{"id":"https://openalex.org/W4412945732","doi":"https://doi.org/10.18653/v1/2025.acl-long.266","title":"Pattern Recognition or Medical Knowledge? The Problem with Multiple-Choice Questions in Medicine","display_name":"Pattern Recognition or Medical Knowledge? The Problem with Multiple-Choice Questions in Medicine","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412945732","doi":"https://doi.org/10.18653/v1/2025.acl-long.266"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.266","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.266","pdf_url":"https://aclanthology.org/2025.acl-long.266.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.acl-long.266.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5098787776","display_name":"Maxime Griot","orcid":"https://orcid.org/0009-0000-8180-4737"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Maxime Griot","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5043868402","display_name":"Jean Vanderdonckt","orcid":"https://orcid.org/0000-0003-3275-3333"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jean Vanderdonckt","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5042651201","display_name":"Demet Y\u00fcksel","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Demet Yuksel","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5102777591","display_name":"Coralie Hemptinne","orcid":"https://orcid.org/0000-0003-0717-3804"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Coralie Hemptinne","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":2.5432,"has_fulltext":true,"cited_by_count":5,"citation_normalized_percentile":{"value":0.89724378,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":96,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"5321","last_page":"5341"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11710","display_name":"Biomedical Text Mining and Ontologies","score":0.4562999904155731,"subfield":{"id":"https://openalex.org/subfields/1312","display_name":"Molecular Biology"},"field":{"id":"https://openalex.org/fields/13","display_name":"Biochemistry, Genetics and Molecular Biology"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},"topics":[{"id":"https://openalex.org/T11710","display_name":"Biomedical Text Mining and Ontologies","score":0.4562999904155731,"subfield":{"id":"https://openalex.org/subfields/1312","display_name":"Molecular Biology"},"field":{"id":"https://openalex.org/fields/13","display_name":"Biochemistry, Genetics and Molecular Biology"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6230877637863159},{"id":"https://openalex.org/keywords/medical-knowledge","display_name":"Medical knowledge","score":0.4760054051876068},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.38541510701179504},{"id":"https://openalex.org/keywords/knowledge-management","display_name":"Knowledge management","score":0.37200334668159485},{"id":"https://openalex.org/keywords/data-science","display_name":"Data science","score":0.36189889907836914},{"id":"https://openalex.org/keywords/medical-education","display_name":"Medical education","score":0.25578421354293823},{"id":"https://openalex.org/keywords/medicine","display_name":"Medicine","score":0.204928457736969}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6230877637863159},{"id":"https://openalex.org/C2985722590","wikidata":"https://www.wikidata.org/wiki/Q11190","display_name":"Medical knowledge","level":2,"score":0.4760054051876068},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.38541510701179504},{"id":"https://openalex.org/C56739046","wikidata":"https://www.wikidata.org/wiki/Q192060","display_name":"Knowledge management","level":1,"score":0.37200334668159485},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.36189889907836914},{"id":"https://openalex.org/C509550671","wikidata":"https://www.wikidata.org/wiki/Q126945","display_name":"Medical education","level":1,"score":0.25578421354293823},{"id":"https://openalex.org/C71924100","wikidata":"https://www.wikidata.org/wiki/Q11190","display_name":"Medicine","level":0,"score":0.204928457736969}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.266","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.266","pdf_url":"https://aclanthology.org/2025.acl-long.266.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},{"id":"pmh:oai:dial.uclouvain.be:boreal:302262","is_oa":true,"landing_page_url":"http://hdl.handle.net/2078.1/302262","pdf_url":null,"source":{"id":"https://openalex.org/S4306400166","display_name":"Digital Access to Libraries","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I2800384882","host_organization_name":"Harris County Public Library","host_organization_lineage":["https://openalex.org/I2800384882"],"host_organization_lineage_names":[],"type":"repository"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"info:eu-repo/semantics/conferenceObject"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.acl-long.266","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.266","pdf_url":"https://aclanthology.org/2025.acl-long.266.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320321584","display_name":"F\u00e9d\u00e9ration Wallonie-Bruxelles","ror":"https://ror.org/04q01pf08"},{"id":"https://openalex.org/F4320326798","display_name":"Fondation Saint Luc","ror":null}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412945732.pdf","grobid_xml":"https://content.openalex.org/works/W4412945732.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"such":[4,28],"as":[5],"Chat-GPT":[6],"demonstrate":[7],"significant":[8],"potential":[9],"in":[10,70,84,108,112],"the":[11,26,55,88,132,149],"medical":[12,48,103],"domain":[13],"and":[14,39,68,72,81,114,127,147],"are":[15],"often":[16],"evaluated":[17,78],"using":[18,74],"multiple-choice":[19],"questions":[20],"(MCQs)":[21],"modeled":[22],"on":[23,51,122],"exams":[24],"like":[25],"USMLE.However,":[27],"benchmarks":[29],"may":[30,141],"overestimate":[31],"true":[32],"clinical":[33,145],"understanding":[34],"by":[35],"rewarding":[36],"pattern":[37],"recognition":[38],"test-taking":[40,125],"heuristics.To":[41],"investigate":[42],"this,":[43],"we":[44],"created":[45],"a":[46,85],"fictional":[47,89],"benchmark":[49],"centered":[50],"an":[52,93],"imaginary":[53],"organ,":[54],"Glianorex,":[56],"allowing":[57],"us":[58],"to":[59,130],"separate":[60],"memorized":[61],"knowledge":[62],"from":[63],"reasoning":[64,129,146],"ability.We":[65],"generated":[66],"textbooks":[67],"MCQs":[69],"English":[71,109],"French":[73],"leading":[75],"LLMs,":[76],"then":[77],"proprietary,":[79],"open-source,":[80],"domain-specific":[82],"models":[83,91,104,107,119],"zero-shot":[86],"setting.Despite":[87],"content,":[90],"achieved":[92],"average":[94],"score":[95],"of":[96],"64%,":[97],"while":[98],"physicians":[99],"scored":[100],"only":[101],"27%.Fine-tuned":[102],"outperformed":[105],"base":[106],"but":[110],"not":[111,142],"French.Ablation":[113],"interpretability":[115],"analyses":[116],"revealed":[117],"that":[118,137],"frequently":[120],"relied":[121],"shallow":[123],"cues,":[124],"strategies,":[126],"hallucinated":[128],"identify":[131],"correct":[133],"choice.These":[134],"results":[135],"suggest":[136],"standard":[138],"MCQ-based":[139],"evaluations":[140],"effectively":[143],"measure":[144],"highlight":[148],"need":[150],"for":[151,158],"more":[152],"robust,":[153],"clinically":[154],"meaningful":[155],"assessment":[156],"methods":[157],"LLMs.":[159]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":3}],"updated_date":"2026-07-02T09:51:11.867554","created_date":"2025-10-10T00:00:00"}