{"about":{"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.","site":"https://codewithpapers.app","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","syntology":{"site":"https://syntology.ai","developers":"https://syntology.ai/developers","mcp":{"server":"https://syntology.ai/mcp","transport":"streamable-http","server_card":"https://syntology.ai/.well-known/mcp/server-card.json","auth":{"type":"trial token, no account","trial_token":"https://syntology.ai/api/oauth/trial/token","method":"POST","docs":"https://syntology.ai/developers"}},"have":"https://syntology.ai/api/graph/have?x=<method, arXiv id or title> (free, answers coverage only)","paper_base":"https://syntology.ai/paper/","atlas_base":"https://app.syntology.ai/?focus="},"machine_readable":[{"url":"https://codewithpapers.app/llms.txt","what":"the machine catalog: every machine-readable file, counted"},{"url":"https://codewithpapers.app/index/manifest.json","what":"paper-to-code index by arXiv id, with Syntology's counts"},{"url":"https://codewithpapers.app/search/manifest.json","what":"site search index (titles, authors) and its files"},{"url":"https://codewithpapers.app/download","what":"bulk files: Syntology's layer, described there"},{"url":"https://codewithpapers.app/build_manifest.json","what":"the build record: inputs, counts, exclusions, probes"}]},"url":"/dataset/radcases","name":"RadCases","full_name":null,"description_markdown":"# RadCases Dataset\r\n\r\nThis HuggingFace (HF) dataset contains the raw case labels for input patient \"one-liner\" case summaries according to the [ACR Appropriateness Criteria](https://www.acr.org/Clinical-Resources/ACR-Appropriateness-Criteria). Because many of the sources of data used to construct the RadCases dataset require credentialed access, we cannot publicly release the input patient case summaries. Instead, the \"cases\" included in this publicly available dataset are the cryptographically secure SHA-512 hashes of the original, \"human-readable\" cases. In this way, the hashes cannot be used to reconstruct the original RadCases dataset, but *can* instead be used as a lookup key to determine the ground-truth label for the dataset.\r\n\r\n## Setup\r\n\r\nPrior to using this dataset, you need to download the raw source of patient one-liners first in compliance with each of the source-specific licenses and data usage agreements. The setup process is different for each of the different dataset sources:\r\n\r\n  1. **Synthetic**: The Synthetic dataset is composed of patient one-liners synthetically generated by OpenAI's ChatGPT. You can find the raw dataset at [this GitHub link](https://github.com/michael-s-yao/radGPT/blob/74700a8b762232729caec61455a3169c5499b14c/radgpt/data/synthetic.csv). No additional setup steps are required for the Synthetic RadCases dataset.\r\n  2. **USMLE**: The USMLE dataset is comprised of practice USMLE Step- 2 and 3 cases from Medbullets that are made available by [Chen et al. (2024).](https://doi.org/10.48550/arXiv.2402.18060) The dataset is made publicly available by the cited authors at [this GitHub link](https://raw.githubusercontent.com/HanjieChen/ChallengeClinicalQA/dc1bc9f6923ea0ecafe2a033346b8f898e8a622c) - we extract the first sentence of each question stem to use as an input patient one-liner in the RadCases dataset.\r\n  3. **JAMA**: The JAMA dataset is comprised of challenging patient one-liners derived from the [JAMA Clinical Challenges](https://jamanetwork.com/collections/44038/clinical-challenge) from the Journal of the American Medical Association (JAMA). Please follow\r\n    the instructions from [`@HanjieChen`](https://github.com/HanjieChen) [here](https://github.com/HanjieChen/ChallengeClinicalQA/blob/main/README.md) to first download the dataset. We extract the first sentence of each clinical challenge to use as the input patient one-liner in the RadCases dataset.\r\n  4. **NEJM**: The NEJM dataset is comprised of challenging patient one-liners derived from the [NEJM Case Records of the Massachusetts General Hospital](https://www.nejm.org/browse/nejm-article-type/case-records-of-the-massachusetts-general-hospital) from the New England Journal of Medicine (NEJM). We provide a script [`build_nejm_dataset.py`](https://github.com/michael-s-yao/radGPT/blob/d97d4617297a1df6e3712f2c8842600ef8b826dc/scripts/build_nejm_dataset.py) to scrape the case records from the DOIs listed [here](https://github.com/michael-s-yao/radGPT/blob/d97d4617297a1df6e3712f2c8842600ef8b826dc/radgpt/data/nejm.txt), which are the same as those used by [Savage et al. (2024).](https://www.nature.com/articles/s41746-024-01010-1). The resulting `nejm.jsonl` file generated by the script should then be added to the `radGPT` home directory.\r\n  5. **BIDMC**: The Beth Israel Deaconess Medical Center (BIDMC) dataset is comprised of real anonymized, de-identified patient one-liners derived from the [MIMIC-IV Dataset](https://www.nature.com/articles/s41597-022-01899-x). Please request access to the MIMIC-IV dataset [here](https://physionet.org/content/mimic-iv-note/2.2/). The `discharge.csv.gz` file should then be added to the [`radGPT/radgpt/data`](https://github.com/michael-s-yao/radGPT/tree/main/radgpt/data) directory.\r\n\r\n## Dataset Structure\r\n\r\nEach row of the dataset is a (SHA-512 hash of a) patient \"one-liner\" **case** mapping to an ACR Appropriateness Criteria **topic**, and also the parent **panel** of that topic.