Datasets:

ClarusC64
/

clinical-quad-stress-buffer-lag-coupling-icu-collapse-v0.9

scenario_id string	stress_load float64	physiologic_buffer float64	inflammatory_lag float64	organ_coupling float64	drift_gradient float64	drift_velocity float64	drift_acceleration float64	boundary_distance float64	secondary_boundary_distance float64	boundary_competition_ratio float64	boundary_uncertainty float64	trajectory_uncertainty float64	regime_confidence float64	regime_transition_score float64	transition_direction string	regime_separation_margin float64	transition_uncertainty float64	transition_velocity float64	intervention_leverage_score float64	intervention_alignment_score float64	rescue_window_width float64	pathway_divergence_margin float64	intervention_competition_ratio float64	primary_intervention_path string	secondary_intervention_path string	intervention_uncertainty float64	pathway_switch_velocity float64	perturbation_radius float64	collapse_trigger string	recovery_distance float64	recovery_gradient float64	return_feasibility float64	delta_stress_load float64	delta_physiologic_buffer float64	delta_inflammatory_lag float64	delta_organ_coupling float64	trajectory_shift float64	minimal_intervention_path string	stabilization_success int64	label_icu_collapse_stabilization int64
ICU_001	0.89	0.24	0.71	0.82	0.73	0.66	0.24	0.15	0.24	0.69	0.18	0.25	0.86	0.79	toward_shock_escalation	0.27	0.19	0.2	0.88	0.91	0.22	0.34	0.43	fluids_then_vasopressors	source_control_then_antibiotics	0.13	0.1	0.79	vasodilatory_hypotension	0.18	-0.49	0.86	0.22	-0.09	0.14	0.16	-0.41	fluids_then_vasopressors	1	1
ICU_002	0.82	0.29	0.63	0.75	0.65	0.58	0.18	0.2	0.27	0.8	0.22	0.24	0.78	0.69	toward_distributive_instability	0.19	0.23	0.15	0.63	0.55	0.24	0.1	0.9	fluids_then_vasopressors	source_control_then_antibiotics	0.28	0.16	0.7	uncertain_sepsis_hypotension_mix	0.34	-0.18	0.48	0.15	-0.05	0.11	0.1	-0.08	fluids_then_vasopressors	0	0
ICU_003	0.74	0.38	0.59	0.62	0.39	0.33	0.09	0.36	0.47	0.77	0.15	0.17	0.72	0.55	toward_respiratory_fatigue	0.24	0.16	0.08	0.79	0.84	0.39	0.23	0.59	nivincrease_then_airway_clearance	intubation_then_lung_protective_ventilation	0.14	0.11	0.55	retained_secretions_with_fatigue	0.27	-0.32	0.79	0.1	-0.04	0.08	0.07	-0.27	nivincrease_then_airway_clearance	1	1
ICU_004	0.91	0.22	0.76	0.84	0.76	0.69	0.25	0.12	0.21	0.67	0.19	0.28	0.88	0.82	toward_multiorgan_transition	0.3	0.18	0.22	0.91	0.94	0.2	0.37	0.39	source_control_then_antibiotics	fluids_then_vasopressors	0.12	0.19	0.83	occult_septic_source_with_mof_drift	0.16	-0.53	0.9	0.23	-0.1	0.18	0.17	-0.45	source_control_then_antibiotics	1	1
ICU_005	0.68	0.46	0.53	0.57	0.27	0.21	0.05	0.45	0.53	0.85	0.12	0.14	0.66	0.46	toward_low_grade_icu_instability	0.23	0.18	0.04	0.44	0.57	0.42	0.06	0.95	oxygen_then_observation	fluids_then_observation	0.24	0.08	0.46	mild_reserve_erosion	0.4	-0.1	0.58	0.06	-0.03	0.04	0.03	-0.05	oxygen_then_observation	0	0
ICU_006	0.87	0.27	0.68	0.73	0.7	0.62	0.21	0.17	0.24	0.71	0.2	0.23	0.82	0.74	toward_renal_perfursion_failure	0.24	0.2	0.17	0.84	0.87	0.27	0.12	0.86	fluids_then_perfusion_support	dialysis_then_perfusion_support	0.26	0.17	0.76	aki_with_hypoperfusion	0.28	-0.16	0.6	0.17	-0.06	0.11	0.1	-0.09	fluids_then_perfusion_support	0	0
ICU_007	0.79	0.34	0.66	0.65	0.51	0.43	0.13	0.29	0.38	0.76	0.17	0.19	0.74	0.62	toward_airway_obstruction	0.21	0.17	0.1	0.75	0.8	0.34	0.19	0.66	bronchodilation_then_niv	nivincrease_then_airway_clearance	0.16	0.09	0.6	bronchospasm_with_retained_secretions	0.27	-0.27	0.78	0.12	-0.04	0.09	0.06	-0.22	bronchodilation_then_niv	1	1
ICU_008	0.84	0.28	0.74	0.79	0.69	0.62	0.22	0.16	0.22	0.73	0.18	0.25	0.84	0.78	toward_infective_hemodynamic_transition	0.2	0.2	0.18	0.82	0.65	0.23	0.07	0.9	source_control_then_antibiotics	fluids_then_vasopressors	0.29	0.19	0.78	line_sepsis_or_abdominal_source	0.33	-0.14	0.57	0.15	-0.05	0.12	0.11	-0.07	fluids_then_vasopressors	0	0
ICU_009	0.73	0.39	0.58	0.61	0.38	0.32	0.08	0.35	0.47	0.74	0.14	0.16	0.73	0.56	toward_hypoxemic_secretory_instability	0.23	0.16	0.07	0.78	0.83	0.39	0.22	0.61	airway_clearance_then_niv	high_flow_then_airway_clearance	0.15	0.12	0.55	retained_secretions_with_falling_sats	0.28	-0.29	0.79	0.1	-0.04	0.08	0.07	-0.24	airway_clearance_then_niv	1	1

