{"paper":{"title":"How to \"train\" a surgeon? Benchmarking generalist agents in surgical scene understanding","arxiv_id":"2611.11437","primary_cat":"cs.CV","submitted_at":"2026-11-08T15:00:44+00:00","version_compiled":"v1"},"notice":"Every quoted sentence is the authors' text from the arXiv listing abstract. Labels and probabilities are the outputs of classifier-1.13.0 on the question bank abstract-p1-20260923, exactly as returned. No sentence in this record was written by a language model.","schema":"pith.compiled-record/v0","status":"compiled","abstract":{"source":"arXiv listing abstract v1","sentences":[{"i":0,"p":0.61,"id":"s0","own":false,"role":"gap","text":"Evaluating general-purpose multimodal large language models (MLLMs) for surgical scene understanding remains limited due to scarce surgical data and fragmented evaluation protocols.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.06,"main_claim_p":0.02,"quantitative":false,"quantitative_p":0.11,"role_probabilities":{"gap":0.61,"method":0.0,"result":0.0,"background":0.39,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":1,"p":0.78,"id":"s1","own":true,"role":"contribution","text":"We address these challenges through a two-stage evaluation framework for surgical scene understanding.","hedge":{"p":0.98,"label":"asserted"},"own_p":0.83,"main_claim_p":0.03,"quantitative":false,"quantitative_p":0.13,"role_probabilities":{"gap":0.0,"method":0.22,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.78,"interpretation":0.0}},{"i":2,"p":0.55,"id":"s2","own":true,"role":"method","text":"In Stage I, we introduce SurgKnowBench, a diagnostic benchmark that harmonizes public surgical datasets across procedures and tasks and evaluates proprietary and open-source MLLMs to map the boundary of pretrained surgical competence.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.93,"main_claim_p":0.04,"quantitative":false,"quantitative_p":0.13,"role_probabilities":{"gap":0.0,"method":0.55,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.45,"interpretation":0.0}},{"i":3,"p":0.98,"id":"s3","own":true,"role":"result","text":"SurgKnowBench reveals that current models retain partial visual priors but remain weak on relational and procedural understanding and workflow assessment.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.92,"main_claim_p":0.17,"quantitative":false,"quantitative_p":0.38,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.98,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.02}},{"i":4,"p":0.69,"id":"s4","own":true,"role":"contribution","text":"Building on this failure frontier and motivated by the scaffolded nature of surgical training, we introduce SurgAgentGym, an agentic environment where general-purpose MLLM agents can use bounded knowledge search over SurgWiki and general-purpose visual grounding for surgical tasks.","hedge":{"p":0.95,"label":"asserted"},"own_p":0.82,"main_claim_p":0.04,"quantitative":false,"quantitative_p":0.29,"role_probabilities":{"gap":0.0,"method":0.31,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.69,"interpretation":0.0}},{"i":5,"p":0.56,"id":"s5","own":true,"role":"method","text":"SurgAgentGym evaluates whether generalist agents can improve surgical understanding at inference time through tool-mediated evidence use, without additional surgical training.","hedge":{"p":0.39,"label":"asserted"},"own_p":0.53,"main_claim_p":0.07,"quantitative":false,"quantitative_p":0.1,"role_probabilities":{"gap":0.0,"method":0.56,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.44,"interpretation":0.0}},{"i":6,"p":0.94,"id":"s6","own":true,"role":"result","text":"Our results show that agentic inference narrows the gap mainly for perception-heavy tasks, while fine-grained localization and expert-style assessment remain persistent bottlenecks.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.96,"main_claim_p":0.7,"quantitative":true,"quantitative_p":0.68,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.94,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.06}},{"i":7,"p":0.99,"id":"s7","own":false,"role":"future_work","text":"This opens a path toward future studies of agentic surgical understanding.","hedge":{"p":0.72,"label":"hedged_possibility"},"own_p":0.09,"main_claim_p":0.05,"quantitative":false,"quantitative_p":0.03,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.99,"contribution":0.0,"interpretation":0.01}}]},"compiled":{"main_claim":{"p":1.0,"text":"Our results show that agentic inference narrows the gap mainly for perception-heavy tasks, while fine-grained localization and expert-style assessment remain persistent bottlenecks.","location":"abstract s6","hedge_strength":{"p":1.0,"label":"asserted"}},"provenance":{"route":{"via":"TypeSafe API","model":"classifier-1.13.0","endpoint":"https://api.typesafe.ai/v1/systemone","provider":"TypeSafe"},"packing":"one request per paper","requests":1,"input_tokens":6259,"cost_usd_reported":0.0},"compiled_at":"2026-11-10","bank_version":"abstract-p1-20260923","model_version":"classifier-1.13.0","abstract_states":{"claims_first":0.09,"states_limitation":0.92,"human_participants":0.1,"controlled_comparison":0.5,"randomized_assignment":0.01,"states_quantitative_result":0.05,"compares_with_prior_results":0.75,"mentions_code_or_data_availability":0.08},"evidential_basis":{"p":0.95,"value":"observation or experiment","probabilities":{"argument":0.0,"literature":0.0,"proof_or_derivation":0.0,"analysis_of_existing_data":0.01,"observation_or_experiment":0.95,"simulation_or_computation":0.04}},"contribution_type":{"p":0.86,"value":"dataset or benchmark","probabilities":{"review":0.0,"method_or_tool":0.11,"empirical_study":0.03,"simulation_study":0.0,"theoretical_result":0.0,"dataset_or_benchmark":0.86,"position_or_commentary":0.0}},"contribution_statement":{"p":0.44,"text":"We address these challenges through a two-stage evaluation framework for surgical scene understanding.","location":"abstract s1"}},"coverage":{"tier":"abstract","levels_run":["P abstract"],"not_examined":["full text","figures and tables","references and citing sentences","relations to other papers"],"sources_read":["arXiv listing abstract v1"],"levels_not_run":["S sections","T body sentences","C citations","R cross-paper relations"],"sentences_examined":8,"sentences_in_abstract":8},"measured_accuracy":{"url":"https://pith.science/accuracy","measured_on":"2026-10-02","method":"two independent model families labelled a fresh random sample blind; the figure is how often the published label matches their consensus","sentence_role_agreement":0.92,"main_claim_agreement":0.908,"typed_citation_agreement":0.933,"typed_citation_agreement_95_interval":[0.917,0.945],"review_comments_accurate":0.758,"review_comments_accurate_95_interval":[0.684,0.82],"review_caught_admitted_error_found_or_related":0.448,"review_comments_judged":{"judged_by":["gpt-5.6-luna","grok-4.7"],"judged_against":"Pith's own text copy of each paper: text extracted from the PDF, the first 90,000 characters","measured_on":"2026-10-04","accurate":113,"decided":149,"sampled_reviews":200,"split_by_lane":false,"by_lane":"https://pith.science/accuracy#by-lane"}}}