\r\n\r\n  - `case`: the SHA-512 hash of the patient one-liner\r\n  - `panel`: the ACR Appropriateness Criteria panel label of the patient one-liner\r\n  - `topic`: the ACR Appropriateness Criteria topic label of the patient one-liner\r\n\r\n## Retrieving A Label\r\n\r\nTo retrieve a ground-truth ACR label from this dataset, you can use the following source code:\r\n\r\n```python\r\nimport hashlib\r\n\r\nprompt = input(\"Patient One-Liner Case: \")\r\nhash_gen = hashlib.sha512()\r\nhash_gen.update(prompt.encode())\r\nhash_val = str(hash_gen.hexdigest())\r\n```\r\n\r\nThe corresponding `hash_val` variable can then be used to lookup the corresponding `panel` or `topic` by matching `hash_val` with the `case` value in the RadCases dataset.\r\n\r\n## Direct Dataset Usage\r\n\r\nYou can download the contents of this dataset using the following terminal command:\r\n\r\n```\r\ngit clone https://huggingface.co/datasets/michaelsyao/RadCases\r\n```","description_withheld":null,"homepage":"https://huggingface.co/datasets/michaelsyao/RadCases","introduced_date":"2024-09-27","introduced_date_note":null,"introduced_by":{"paper":"/paper/evidence-is-all-you-need-ordering-imaging","title":"Evidence Is All You Need: Ordering Imaging Studies via Language Model Alignment with the ACR Appropriateness Criteria","first_author":"Michael S. Yao","url":null},"license":{"name":"MIT License","url":null},"modalities":[{"name":"Texts","url":"/datasets/modality/texts"},{"name":"Medical","url":"/datasets/modality/medical"}],"tasks":[{"name":"text-classification","url":"/task/text-classification-1","datasets_with_task":"/datasets/task/text-classification-1"}],"languages":[{"name":"English","url":"/datasets/language/english"}],"variants":["RadCases"],"data_loaders":[],"num_papers_in_archive":1,"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28"},"benchmarks":[],"papers":{"order":"date (newest first), then slug","population":"every paper with a leaderboard row on this dataset's benchmarks (the benchmark-backed subset): the archive's own papers-using-this-dataset list was never published, so this is not that list; num_papers_in_archive is the archive's own count","page":1,"pages_in_order":1,"rows_per_page":100,"shown":0,"of":0,"counts":{"papers_with_a_benchmark_row":0,"with_a_code_link":0,"where_syntology_ran_a_sample":0,"not_listed_spam_title":0,"listed":0,"listed_where_code_ran":0,"where_syntology_ran_a_sample_split":{"with_a_run_with_no_instrument_failure":0,"every_run_a_failure_of_syntologys_instrument":0,"listed_with_a_run_with_no_instrument_failure":0,"listed_every_run_a_failure_of_syntologys_instrument":0,"filter":{"states":["a run with no instrument failure","any run, instrument failures included"],"default":"a run with no instrument failure","note":"on the 'only where code ran' pages the default hides, in the browser, the rows where every run was a failure of Syntology's instrument; the second state shows them again. Rows are hidden, never re-ordered; these twins list every row"}},"definition":"distinct papers with at least one leaderboard row on this dataset's benchmarks; 'where Syntology ran a sample' counts papers with at least one harvested sample that ran, which is not a correctness claim"},"items":[],"full_list":"papers_with_a_benchmark_row: every listed paper, in this order"},"papers_with_a_benchmark_row":[],"papers_where_code_ran":0,"papers_where_code_ran_first_page":null,"papers_where_code_ran_caption":"We ran code from the paper's repository; we did not run it on this dataset or check it against this dataset's benchmarks.","pages":[],"row_route":"each paper's syntology object carries sample_list (this site's paper page, #ran), syntology_url (syntology.ai's page for the paper, only when that route probed live) and mcp (the agent call that lists the paper's harvested samples; not a verdict on the paper)","developers":"https://syntology.ai/developers","syntology_totals":{"read_at":"2026-10-05T00:36:11+00:00","papers_with_samples":0,"samples_harvested":0,"samples_ran":0,"samples_constructed":0,"samples_ran_checked":0,"samples_ran_instrument_failed":0,"samples_unverified":0,"pointer_only_for_licence":0,"papers_with_no_sample_that_ran":0,"papers_with_a_sample_that_ran":0,"papers_with_a_run_with_no_instrument_failure":0,"papers_where_every_run_was_an_instrument_failure":0,"note":"the per-paper counts of every listed paper (all pages), summed; not a rate"},"key_notes":{"samples_ran_checked":"legacy name, kept unchanged so existing readers do not break: it counts the samples that ran with no instrument failure (honoured, violated, and ran with no contract checked); it does not mean a contract was checked, and the pages print it as 'K with no instrument failure', not 'K checked'","samples_constructed":"a sub-count of the samples that ran, never subtracted from them and never a failure: an executed sample whose run returned an instance of its own class (fixture_out_type equals the entry name): the run built an object and did not compute a result (Syntology's RAN record, counts.constructed)"},"since_archive_judged":{"papers":0,"papers_placed":0,"judged_by":{"model":"a model","models_as_recorded":[],"from":"lb_rows_run_verified.jsonl: each record's `model` field (the extractor's call that wrote decisions[])"},"note":"a model's judgement from the paper's own tables that they report on this leaderboard, not a result: no value from a paper that was not placed is printed or carried here; papers counts each paper once over these leaderboards","backfill":"8,886 of 18,637 newer papers checked so far (733 more had nothing to place; 41 could not be read)","backfill_complete":false,"leaderboards":[]},"papers_note":"The archive never published its papers-using-dataset list; these are papers with a leaderboard row on this dataset's benchmarks.","record_sha256":"41582cff62881e1a2784abf68662e769e35a7671d04914416e3552cd7c78fba7","record_changed_at":"2026-10-05","record_changed_at_basis":"changed"}