{"id":"https://openalex.org/W7153007385","doi":"https://doi.org/10.48550/arxiv.2604.08333","title":"Lost in the Hype: Revealing and Dissecting the Performance Degradation of Medical Multimodal Large Language Models in Image Classification","display_name":"Lost in the Hype: Revealing and Dissecting the Performance Degradation of Medical Multimodal Large Language Models in Image Classification","publication_year":2026,"publication_date":"2026-04-09","ids":{"openalex":"https://openalex.org/W7153007385","doi":"https://doi.org/10.48550/arxiv.2604.08333"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.08333","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.08333","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.08333","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5133381947","display_name":"Xun Zhu","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Zhu, Xun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5078888279","display_name":"Fan Mo","orcid":"https://orcid.org/0000-0001-6272-781X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mo, Fanbin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133346417","display_name":"Xi Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Xi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133384949","display_name":"Kaili Zheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Kaili","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133323227","display_name":"Shaoshuai Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Shaoshuai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133334187","display_name":"Yiming Shi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shi, Yiming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133331442","display_name":"Jian Gao","orcid":"https://orcid.org/0009-0005-0754-9519"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gao, Jian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100677700","display_name":"Miao Li","orcid":"https://orcid.org/0000-0001-6196-2041"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Miao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5133317594","display_name":"Ji Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Ji","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":9,"corresponding_author_ids":["https://openalex.org/A5133381947"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11636","display_name":"Artificial Intelligence in Healthcare and Education","score":0.2842000126838684,"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"}},"topics":[{"id":"https://openalex.org/T11636","display_name":"Artificial Intelligence in Healthcare and Education","score":0.2842000126838684,"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"}},{"id":"https://openalex.org/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.22509999573230743,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.11180000007152557,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.5985000133514404},{"id":"https://openalex.org/keywords/field","display_name":"Field (mathematics)","score":0.5845000147819519},{"id":"https://openalex.org/keywords/comprehension","display_name":"Comprehension","score":0.5521000027656555},{"id":"https://openalex.org/keywords/visualization","display_name":"Visualization","score":0.49549999833106995},{"id":"https://openalex.org/keywords/semantics","display_name":"Semantics (computer science)","score":0.4675999879837036},{"id":"https://openalex.org/keywords/fidelity","display_name":"Fidelity","score":0.4571000039577484},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.4553000032901764},{"id":"https://openalex.org/keywords/feature-extraction","display_name":"Feature extraction","score":0.3644999861717224}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6798999905586243},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.5985000133514404},{"id":"https://openalex.org/C9652623","wikidata":"https://www.wikidata.org/wiki/Q190109","display_name":"Field (mathematics)","level":2,"score":0.5845000147819519},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5526999831199646},{"id":"https://openalex.org/C511192102","wikidata":"https://www.wikidata.org/wiki/Q5156948","display_name":"Comprehension","level":2,"score":0.5521000027656555},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.49549999833106995},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.4675999879837036},{"id":"https://openalex.org/C2776459999","wikidata":"https://www.wikidata.org/wiki/Q2119376","display_name":"Fidelity","level":2,"score":0.4571000039577484},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.4553000032901764},{"id":"https://openalex.org/C52622490","wikidata":"https://www.wikidata.org/wiki/Q1026626","display_name":"Feature extraction","level":2,"score":0.3644999861717224},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.353300005197525},{"id":"https://openalex.org/C2779679103","wikidata":"https://www.wikidata.org/wiki/Q5251805","display_name":"Degradation (telecommunications)","level":2,"score":0.3513000011444092},{"id":"https://openalex.org/C75294576","wikidata":"https://www.wikidata.org/wiki/Q5165192","display_name":"Contextual image classification","level":3,"score":0.3488999903202057},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.3361999988555908},{"id":"https://openalex.org/C113364801","wikidata":"https://www.wikidata.org/wiki/Q26674","display_name":"High fidelity","level":2,"score":0.3230000138282776},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3111000061035156},{"id":"https://openalex.org/C31601959","wikidata":"https://www.wikidata.org/wiki/Q931309","display_name":"Medical imaging","level":2,"score":0.30660000443458557},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.2924000024795532},{"id":"https://openalex.org/C172367668","wikidata":"https://www.wikidata.org/wiki/Q6504956","display_name":"Data visualization","level":3,"score":0.2915000021457672},{"id":"https://openalex.org/C2777561058","wikidata":"https://www.wikidata.org/wiki/Q2652119","display_name":"Program comprehension","level":4,"score":0.275299996137619},{"id":"https://openalex.org/C2780878386","wikidata":"https://www.wikidata.org/wiki/Q1659648","display_name":"Visual language","level":2,"score":0.2712000012397766},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.25699999928474426}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.08333","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.08333","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.08333","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.08333","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.5458816289901733}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0],"rise":[1],"of":[2,13,18,25,109,123,173,185],"multimodal":[3],"large":[4],"language":[5],"models":[6],"(MLLMs)":[7],"has":[8],"sparked":[9],"an":[10],"unprecedented":[11],"wave":[12],"applications":[14],"in":[15,58,143,155,161,167],"the":[16,26,73,106,116,135,183,203,225,228],"field":[17],"medical":[19,36,44,87,144,209],"imaging":[20],"analysis.":[21],"However,":[22],"as":[23],"one":[24],"earliest":[27],"and":[28,61,113,125,170,194,239],"most":[29],"fundamental":[30],"tasks":[31],"integrated":[32],"into":[33],"this":[34,78],"paradigm,":[35],"image":[37,92],"classification":[38,93,127,140],"reveals":[39],"a":[40,67],"sobering":[41],"reality:":[42],"state-of-the-art":[43],"MLLMs":[45,88,193,210,236],"consistently":[46],"underperform":[47],"compared":[48],"to":[49,104,138,233],"traditional":[50],"deep":[51],"learning":[52],"models,":[53],"despite":[54],"their":[55,213],"overwhelming":[56],"advantages":[57],"pre-training":[59],"data":[60],"model":[62],"parameters.":[63],"This":[64],"paradox":[65],"prompts":[66],"critical":[68,204],"rethinking:":[69],"where":[70,124],"exactly":[71],"does":[72],"performance":[74,98,141],"degradation":[75,142],"originate?":[76],"In":[77],"paper,":[79],"we":[80,100,177,197],"conduct":[81],"extensive":[82],"experiments":[83],"on":[84,202],"14":[85],"open-source":[86],"across":[89,191],"three":[90],"representative":[91],"datasets.":[94,195],"Moving":[95],"beyond":[96],"superficial":[97],"benchmarking,":[99],"employ":[101],"feature":[102,186],"probing":[103],"track":[105],"information":[107],"flow":[108],"visual":[110,156],"features":[111],"module-by-module":[112],"layer-by-layer":[114],"throughout":[115],"entire":[117],"MLLM":[118],"pipeline,":[119],"enabling":[120,188],"explicit":[121],"visualization":[122],"how":[126],"signals":[128],"are":[129],"distorted,":[130],"diluted,":[131],"or":[132],"overridden.":[133],"As":[134],"first":[136],"attempt":[137],"dissect":[139],"MLLMs,":[145],"our":[146,220],"findings":[147],"reveal":[148],"four":[149],"failure":[150],"modes:":[151],"1)":[152],"quality":[153],"limitation":[154],"representation,":[157],"2)":[158],"fidelity":[159],"loss":[160],"connector":[162],"projection,":[163],"3)":[164],"comprehension":[165],"deficit":[166],"LLM":[168],"reasoning,":[169],"4)":[171],"misalignment":[172],"semantic":[174],"mapping.":[175],"Meanwhile,":[176],"introduce":[178],"quantitative":[179],"scores":[180],"that":[181,206,219,227],"characterize":[182],"healthiness":[184],"evolution,":[187],"principled":[189],"comparisons":[190],"diverse":[192],"Furthermore,":[196],"provide":[198],"insightful":[199],"discussions":[200],"centered":[201],"barriers":[205],"prevent":[207],"current":[208],"from":[211,230],"fulfilling":[212],"promised":[214],"clinical":[215],"potential.":[216],"We":[217],"hope":[218],"work":[221],"provokes":[222],"rethinking":[223],"within":[224],"community-highlighting":[226],"road":[229],"high":[231],"expectations":[232],"clinically":[234],"deployable":[235],"remains":[237],"long":[238],"winding.":[240]},"counts_by_year":[],"updated_date":"2026-04-29T09:16:38.111599","created_date":"2026-04-11T00:00:00"}