Datasets:

ClarusC64
/

clarus-endpoint-definition-integrity-v0.1.1

Tasks:

Text Classification

Modalities:

Formats:

Languages:

Size:

Tags:

clinical-trials

evidence-synthesis

Libraries:

License:

Dataset card Data Studio Files Files and versions

Dataset Viewer

Auto-converted to Parquet Duplicate

Split (1)

train · 10 rows

sample_id string	trial_a_id string	trial_b_id string	synth_doc_id string	concept string	trial_a_context string	trial_b_context string	claim string	option_a string	option_b string	task string	correct_option string	required_note string	required_keywords string	axis string	difficulty int64
CEDI11-0001	NCTX-E01	NCTX-E02	META-101	PFS death handling	PFS counts death as an event	PFS censors death	Pool PFS as identical across trials	Invalid mapping, event rule differs	Partial mapping, state rule difference and adjust	Choose mapping and write the required note	A	State death-as-event vs death-censored and do not treat PFS as identical	death as event\|censors death\|PFS\|not identical\|event rule	event_rule	3
CEDI11-0002	NCTX-E03	NCTX-E04	META-102	ORR confirmation	ORR per RECIST 1.1 confirmed response	ORR unconfirmed best overall response	Treat ORR as equal	Partial mapping, confirmation differs	Direct mapping, both ORR	Choose mapping and write the required note	A	State confirmed vs unconfirmed response and flag non-equivalence unless bridged	confirmed\|unconfirmed\|ORR\|RECIST\|non-equivalent	responder_rule	2
CEDI11-0003	NCTX-E05	NCTX-E06	META-103	OS time origin	OS measured from randomization	OS measured from first dose	Pool OS without notes	Partial mapping, time zero differs	Direct mapping, OS label matches	Choose mapping and write the required note	A	State randomization vs first dose and define time zero used in synthesis	randomization\|first dose\|time zero\|OS\|define	time_origin	2
CEDI11-0004	NCTX-E07	NCTX-E08	META-104	Composite endpoint components	MACE includes hospitalization	MACE excludes hospitalization	Treat MACE as same composite endpoint	Invalid mapping, components differ	Partial mapping, restate components and bridge	Choose mapping and write the required note	A	State which components differ and do not pool as identical composite	MACE\|hospitalization\|components\|composite\|not identical	composite_components	3
CEDI11-0005	NCTX-E09	NCTX-E10	META-105	AE window definition	Grade 3+ AEs treatment-emergent (TEAE)	Grade 3+ AEs any time on study	Pool AE rates directly	Partial mapping, window differs	Direct mapping, same grade threshold	Choose mapping and write the required note	A	State TEAE vs any-time window and specify the window used	TEAE\|any time\|window\|Grade 3+\|specify	window_definition	2
CEDI11-0006	NCTX-E11	NCTX-E12	META-106	Censoring method	Censor at last known assessment	Impute event at 30 days after LTFU	Treat survival curves as comparable	Invalid mapping, censoring differs	Partial mapping, state method and adjust	Choose mapping and write the required note	A	State censoring vs imputed event and do not treat curves as equivalent	censor at last\|impute event\|30 days\|LTFU\|not equivalent	censoring_handling	3
CEDI11-0007	NCTX-E13	NCTX-E14	META-107	Endpoint assessment schedule	Tumor scans every 6 weeks	Tumor scans every 12 weeks	Pool PFS hazard rates as equal	Partial mapping, assessment frequency differs	Direct mapping, both PFS	Choose mapping and write the required note	A	State scan schedule difference and note detection bias on PFS	6 weeks\|12 weeks\|scan schedule\|detection bias\|PFS	assessment_schedule	2
CEDI11-0008	NCTX-E15	NCTX-E16	META-108	PFS censoring at new therapy	Censor at start of new anti-cancer therapy	Continue follow-up through new therapy	Pool PFS outcomes	Invalid mapping, censor rule differs	Partial mapping, state rule and adjust	Choose mapping and write the required note	A	State censor-at-new-therapy vs continue-follow-up and do not pool as identical	censor at new therapy\|continue follow-up\|PFS\|censor rule\|not identical	censor_rule	3
CEDI11-0009	NCTX-E17	NCTX-E18	META-109	PRO endpoint scale	QoL mean change from baseline at week 12	QoL responder rate at week 12	Treat PRO endpoints as the same effect size	Invalid mapping, scale differs	Partial mapping, map with stated transform	Choose mapping and write the required note	A	State mean change vs responder rate and do not treat as same scale	mean change\|responder rate\|QoL\|scale\|not same	endpoint_scale	3
CEDI11-0010	NCTX-E19	NCTX-E20	META-110	OS vs PFS swap	Primary endpoint OS	Primary endpoint PFS	Use OS from one trial and PFS from the other as interchangeable benefit	Invalid mapping, endpoints differ	Partial mapping, state and avoid direct swap	Choose mapping and write the required note	A	State OS vs PFS difference and do not swap endpoints without explicit model	OS\|PFS\|not interchangeable\|endpoint\|model	endpoint_identity	3

Clarus Endpoint Definition Integrity v0.1.1

What this measures

You stop endpoint-label reuse from hiding definition changes
You force the model to name the missing endpoint details

Task per row

Pick A or B
Write the required note with the key definition difference

Gold support

required_keywords is a "|" list of key terms your note must include

Predictions file

CSV columns
- sample_id
- predicted_option
- predicted_note

Downloads last month: 13

Size of downloaded dataset files:

Size of the auto-converted Parquet files:

Number of rows: