,"provider":"typesafe","model":"jev-1.13.0","inputHash":"1a0dae684f926f6083138a26cfc16ee386a002159908fe1ac300fb082444f5a6","decidedAt":"2026-11-12T09:02:53.160Z"},"provenance":{"textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observationId":"fefde60808a6f0ea3016aaf2d867ae508b80c8312f9026b19afe348cda27bd76","retrievedAt":"2026-11-12T09:02:50.853Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"8a4eaa5c-bdd9-47a4-8d61-3d06432e668f"},"reviewer":{"model":"gpt-6-luna","runId":"f84e2702-511f-403f-8182-7df19a62f3b1"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-12T09:03:44.716Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Peng, Jiachuan","url":null,"kind":"unknown"},{"name":"Lam, Kyle","url":null,"kind":"unknown"},{"name":"Qu, Jierui","url":null,"kind":"unknown"},{"name":"Li, Lin","url":null,"kind":"unknown"},{"name":"Zheng, Haining","url":null,"kind":"unknown"},{"name":"Zheng, Bang","url":null,"kind":"unknown"},{"name":"Agarwal, Aniruddha","url":null,"kind":"unknown"},{"name":"Almallah, Zaki","url":null,"kind":"unknown"},{"name":"Chen, Min","url":null,"kind":"unknown"},{"name":"Grau, Vicente","url":null,"kind":"unknown"},{"name":"Rossman, Hagai","url":null,"kind":"unknown"},{"name":"Segal, Eran","url":null,"kind":"unknown"},{"name":"Qiu, Jianing","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Peng, Jiachuan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Peng, Jiachuan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Lam, Kyle","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lam, Kyle","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Qu, Jierui","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Qu, Jierui","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Li, Lin","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Li, Lin","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Zheng, Haining","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zheng, Haining","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Zheng, Bang","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zheng, Bang","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Agarwal, Aniruddha","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Agarwal, Aniruddha","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:6","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Almallah, Zaki","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Almallah, Zaki","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:7","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Chen, Min","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Chen, Min","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:8","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Grau, Vicente","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Grau, Vicente","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:9","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Rossman, Hagai","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Rossman, Hagai","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:10","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Segal, Eran","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Segal, Eran","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:11","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Qiu, Jianing","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Qiu, Jianing","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:12","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}}],"participationCoverage":{"total":13,"returned":13,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.10484","snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z","basis":"publisher-page-metadata","sourceLinks":52,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/0a01eb10baec3cbfd34a62fbf950daaf0a9678964bbf462b49d69a3c73c1768c/source"}},"canonicalUrl":"https://escapemeridian.com/readings/arxiv-cs-ai-can-general-purpose-generative-ai-supply-training-0a01eb10","formats":{"html":"/readings/arxiv-cs-ai-can-general-purpose-generative-ai-supply-training-0a01eb10","markdown":"/readings/arxiv-cs-ai-can-general-purpose-generative-ai-supply-training-0a01eb10.md","json":"/readings/arxiv-cs-ai-can-general-purpose-generative-ai-supply-training-0a01eb10.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":[{"id":"term-synthetic-data","name":"Synthetic data","explanation":"Training or test data generated by software or another model rather than collected from the world.","referenceUrl":"https://en.wikipedia.org/wiki/Synthetic_data"}],"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"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"0a01eb10baec3cbfd34a62fbf950daaf0a9678964bbf462b49d69a3c73c1768c","revision":"2a0f11a1-0480-4534-8082-229a97ef3f34","revisionKind":"publication-id","title":"arXiv cs.AI: Can General-Purpose Generative AI Supply Training Data for Medical Vision Models?","url":"https://escapemeridian.com/readings/arxiv-cs-ai-can-general-purpose-generative-ai-supply-training-0a01eb10"},"source":{"type":"source","id":"https://arxiv.org/abs/2611.10484","revision":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","revisionKind":"sha256-source-text","title":"Can General-Purpose Generative AI Supply Training Data for Medical Vision Models? A Multimodal Retrospective Study","url":"https://arxiv.org/abs/2611.10484"},"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":"0a01eb10baec3cbfd34a62fbf950daaf0a9678964bbf462b49d69a3c73c1768c","title":"arXiv cs.AI: Can General-Purpose Generative AI Supply Training Data for Medical Vision Models?","argument":"The paper’s authors investigate whether Nano Banana, a general-purpose image generation and editing model, can supply synthetic medical images for training and testing medical vision models. They report results across chest radiography, fundus photography, CT, and carotid ultrasound, including tests on real-world data. Models trained only on synthetic images achieved AUROCs comparable to models trained on real data for pneumonia, retinal detachment, and renal calculus. But the authors say calibration and net benefit were not consistently more favorable, cautioning against judging clinical value by discrimination alone. In carotid ultrasound, mixed synthetic-real training produced an ordinal vascular age index that stratified external subjects into groups differing in BMI and glucose. The authors also report that editing a pattern into fundus photographs reduced a model’s true-class probability for central serous chorioretinopathy, and clinician-guided editing improved the plausibility ratings of low-quality CT images. They interpret these results as evidence that synthetic data can complement real data in training, subgroup representation, and model testing. Their conclusion is qualified: assessment should include calibration, decision curve analysis, and clinician review, not discrimination alone.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The evidence is a retrospective study; the abstract does not establish how these results would transfer to other models, datasets, or clinical use. The abstract reports that calibration and net benefit were not consistently more favorable, so comparable AUROCs alone do not establish comparable clinical value.","speaker":{"name":"Peng, Jiachuan, Lam, Kyle, Qu, Jierui, Li, Lin, Zheng, Haining, Zheng, Bang, Agarwal, Aniruddha, Almallah, Zaki, Chen, Min, Grau, Vicente, Rossman, Hagai, Segal, Eran, Qiu, Jianing","roleAtSource":"Article author"},"source":{"title":"Can General-Purpose Generative AI Supply Training Data for Medical Vision Models? A Multimodal Retrospective Study","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2611.10484","publishedAt":"2026-11-07","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors retrospectively evaluated Nano Banana-generated medical images for model training, subgroup augmentation, and auditing across imaging modalities.","excerpt":"We evaluated Nano Banana, a general-purpose image generation and editing model, as a source of synthetic medical images for model training, subgroup augmentation, and model auditing.","url":"https://arxiv.org/abs/2611.10484","relationship":"Source-stated"},{"text":"For three conditions, synthetic-only models achieved AUROCs comparable to real-data-trained models, while calibration and net benefit were not consistently more favorable.","excerpt":"Synthetic-only models achieved AUROCs of 0.877 (95% CI 0.843-0.910) for pneumonia (n=14863), 0.963 (0.948-0.975) for retinal detachment (n=1149), and 0.711 (0.600-0.823) for renal calculus (n=4122), comparable to real-data-trained models, but their calibration and net benefit were not consistently","url":"https://arxiv.org/abs/2611.10484","relationship":"Source-stated"},{"text":"With mixed synthetic-real training, the carotid ultrasound model's vascular age index separated external subjects into subgroups with differing BMI and glucose.","excerpt":"a carotid ultrasound model produced an ordinal vascular age index that stratified 633 external subjects into subgroups differing in BMI and glucose (both p","url":"https://arxiv.org/abs/2611.10484","relationship":"Source-stated"},{"text":"Editing a pattern into fundus photographs reduced the model’s true-class probability for central serous chorioretinopathy.","excerpt":"editing a moderate tessellated pattern into fundus photographs reduced the true-class probability for central serous chorioretinopathy from 0.996 to 0.822.","url":"https://arxiv.org/abs/2611.10484","relationship":"Source-stated"},{"text":"The authors conclude synthetic data can complement real data, but recommend assessing clinical value with measures beyond discrimination.","excerpt":"Synthetic data can complement real data in model training, subgroup representation, and model testing via controlled editing, but its clinical value should be assessed with calibration, decision curve analysis, and clinician review, not discrimination alone.","url":"https://arxiv.org/abs/2611.10484","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.10484","checkedAt":"2026-11-12T09:34:11.624Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"Across further evaluations, would synthetic-only models show calibration and net benefit as favorable as their discrimination?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv cs.AI: Can General-Purpose Generative AI Supply Training Data for Medical Vision Models?","summary":"The paper’s authors investigate whether Nano Banana, a general-purpose image generation and editing model, can supply synthetic medical images for training and testing medical vision models. They report results across chest radiography, fundus photography, CT, and carotid ultrasound, including tests on real-world data. Models trained only on synthetic images achieved AUROCs comparable to models trained on real data for pneumonia, retinal detachment, and renal calculus. But the authors say calibration and net benefit were not consistently more favorable, cautioning against judging clinical value by discrimination alone. In carotid ultrasound, mixed synthetic-real training produced an ordinal vascular age index that stratified external subjects into groups differing in BMI and glucose. The authors also report that editing a pattern into fundus photographs reduced a model’s true-class probability for central serous chorioretinopathy, and clinician-guided editing improved the plausibility ratings of low-quality CT images. They interpret these results as evidence that synthetic data can complement real data in training, subgroup representation, and model testing. Their conclusion is qualified: assessment should include calibration, decision curve analysis, and clinician review, not discrimination alone.","question":"Across further evaluations, would synthetic-only models show calibration and net benefit as favorable as their discrimination?","claims":[{"text":"The authors retrospectively evaluated Nano Banana-generated medical images for model training, subgroup augmentation, and auditing across imaging modalities.","excerpt":"We evaluated Nano Banana, a general-purpose image generation and editing model, as a source of synthetic medical images for model training, subgroup augmentation, and model auditing.","support":"source-stated"},{"text":"For three conditions, synthetic-only models achieved AUROCs comparable to real-data-trained models, while calibration and net benefit were not consistently more favorable.","excerpt":"Synthetic-only models achieved AUROCs of 0.877 (95% CI 0.843-0.910) for pneumonia (n=14863), 0.963 (0.948-0.975) for retinal detachment (n=1149), and 0.711 (0.600-0.823) for renal calculus (n=4122), comparable to real-data-trained models, but their calibration and net benefit were not consistently","support":"source-stated"},{"text":"With mixed synthetic-real training, the carotid ultrasound model's vascular age index separated external subjects into subgroups with differing BMI and glucose.","excerpt":"a carotid ultrasound model produced an ordinal vascular age index that stratified 633 external subjects into subgroups differing in BMI and glucose (both p","support":"source-stated"},{"text":"Editing a pattern into fundus photographs reduced the model’s true-class probability for central serous chorioretinopathy.","excerpt":"editing a moderate tessellated pattern into fundus photographs reduced the true-class probability for central serous chorioretinopathy from 0.996 to 0.822.","support":"source-stated"},{"text":"The authors conclude synthetic data can complement real data, but recommend assessing clinical value with measures beyond discrimination.","excerpt":"Synthetic data can complement real data in model training, subgroup representation, and model testing via controlled editing, but its clinical value should be assessed with calibration, decision curve analysis, and clinician review, not discrimination alone.","support":"source-stated"}],"limitations":["The evidence is a retrospective study; the abstract does not establish how these results would transfer to other models, datasets, or clinical use.","The abstract reports that calibration and net benefit were not consistently more favorable, so comparable AUROCs alone do not establish comparable clinical value."],"mediaId":"67fbe957d652e39b266d21fe8a4221afccef5c3f9da7a64b2e5ff596dcbab29d","correctionReason":null,"id":"0a01eb10baec3cbfd34a62fbf950daaf0a9678964bbf462b49d69a3c73c1768c","publicationId":"2a0f11a1-0480-4534-8082-229a97ef3f34","version":1,"sourceUrl":"https://arxiv.org/abs/2611.10484","sourceTitle":"Can General-Purpose Generative AI Supply Training Data for Medical Vision Models? A Multimodal Retrospective Study","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-11-07","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-10T08:42:32.173Z","createdAt":"2026-11-12T09:03:44.716Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"67fbe957d652e39b266d21fe8a4221afccef5c3f9da7a64b2e5ff596dcbab29d","kind":"article","url":"https://arxiv.org/abs/2611.10484","sourceUrl":"https://arxiv.org/abs/2611.10484","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-12T09:02:51.141Z","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/2611.10484","checkedAt":"2026-11-12T09:34:11.624Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v6","mediaId":"67fbe957d652e39b266d21fe8a4221afccef5c3f9da7a64b2e5ff596dcbab29d","confidence":0.9,"provider":"typesafe","model":"jev-1.13.0","inputHash":"1a0dae684f926f6083138a26cfc16ee386a002159908fe1ac300fb082444f5a6","decidedAt":"2026-11-12T09:02:53.160Z"},"provenance":{"textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observationId":"fefde60808a6f0ea3016aaf2d867ae508b80c8312f9026b19afe348cda27bd76","retrievedAt":"2026-11-12T09:02:50.853Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"8a4eaa5c-bdd9-47a4-8d61-3d06432e668f"},"reviewer":{"model":"gpt-6-luna","runId":"f84e2702-511f-403f-8182-7df19a62f3b1"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-12T09:03:44.716Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Peng, Jiachuan","url":null,"kind":"unknown"},{"name":"Lam, Kyle","url":null,"kind":"unknown"},{"name":"Qu, Jierui","url":null,"kind":"unknown"},{"name":"Li, Lin","url":null,"kind":"unknown"},{"name":"Zheng, Haining","url":null,"kind":"unknown"},{"name":"Zheng, Bang","url":null,"kind":"unknown"},{"name":"Agarwal, Aniruddha","url":null,"kind":"unknown"},{"name":"Almallah, Zaki","url":null