Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringclasses
8 values
patient_profile
stringclasses
8 values
timeline
stringclasses
8 values
longitudinal_omics
stringclasses
8 values
interventions
stringclasses
8 values
response_trajectory
stringclasses
8 values
model_narrative
stringclasses
8 values
coherence_checks
stringclasses
8 values
coherence_label
stringclasses
3 values
coherence_failure_signature
stringclasses
3 values
constraints
stringclasses
1 value
gold_checklist
stringclasses
8 values
MOCA-NC-001
38F, post-viral dysautonomia
T0 onset; T1 persistent; T2 improved
T0 IFN high, IL-6 high, kynurenine high, diversity low; T2 IFN normal, IL-6 low, kynurenine mid, diversity improved
anti-inflammatory + sleep protocol starting T1
Orthostatic intolerance + brain fog improve by T2
Narrative links IFN-driven inflammation to symptoms at T0/T1, and improvement tracks normalization by T2 after intervention
Checks: time alignment; intervention-response alignment; no missing reversal
coherent
none
Under 170 words.
1 mentions T0/T2; 2 links to response
MOCA-NC-002
45M, bile acid syndrome
T0 baseline; T1 after binder
T0 bile acids distorted, FXR targets low, FGF19 low, bile-tolerant taxa high; T1 bile acids normalize, FXR targets rise
bile acid binder at T0
Steatorrhea and RUQ pain improve at T1
Narrative: FXR feedback collapse drives symptoms; binder reduces bile acids and symptoms improve; microbiome shifts follow
Checks: mechanism to symptoms; treatment explains omics shift
coherent
none
Under 170 words.
1 ties binder to omics and symptoms
MOCA-NC-003
34M, hyperammonemia episodes
T0 after protein meal; T1 after lactulose
T0 ammonia high, urea genes down, urease taxa high; T1 ammonia down
lactulose at T0
Confusion resolves at T1
Narrative: gut urease load increases ammonia; urea bottleneck limits clearance; lactulose reduces gut ammonia leading to symptom relief
Checks: intervention explains outcome
coherent
none
Under 170 words.
1 lactulose explains ammonia drop
MOCA-NC-004
52M, cardiometabolic
T0 baseline; T1 after metformin + weight loss
T0 BCAAs high, insulin signaling down, CRP high, low butyrate taxa; T1 CRP down, insulin signaling improves
metformin + weight loss
A1c improves
Narrative: insulin resistance loop explains T0; intervention improves insulin signaling and inflammation by T1
Checks: directionality matches
coherent
none
Under 170 words.
1 correct directions
MOCA-NC-005
29F, mast cell flares
T0 flare; T1 antihistamine; T2 partial improvement
T0 mast cell signature high; T1 histamine proteins lower; methylation markers remain low
antihistamine at T0
Hives reduce at T1 but fatigue persists
Narrative: antihistamine reduces mediator symptoms but methylation bottleneck maintains background susceptibility, explaining partial response
Checks: partial response explained
coherent
none
Under 170 words.
1 explains partial
MOCA-NC-006
50F, anemia pattern
T0 baseline; T1 iron trial
T0 ferritin high, heme genes down; T1 ferritin high persists
oral iron at T0
Symptoms unchanged, nausea worse
Narrative claims iron deficiency and expects improvement at T1
Checks fail: ferritin high contradicts; response mismatch
incoherent
wrong-mechanism
Under 170 words.
1 must not expect iron benefit
MOCA-NC-007
27F, immune deficiency
T0 baseline; T1 Ig replacement + diet
T0 IgA low, pathobionts high, SCFAs low; T1 infections reduce, SCFAs rise
Ig replacement + diet
Fewer infections and better stool at T1
Narrative: immune deficit drives dysbiosis; replacement reduces infections; diet restores SCFA
Checks: sequence matches
coherent
none
Under 170 words.
1 intervention matches
MOCA-NC-008
46M, unstable symptoms
T0,T1 mixed
No stable longitudinal pattern
multiple trials no response
No clear change
Narrative claims clear causal story and improvement
Checks fail: timeline not supported
posthoc-story
overconfident-timeline
Under 170 words.
1 should state uncertainty

What this dataset tests

Whether the biology explains the clinical story across time.

It requires alignment between:

  • symptom timeline
  • longitudinal omic shifts
  • interventions
  • response trajectory

Labels

  • coherent
  • partially-coherent
  • incoherent
  • posthoc-story

Posthoc-story means the narrative sounds plausible
but does not match the timeline or measurements.

Typical failures

  • timepoint inversion
  • claiming an intervention caused a shift that occurred earlier
  • ignoring non-response
  • forcing a story when data is unstable

Suggested prompt wrapper

System

You evaluate clinical–biological narrative coherence.

User

Patient Profile
{patient_profile}

Timeline
{timeline}

Longitudinal Omics
{longitudinal_omics}

Interventions
{interventions}

Response
{response_trajectory}

Model Narrative
{model_narrative}

Return

  • one label
  • one sentence naming what aligns or conflicts

Citation

ClarusC64 dataset family

Downloads last month
13