Datasets:

ClarusC64
/

ffr-physiology-prediction-coherence-baseline-mapping-v0.1

id stringclasses 6 values	case_context stringclasses 6 values	ai_ffr_prediction float64 0.72 0.84	myocardial_perfusion_index float64 0.45 0.88	wall_motion_score float64 0.5 1	stress_test_result stringclasses 4 values	heart_rate int64 62 90	blood_pressure_systolic int64 118 145	physiology_alignment_error float64 0.02 0.25	physiological_coherence_score float64 0.35 0.95	baseline_label stringclasses 6 values	notes stringclasses 6 values	constraints stringclasses 1 value	gold_checklist stringclasses 1 value
PPCB-001	normal physiology	0.84	0.88	1	normal	62	118	0.02	0.95	high-coherence	aligned	<=250 words	coherence+error+baseline
PPCB-002	mild perfusion drop	0.8	0.75	1	mild defect	70	122	0.05	0.85	stable	expected mild shift	<=250 words	coherence+error+baseline
PPCB-003	wall motion mild abnormal	0.78	0.82	0.8	borderline	74	128	0.07	0.78	stable-low	minor discordance	<=250 words	coherence+error+baseline
PPCB-004	perfusion abnormal	0.76	0.6	0.7	abnormal	80	132	0.12	0.6	edge	baseline drift	<=250 words	coherence+error+baseline
PPCB-005	multi-modal mismatch	0.74	0.58	0.6	abnormal	85	138	0.18	0.48	low-coherence	unstable baseline	<=250 words	coherence+error+baseline
PPCB-006	severe mismatch	0.72	0.45	0.5	abnormal	90	145	0.25	0.35	fragile	physiology conflict	<=250 words	coherence+error+baseline

Goal

Define the baseline coherence
between AI-derived FFR predictions
and real physiological signals.

Signals include:

This dataset establishes
what physiologically plausible alignment
looks like.

Without this baseline
implausibility cannot be detected.

The model must provide:

AI-FFR can remain stable numerically
while becoming physiologically implausible.

This dataset detects the moment
prediction and physiology
stop telling the same story.

It enables:

The scorer checks:

Future versions will include
true regression scoring
against ground-truth coherence values.