{"id":"https://openalex.org/W7164314040","doi":"https://doi.org/10.48550/arxiv.2606.12407","title":"How Seemingly Inconsequential Design Choices Dictate Performance of LLMs in Pathology","display_name":"How Seemingly Inconsequential Design Choices Dictate Performance of LLMs in Pathology","publication_year":2026,"publication_date":"2026-06-10","ids":{"openalex":"https://openalex.org/W7164314040","doi":"https://doi.org/10.48550/arxiv.2606.12407"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.12407","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.12407","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.12407","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5030384719","display_name":"Kian R. Weihrauch","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Weihrauch, Kian R.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138473617","display_name":"Thomas A. Buckley","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Buckley, Thomas A.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138412568","display_name":"William Lotter","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lotter, William","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5136059686","display_name":"Arjun K. Manrai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Manrai, Arjun K.","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":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/T10862","display_name":"AI in cancer detection","score":0.9606000185012817,"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/T10862","display_name":"AI in cancer detection","score":0.9606000185012817,"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":0.005400000140070915,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.004000000189989805,"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/context","display_name":"Context (archaeology)","score":0.6973000168800354},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5863999724388123},{"id":"https://openalex.org/keywords/adaptation","display_name":"Adaptation (eye)","score":0.49399998784065247},{"id":"https://openalex.org/keywords/perception","display_name":"Perception","score":0.4562000036239624},{"id":"https://openalex.org/keywords/perspective","display_name":"Perspective (graphical)","score":0.2842000126838684}],"concepts":[{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.6973000168800354},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5863999724388123},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.49399998784065247},{"id":"https://openalex.org/C180747234","wikidata":"https://www.wikidata.org/wiki/Q23373","display_name":"Cognitive psychology","level":1,"score":0.460999995470047},{"id":"https://openalex.org/C26760741","wikidata":"https://www.wikidata.org/wiki/Q160402","display_name":"Perception","level":2,"score":0.4562000036239624},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.39309999346733093},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.3880999982357025},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.35179999470710754},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.2842000126838684},{"id":"https://openalex.org/C77805123","wikidata":"https://www.wikidata.org/wiki/Q161272","display_name":"Social psychology","level":1,"score":0.2750000059604645},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2639000117778778},{"id":"https://openalex.org/C2779816988","wikidata":"https://www.wikidata.org/wiki/Q305418","display_name":"Abandonment (legal)","level":2,"score":0.2578999996185303}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.12407","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.12407","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.12407","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.12407","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"General-purpose":[0],"large":[1],"language":[2],"models":[3,14,108,172],"(LLMs)":[4],"are":[5],"routinely":[6,28],"used":[7],"as":[8,47],"baselines":[9,27],"when":[10],"evaluating":[11],"specialized":[12,59,107],"pathology":[13,73],"on":[15,140,149],"whole-slide":[16],"images":[17],"(WSIs).":[18],"Because":[19],"WSIs":[20],"exceed":[21],"contemporary":[22],"model":[23],"context":[24],"limits,":[25],"LLM":[26],"use":[29],"small,":[30],"high-magnification":[31],"patches":[32,128],"processed":[33,132],"independently":[34],"via":[35],"majority":[36],"voting,":[37],"without":[38,190],"systematic":[39,81],"evaluation":[40],"of":[41,84],"seemingly":[42],"inconsequential":[43],"design":[44,87],"choices":[45],"such":[46],"patch":[48,50,91,95],"size,":[49,92],"count,":[51],"and":[52,94,109,144,162,173],"magnification.":[53],"Generalist":[54],"LLMs":[55,111],"have":[56,102],"consistently":[57],"underperformed":[58],"systems,":[60],"reinforcing":[61],"the":[62,104,118],"perception":[63],"that":[64,99],"domain-specific":[65],"training":[66],"or":[67],"architectural":[68],"adaptation":[69],"is":[70],"necessary":[71],"for":[72],"tasks":[74],"involving":[75],"WSIs.":[76],"Here,":[77],"we":[78],"conduct":[79],"a":[80,123,175],"factorial":[82],"analysis":[83],"four":[85],"input":[86,115],"factors:":[88],"inference":[89],"mode,":[90],"magnification,":[93,131],"count.":[96],"We":[97],"demonstrate":[98],"prior":[100],"studies":[101],"overstated":[103],"gap":[105],"between":[106],"general-purpose":[110],"by":[112,186],"choosing":[113],"non-optimized":[114],"configurations.":[116],"On":[117],"MultiPathQA":[119],"benchmark,":[120],"switching":[121],"to":[122,138,147,159,169,174],"single":[124],"balanced":[125],"configuration":[126,167],"(large":[127],"at":[129],"lower":[130],"jointly)":[133],"raises":[134],"GPT-5":[135],"from":[136,145],"15.1%":[137],"39.5%":[139],"cancer-type":[141],"classification":[142,151],"(TCGA)":[143,161],"38.1%":[146],"62.9%":[148],"organ":[150],"(GTEx).":[152,164],"Per-task":[153],"optimization":[154],"yields":[155],"further":[156],"gains":[157],"up":[158],"43.9%":[160],"71.6%":[163],"The":[165],"same":[166],"generalizes":[168],"two":[170],"other":[171],"fully":[176],"held-out":[177],"CPTAC":[178],"cohort,":[179],"where":[180],"it":[181],"improves":[182],"Gemini":[183],"3":[184],"Flash":[185],"23.4":[187],"percentage":[188],"points":[189],"any":[191],"task-specific":[192],"tuning.":[193]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-12T00:00:00"}