{"id":"https://openalex.org/W4416037839","doi":"https://doi.org/10.18653/v1/2025.arabicnlp-sharedtasks.19","title":"NYUAD at AraHealthQA Shared Task: Benchmarking the Medical Understanding and Reasoning of Large Language Models in Arabic Healthcare Tasks","display_name":"NYUAD at AraHealthQA Shared Task: Benchmarking the Medical Understanding and Reasoning of Large Language Models in Arabic Healthcare Tasks","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416037839","doi":"https://doi.org/10.18653/v1/2025.arabicnlp-sharedtasks.19"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.arabicnlp-sharedtasks.19","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.arabicnlp-sharedtasks.19","pdf_url":"https://aclanthology.org/2025.arabicnlp-sharedtasks.19.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 Third Arabic Natural Language Processing Conference: Shared Tasks","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.arabicnlp-sharedtasks.19.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5037300629","display_name":"Nouar AlDahoul","orcid":"https://orcid.org/0000-0001-5522-0033"},"institutions":[{"id":"https://openalex.org/I120250893","display_name":"New York University Abu Dhabi","ror":"https://ror.org/00e5k0821","country_code":"AE","type":"education","lineage":["https://openalex.org/I120250893","https://openalex.org/I57206974"]}],"countries":["AE"],"is_corresponding":false,"raw_author_name":"Nouar AlDahoul","raw_affiliation_strings":["Computer Science Department New York University Abu Dhabi , UAE"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Computer Science Department New York University Abu Dhabi , UAE","institution_ids":["https://openalex.org/I120250893"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5018129441","display_name":"Yasir Zaki","orcid":"https://orcid.org/0000-0001-8078-6944"},"institutions":[{"id":"https://openalex.org/I120250893","display_name":"New York University Abu Dhabi","ror":"https://ror.org/00e5k0821","country_code":"AE","type":"education","lineage":["https://openalex.org/I120250893","https://openalex.org/I57206974"]}],"countries":["AE"],"is_corresponding":false,"raw_author_name":"Yasir Zaki","raw_affiliation_strings":["Computer Science Department New York University Abu Dhabi , UAE"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Computer Science Department New York University Abu Dhabi , UAE","institution_ids":["https://openalex.org/I120250893"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I120250893"],"apc_list":null,"apc_paid":null,"fwci":1.673,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":{"value":0.87326786,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":94,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"119","last_page":"125"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T13702","display_name":"Machine Learning in Healthcare","score":0.296099990606308,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T13702","display_name":"Machine Learning in Healthcare","score":0.296099990606308,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.18700000643730164,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11636","display_name":"Artificial Intelligence in Healthcare and Education","score":0.14079999923706055,"subfield":{"id":"https://openalex.org/subfields/2718","display_name":"Health Informatics"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/benchmarking","display_name":"Benchmarking","score":0.6600000262260437},{"id":"https://openalex.org/keywords/health-care","display_name":"Health care","score":0.593999981880188},{"id":"https://openalex.org/keywords/unified-medical-language-system","display_name":"Unified Medical Language System","score":0.3668999969959259},{"id":"https://openalex.org/keywords/arabic","display_name":"Arabic","score":0.35670000314712524},{"id":"https://openalex.org/keywords/comprehension","display_name":"Comprehension","score":0.335099995136261},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.32580000162124634},{"id":"https://openalex.org/keywords/patient-care","display_name":"Patient care","score":0.31790000200271606},{"id":"https://openalex.org/keywords/medline","display_name":"MEDLINE","score":0.2985000014305115}],"concepts":[{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.6600000262260437},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6141999959945679},{"id":"https://openalex.org/C160735492","wikidata":"https://www.wikidata.org/wiki/Q31207","display_name":"Health care","level":2,"score":0.593999981880188},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.46000000834465027},{"id":"https://openalex.org/C56739046","wikidata":"https://www.wikidata.org/wiki/Q192060","display_name":"Knowledge management","level":1,"score":0.3783999979496002},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.37770000100135803},{"id":"https://openalex.org/C69505689","wikidata":"https://www.wikidata.org/wiki/Q455338","display_name":"Unified Medical Language System","level":2,"score":0.3668999969959259},{"id":"https://openalex.org/C96455323","wikidata":"https://www.wikidata.org/wiki/Q13955","display_name":"Arabic","level":2,"score":0.35670000314712524},{"id":"https://openalex.org/C511192102","wikidata":"https://www.wikidata.org/wiki/Q5156948","display_name":"Comprehension","level":2,"score":0.335099995136261},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.32580000162124634},{"id":"https://openalex.org/C2989236134","wikidata":"https://www.wikidata.org/wiki/Q31207","display_name":"Patient care","level":2,"score":0.31790000200271606},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.30480000376701355},{"id":"https://openalex.org/C2779473830","wikidata":"https://www.wikidata.org/wiki/Q1540899","display_name":"MEDLINE","level":2,"score":0.2985000014305115},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.2867000102996826},{"id":"https://openalex.org/C2777601683","wikidata":"https://www.wikidata.org/wiki/Q6499736","display_name":"Vocabulary","level":2,"score":0.26969999074935913},{"id":"https://openalex.org/C145642194","wikidata":"https://www.wikidata.org/wiki/Q870895","display_name":"Health informatics","level":3,"score":0.26750001311302185},{"id":"https://openalex.org/C155092808","wikidata":"https://www.wikidata.org/wiki/Q182557","display_name":"Computational linguistics","level":2,"score":0.26510000228881836},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.26159998774528503},{"id":"https://openalex.org/C2986740045","wikidata":"https://www.wikidata.org/wiki/Q31207","display_name":"Health services","level":3,"score":0.26010000705718994},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.2581999897956848},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2563000023365021},{"id":"https://openalex.org/C161301231","wikidata":"https://www.wikidata.org/wiki/Q3478658","display_name":"Knowledge representation and reasoning","level":2,"score":0.2540999948978424}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.arabicnlp-sharedtasks.19","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.arabicnlp-sharedtasks.19","pdf_url":"https://aclanthology.org/2025.arabicnlp-sharedtasks.19.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 Third Arabic Natural Language Processing Conference: Shared Tasks","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.arabicnlp-sharedtasks.19","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.arabicnlp-sharedtasks.19","pdf_url":"https://aclanthology.org/2025.arabicnlp-sharedtasks.19.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 Third Arabic Natural Language Processing Conference: Shared Tasks","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416037839.pdf","grobid_xml":"https://content.openalex.org/works/W4416037839.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recent":[0],"progress":[1],"in":[2,11,21,43,64,70,88,101,113,121,136,176,197],"large":[3],"language":[4,15],"models":[5,154],"(LLMs)":[6],"has":[7,26],"showcased":[8],"impressive":[9],"proficiency":[10],"numerous":[12],"Arabic":[13,22,53,66,137],"natural":[14],"processing":[16],"(NLP)":[17],"applications.Nevertheless,":[18],"their":[19,46,78],"effectiveness":[20],"medical":[23,54,61],"NLP":[24,67],"domains":[25],"received":[27],"limited":[28],"investigation.This":[29],"research":[30],"examines":[31],"the":[32,65,97,129,146,177,185],"degree":[33],"to":[34,80,168,193],"which":[35],"state-of-the-art":[36],"LLMs":[37,58,74,100,135,190],"demonstrate":[38,194],"and":[39,92,118,131,161,171,202],"articulate":[40],"healthcare":[41],"knowledge":[42],"Arabic,":[44],"assessing":[45],"capabilities":[47],"across":[48],"a":[49,60,204],"varied":[50],"array":[51],"of":[52,99,124,133,199,207],"tasks.We":[55],"benchmark":[56],"several":[57,189],"using":[59],"dataset":[62],"proposed":[63,147],"AraHealthQA":[68],"challenge":[69,178],"MedArabiQ2025":[71],"track.Various":[72],"base":[73,153],"were":[75,191],"assessed":[76],"on":[77],"ability":[79],"accurately":[81],"provide":[82],"correct":[83,114],"answers":[84],"from":[85],"existing":[86],"choices":[87],"multiple-choice":[89],"questions":[90,104,187],"(MCQs)":[91],"fill-in-the-blank":[93],"scenarios.Additionally,":[94],"we":[95],"evaluated":[96],"capacity":[98],"answering":[102],"open-ended":[103,186],"aligned":[105],"with":[106],"expert":[107],"answers.Our":[108],"results":[109],"reveal":[110],"significant":[111],"variations":[112,120],"answer":[115],"prediction":[116],"accuracy":[117,170],"low":[119],"semantic":[122,200],"alignment":[123,201],"generated":[125],"answers,":[126],"highlighting":[127],"both":[128],"potential":[130],"limitations":[132],"current":[134],"clinical":[138],"contexts.Our":[139],"analysis":[140],"shows":[141],"that":[142],"for":[143,184],"MCQs":[144],"task,":[145,188],"majority":[148],"voting":[149],"solution,":[150],"leveraging":[151],"three":[152],"(Gemini":[155],"Flash":[156],"2.5,":[157,160],"Gemini":[158],"Pro":[159],"GPT":[162],"o3),":[163],"outperforms":[164],"others,":[165],"achieving":[166],"up":[167],"77%":[169],"securing":[172],"first":[173],"place":[174],"overall":[175],"1":[179],"(Alhuzali":[180],"et":[181],"al.,":[182],"2025).Moreover,":[183],"able":[192],"excellent":[195],"performance":[196],"terms":[198],"achieve":[203],"maximum":[205],"BERTScore":[206],"86.44%.":[208]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-11-08T00:00:00"}