id stringclasses 8 values | clinical_context stringclasses 8 values | narrative_text stringclasses 8 values | timeline_events stringclasses 8 values | temporal_inconsistencies stringclasses 8 values | causal_timeline_violations stringclasses 8 values | required_timeline_clarifications stringclasses 8 values | patient_safety_or_integrity_risk stringclasses 8 values | notes stringclasses 8 values | constraints stringclasses 1 value | gold_checklist stringclasses 1 value |
|---|---|---|---|---|---|---|---|---|---|---|
CTNI-001 | emergency_department | Day 1 chest pain onset at 08:00. Day 1 ECG at 10:00 normal. Troponin drawn at 11:00 pending. Note says discharged after negative troponin. | 08:00 onset; 10:00 ECG; 11:00 troponin drawn; discharge time not stated | Troponin not resulted at time of claim | Discharge justified by negative troponin before result exists | What was troponin result time. What time was discharge. Any repeat troponin. | missed MI; unsafe discharge | Disposition integrity | Under 220 words. | 4 outputs |
CTNI-002 | clinical_trial_reporting | Registry: Jan 10. Recruitment started Feb 1. Primary endpoint changed Mar 15. Paper claims endpoint prespecified before recruitment. | Jan 10 registration; Feb 1 recruitment start; Mar 15 endpoint change; paper claim | Endpoint change after recruitment contradicts prespecification claim | Causal narrative of prespecification violated by dates | What was original endpoint. When was change approved. Were analysts blinded. | scientific integrity failure; spin | Trial timeline distortion | Under 220 words. | 4 outputs |
CTNI-003 | inpatient | Patient reports symptoms began after medication started. Chart shows medication started 2 days after symptom onset. Discharge summary blames medication. | symptom onset; med start later; discharge summary blame | Blame contradicts order | Medication cannot cause symptoms that predate first dose | Confirm exact symptom onset date. Confirm first dose date and time. | wrong attribution; missed diagnosis | Common charting error | Under 220 words. | 4 outputs |
CTNI-004 | post_op | Fever began post-op Day 2. Antibiotics started Day 5. Note states antibiotics started promptly after fever onset. | Day2 fever; Day5 antibiotics | 3 day delay contradicts prompt claim | Prompt-treatment claim violates interval | Why delay occurred. Any interim deterioration. Any cultures taken earlier. | sepsis risk; misdocumentation | Delay distortion | Under 220 words. | 4 outputs |
CTNI-005 | follow_up | Discharged on Day 4. Follow-up note dated Day 3 says patient improved after discharge. | Day4 discharge; Day3 note | Note describes future event | Improvement after discharge cannot occur before discharge | Is note date incorrect. What is actual contact date. | documentation integrity; care coordination errors | Date stamp error | Under 220 words. | 4 outputs |
CTNI-006 | stroke_pathway | Symptoms started at 07:00. Thrombolysis given at 06:30 per note. Later section says thrombolysis at 09:30. | 07:00 onset; 06:30 thrombolysis; 09:30 thrombolysis | Internal timestamp contradiction | Time-to-treatment logic breaks; affects eligibility narrative | Which time is correct. Which source system time is used. | major safety and legal risk | Critical timestamp integrity | Under 220 words. | 4 outputs |
CTNI-007 | medication_safety | Dose increased at week 8. SAEs increased week 9 onward. Report states SAEs occurred before escalation and were unrelated. | week8 escalation; week9 SAE rise; report claim | Claim conflicts with SAE rise timing | Dismissal of relatedness ignores temporal association | Provide SAE timeline by week. Any lag effects considered. | patient harm; misleading safety profile | Safety timing laundering | Under 220 words. | 4 outputs |
CTNI-008 | diagnostic_workup | CT ordered Day 1. CT performed Day 3. Note says CT ruled out PE before anticoag started Day 2. | Day1 CT order; Day2 anticoag start; Day3 CT performed | CT cannot rule out PE before it happened | Causality narrative reversed | Confirm CT performance time. Was there interim imaging. | wrong anticoag rationale; risk | Order vs result confusion | Under 220 words. | 4 outputs |
What this dataset tests
Whether a clinical narrative preserves time order
and does not invent causality that violates timestamps.
Required outputs
- timeline events
- temporal inconsistencies
- causal timeline violations
- required timeline clarifications
Typical failures
- discharge justified by results not yet available
- endpoint prespecification claims contradicted by registry dates
- medication blamed for symptoms that predate dosing
Suggested prompt wrapper
System
You extract clinical timelines and detect integrity failures.
User
Narrative
{narrative_text}
Return
- timeline events
- temporal inconsistencies
- causal timeline violations
- required clarifications
Citation
ClarusC64 dataset family
- Downloads last month
- 457