id stringclasses 6
values | prompt stringclasses 6
values | gold_json stringclasses 4
values | reference_notes stringclasses 6
values | __index_level_0__ stringclasses 6
values |
|---|---|---|---|---|
Quad adjudication analysis. Phase III cardiology trial. Primary endpoint rate increases at two high-enrolling sites. Adjudication turnaround drops from 14 to 4 days. One adjudicator replaced mid-study. Governance: submission timeline fixed. Determine adjudication integrity risk and action. Return strict JSON only. | {"adjudication_integrity_risk":true,"risk_type":"adjudication_drift_under_operational_pressure","driver_nodes":["endpoint cluster at high-enrolling sites","accelerated adjudication turnaround","adjudicator replacement","fixed submission timeline"],"recommended_action":"adjudication_process_audit","action_detail":["audi... | confidence":0.9} | Turnaround node. | ADJ-001 |
Quad adjudication analysis. Oncology trial. Tumor response rates rise after imaging vendor change. Blinding logs incomplete. Monitoring visit delayed. Governance: interim efficacy look imminent. Determine risk and action. JSON only. | {"adjudication_integrity_risk":true,"risk_type":"vendor_shift_with_blinding_gap","driver_nodes":["response rate increase post-vendor change","imaging vendor switch","incomplete blinding logs","interim timing"],"recommended_action":"vendor_and_blinding_reconciliation","action_detail":["re-review imaging samples","verify... | confidence":0.88} | Vendor node. | ADJ-002 |
Quad adjudication analysis. Neurology trial. Stroke endpoint classification varies by region. Training refresh overdue. Single adjudicator handling 70% of cases. Governance: regulator inspection expected. Determine risk and action. JSON only. | {"adjudication_integrity_risk":true,"risk_type":"regional_variability_with_single_reviewer_bias","driver_nodes":["regional endpoint variability","training overdue","single adjudicator dominance","inspection expectation"],"recommended_action":"regional_recalibration_and_training","action_detail":["compare regional class... | confidence":0.87} | Regional node. | ADJ-003 |
Quad adjudication analysis. Pulmonary trial. Exacerbation endpoints stable overall but decline sharply at remote sites. Remote review introduced recently. Blinding confirmation incomplete. Governance: DSMB review next week. Determine risk and action. JSON only. | {"adjudication_integrity_risk":true,"risk_type":"remote_review_bias_with_blinding_gap","driver_nodes":["remote site endpoint decline","recent remote review shift","incomplete blinding confirmation","DSMB timing"],"recommended_action":"remote_process_and_blinding_audit","action_detail":["audit remote adjudications","ver... | confidence":0.89} | Remote node. | ADJ-004 |
Quad adjudication analysis. Metabolic trial. Secondary endpoint upgraded to co-primary mid-study. Adjudication charter amended. Training not completed at all sites. Governance: submission package in preparation. Determine risk and action. JSON only. | {"adjudication_integrity_risk":true,"risk_type":"charter_change_without_full_training","driver_nodes":["endpoint status upgrade","charter amendment","incomplete training","submission preparation"],"recommended_action":"charter_and_training_completion","action_detail":["complete adjudicator training","document charter c... | confidence":0.88} | Charter node. | ADJ-005 |
Quad adjudication analysis. Vaccine trial. Severe adverse event adjudications decrease at two high-enrolling sites. Reviewer workload increased. Blinding confirmation missing for 10% of cases. Governance: regulator submission deadline fixed. Determine risk and action. JSON only. | {"adjudication_integrity_risk":true,"risk_type":"workload_bias_with_blinding_missingness","driver_nodes":["SAE decline at clustered sites","increased reviewer workload","missing blinding confirmation","fixed submission deadline"],"recommended_action":"workload_and_blinding_review","action_detail":["rebalance adjudicati... | confidence":0.9} | Workload node. | ADJ-006 |
Clarus Clinical Quad Coupling Endpoint Adjudication Integrity v0.1
Purpose
Detect adjudication drift driven by four interacting nodes.
Quad nodes
- Endpoint cluster shift
- Blinding gap or reviewer dominance
- Operational or vendor process change
- Governance submission or review pressure
Input
One vignette.
Output
Strict JSON only.
Required keys
- adjudication_integrity_risk
- risk_type
- driver_nodes
- recommended_action
- action_detail
- rationale
- confidence
Files
data/train.csv
data/test.csv
scorer.py
Run scoring
Create JSONL predictions.
python scorer.py --gold_csv data/test.csv --preds_jsonl preds.jsonl
- Downloads last month
- 6