{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/arxiv-2606-01204","title":"Implicit Geographic Inference in LLM Medical Triage: Language-Driven Disparities in Emergency Recommendations","arxiv_id":"2606.01204","date":"2026-05-31","proceeding":null,"authors":["Qi Han Wong"],"abstract":"We investigate whether large language models produce different medical triage recommendations for identical symptoms based solely on the language of the patient prompt. Using Gemini 3.5 Flash, we evaluate a neurological symptom profile (persistent headache, blurred vision, nausea) across six languages (English, Spanish, Chinese, Hindi, Japanese, Arabic) with 30 runs per condition (n=450 total API calls). We find that the model recommends emergency room visits at rates ranging from 0% (Japanese, Hindi) to 30% (English, Arabic), despite assigning nearly identical severity scores (7.7-8.0/10) across all languages. Adding a single sentence specifying the patient's US location increases ER recommendations by up to 76.7 percentage points for non-English prompts, while the reverse anchor (English prompt with a Tokyo location) reduces the ER rate from 30% to 6.7%. A back-translation control (Japanese to English) produces ER rates comparable to the English baseline, confirming that the disparity is not caused by translation quality but by implicit geographic inference from the input language. We release the complete dataset, experiment code, and results.","url_abs":"https://arxiv.org/abs/2606.01204","url_pdf":"https://arxiv.org/pdf/2606.01204","source":{"archive":null,"snapshot":"2025-07-28","note":"not in the Papers with Code archive (frozen at the snapshot)","row_kind":"graph","title_abstract_authors_date":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)"},"code_links":[],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2606.01204","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2606.01204"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/wongqihan/ai-behavioral-experiments","reach":null}],"summary":{"ran":17,"unverified":4},"by_repo_kind":{"found_in_text":{"samples":21,"ran":17,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"71663d9943983420","entry":"build_questions_prompt","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"grain-reporting/experiment.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/grain-reporting/experiment.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"71663d9943983420"}},{"code_sha256_prefix":"0a53748b580d1e05","entry":"build_scoring_feedback","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"grain-reporting/experiment.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/grain-reporting/experiment.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"0a53748b580d1e05"}},{"code_sha256_prefix":"ab7269c489a15c95","entry":"call_claude","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"customer-refund/run_experiment.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/customer-refund/run_experiment.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"ab7269c489a15c95"}},{"code_sha256_prefix":"deff18135874f621","entry":"call_fable","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"delegation-laundering/run_fable.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/delegation-laundering/run_fable.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"deff18135874f621"}},{"code_sha256_prefix":"3fe2e153ceb4ff0d","entry":"call_gemini","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"customer-refund/run_experiment.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/customer-refund/run_experiment.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"3fe2e153ceb4ff0d"}},{"code_sha256_prefix":"39342c9858467c91","entry":"call_gemini","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"gender-age-triage/run_experiment.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/gender-age-triage/run_experiment.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"39342c9858467c91"}},{"code_sha256_prefix":"788e3001191e9f9c","entry":"extract_price","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"hotel-negotiation/run_a2a_negotiation.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/hotel-negotiation/run_a2a_negotiation.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"788e3001191e9f9c"}},{"code_sha256_prefix":"050f553a09339598","entry":"get_by_category","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"grain-reporting/questions.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/grain-reporting/questions.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"050f553a09339598"}},{"code_sha256_prefix":"969901700d81b4d1","entry":"get_payoffs","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"prisoners-dilemma/run_pd_advanced.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/prisoners-dilemma/run_pd_advanced.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"969901700d81b4d1"}},{"code_sha256_prefix":"44573ff7fdea6231","entry":"is_refusal","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"delegation-laundering/run_fable.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/delegation-laundering/run_fable.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"44573ff7fdea6231"}},{"code_sha256_prefix":"391c1b0d7d59effb","entry":"parse","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"multilingual-medical-triage/run_french_russian.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/multilingual-medical-triage/run_french_russian.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"391c1b0d7d59effb"}},{"code_sha256_prefix":"677780718b58e3e8","entry":"parse","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"multilingual-medical-triage/run_us_anchor_fr_ru.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/multilingual-medical-triage/run_us_anchor_fr_ru.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"677780718b58e3e8"}},{"code_sha256_prefix":"49e53fd218f47011","entry":"parse_json","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"gender-age-triage/run_multimodel.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/gender-age-triage/run_multimodel.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"49e53fd218f47011"}},{"code_sha256_prefix":"0fa89ba396fc0411","entry":"run_batch","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"multilingual-medical-triage/run_experiment.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/multilingual-medical-triage/run_experiment.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"0fa89ba396fc0411"}},{"code_sha256_prefix":"17f783a7488c6a4a","entry":"score_answer","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"grain-reporting/experiment.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/grain-reporting/experiment.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"17f783a7488c6a4a"}},{"code_sha256_prefix":"8ec55e7aa92ddbdb","entry":"to_usd","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"salary-negotiation-language/run_experiment.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/salary-negotiation-language/run_experiment.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"8ec55e7aa92ddbdb"}},{"code_sha256_prefix":"daf150f5e4e5b28b","entry":"wilson_ci","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"gender-age-triage/run_experiment.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/gender-age-triage/run_experiment.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"daf150f5e4e5b28b"}},{"code_sha256_prefix":"2ab23adcd68fd9fd","entry":"call_claude","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"gender-age-triage/run_multimodel.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/gender-age-triage/run_multimodel.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"2ab23adcd68fd9fd"}},{"code_sha256_prefix":"34cc7029f1e0e1cb","entry":"call_gemini","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"hotel-negotiation/run_a2a_negotiation.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/hotel-negotiation/run_a2a_negotiation.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"34cc7029f1e0e1cb"}},{"code_sha256_prefix":"e3d170768035e5fe","entry":"call_gpt","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"gender-age-triage/run_multimodel.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/gender-age-triage/run_multimodel.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"e3d170768035e5fe"}},{"code_sha256_prefix":"880b98e0cc23dd7a","entry":"load_env","repo":"wongqihan/ai-behavioral-experiments","repo_kind":"found_in_text","path":"customer-refund/run_experiment.py","file_url":"https://github.com/wongqihan/ai-behavioral-experiments/blob/HEAD/customer-refund/run_experiment.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"880b98e0cc23dd7a"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.CL","source":"arxiv_2026.jsonl"},"syntology_extracted_results":null}