,"kind":"unknown"},{"name":"Chen, Min","url":null,"kind":"unknown"},{"name":"Grau, Vicente","url":null,"kind":"unknown"},{"name":"Rossman, Hagai","url":null,"kind":"unknown"},{"name":"Segal, Eran","url":null,"kind":"unknown"},{"name":"Qiu, Jianing","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Peng, Jiachuan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Peng, Jiachuan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Lam, Kyle","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lam, Kyle","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Qu, Jierui","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Qu, Jierui","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Li, Lin","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Li, Lin","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Zheng, Haining","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zheng, Haining","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Zheng, Bang","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zheng, Bang","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Agarwal, Aniruddha","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Agarwal, Aniruddha","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:6","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Almallah, Zaki","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Almallah, Zaki","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:7","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Chen, Min","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Chen, Min","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:8","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Grau, Vicente","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Grau, Vicente","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:9","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Rossman, Hagai","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Rossman, Hagai","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:10","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Segal, Eran","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Segal, Eran","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:11","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}},{"actor":{"name":"Qiu, Jianing","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Qiu, Jianing","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.10484","locator":"head/meta[name=author|citation_author]"},"assertionId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8:participation:12","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-12T09:34:11.624Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","sourceUrl":"https://arxiv.org/abs/2611.10484","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z"}}],"participationCoverage":{"total":13,"returned":13,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.10484","snapshotId":"f0b42837c7e6ce32bdec49c3fe2e8ec1b0d78d19b084974e2af5e778de81c7a8","contentSha256":"02a62f6bf4be55f2444a387f4954a7851c9eca1d020799e7951c3e7b040aeee9","textHash":"19108ac5e5ce9c1d33cec2930c90a5c8b53978375ecbedaa9a3f5edc2f7d0248","observedAt":"2026-11-12T09:34:11.624Z","basis":"publisher-page-metadata","sourceLinks":52,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/0a01eb10baec3cbfd34a62fbf950daaf0a9678964bbf462b49d69a3c73c1768c/source"}}}],"audits":[{"publicationId":"2a0f11a1-0480-4534-8082-229a97ef3f34","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The findings and conclusions are attributed to the paper's authors and framed as reported results. The arXiv cs.AI label matches the feed publisher; the paper is primarily cs.CV and cross-listed to cs.AI, which does not misattribute it.","excerpt":"Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)"},{"dimension":"summary","status":"uncertain","explanation":"Central terms go unexplained for a general reader: AUROC, discrimination, calibration, net benefit and ordinal vascular age index. Readers cannot see why comparable AUROC with weaker calibration matters. The fundus result also drops its stated purpose, a shortcut-learning stress test, so the probability drop reads as unexplained. The motivating constraint, scarce annotated data, is also missing. Smallest repairs: add tooltips or plain-language glosses, state the shortcut-learning purpose, and note the data-scarcity motivation. The question compares calibration with discrimination; asking whether synthetic-only models match real-data models on these measures would be clearer.","excerpt":"To stress-test shortcut learning, editing a moderate tessellated pattern into fundus photographs"},{"dimension":"date","status":"pass","explanation":"originalPublished 2026-11-07 matches the v1 submission date. It is not the later feed date.","excerpt":"[Submitted on 7 Nov 2026]"},{"dimension":"support","status":"pass","explanation":"Every claim is directly supported by the abstract, and the excerpts are exact substrings. The limitations stay within the evidence. No benefits or costs are invented, and the authors' qualified conclusion is preserved.","excerpt":"comparable to real-data-trained models, but their calibration and net benefit were not consistently more favorable"},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct arXiv abstract page for this paper, which is the most direct way to reach the source.","excerpt":null}],"createdAt":"2026-11-12T09:03:44.716Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}