Clinical Quad Stress Buffer Lag Coupling ICU Collapse (v0.9)

What this repo does

This dataset implements a Clarus v0.9 intervention-competition benchmark.

Earlier dataset versions focused on detecting:

deterioration
regime transitions
boundary proximity
recovery feasibility

v0.9 extends the ladder.

The benchmark now evaluates whether a model can identify the correct rescue path when multiple interventions compete under narrowing rescue windows.

This reflects real system decision geometry.

In real systems:

several plausible interventions may exist
one intervention may have slightly higher structural leverage
intervention rankings can change rapidly
selecting the wrong path can waste the rescue window

v0.9 explicitly models intervention competition.

Core quad

The instability geometry is defined by four interacting variables.

Quad Variable	Measurement Sources	Typical Indicators
stress_load	Vasopressor need, ventilatory burden, sedation requirement, metabolic demand	Rising pressor need, tachypnoea, agitation, lactate stress
physiologic_buffer	Albumin, haemoglobin, reserve capacity, lactate clearance	Albumin low, Hb < 8, poor reserve, rising lactate
inflammatory_lag	CRP trend, fever curve, WBC trajectory, procalcitonin drift	CRP rising, fever > 38.5, WBC > 12
organ_coupling	Renal drift, hepatic strain, ABG deterioration, hemodynamic spillover	Creatinine rising, PaCO2 rising, bilirubin drift, tachycardia

These four variables define the local stability geometry of the system.

Example intervention competition scenario

A typical v0.9 row captures a situation where two or more rescue paths compete.

Example:

Primary intervention path

fluids_then_vasopressors

Secondary intervention path

source_control_then_antibiotics

Example signals

intervention_alignment_score ≈ 0.82
intervention_leverage_score ≈ 0.79
pathway_divergence_margin ≈ 0.07
intervention_competition_ratio ≈ 0.91
rescue_window_width ≈ 0.24

Interpretation

multiple plausible rescue paths exist
the primary path has slightly stronger structural leverage
the rescue window is narrowing
selecting the wrong path risks losing recoverability

This geometry is exactly what the v0.9 benchmark tests.

Prediction target

Label column

label_icu_collapse_stabilization

Default rule

label = 1 if stabilization_success = 1 AND trajectory_shift < -0.10

Relaxed variant

label = 1 if stabilization_success = 1

Preferred v0.9 rule

label = 1 if stabilization_success = 1 \
AND trajectory_shift < -0.10 \
AND intervention_alignment_score >= 0.60

The v0.9 rule rewards correctly targeted rescue, not merely recovery.

Row structure

Each dataset row describes system geometry.

State variables

stress_load

physiologic_buffer

inflammatory_lag

organ_coupling

Trajectory signals

drift_gradient

drift_velocity

drift_acceleration

Boundary signals

boundary_distance

secondary_boundary_distance

boundary_competition_ratio

Uncertainty signals

boundary_uncertainty

trajectory_uncertainty

intervention_uncertainty

Regime transition signals

regime_transition_score

transition_direction

regime_separation_margin

transition_uncertainty

transition_velocity

Intervention competition signals

intervention_leverage_score

intervention_alignment_score

rescue_window_width

pathway_divergence_margin

intervention_competition_ratio

primary_intervention_path

secondary_intervention_path

intervention_uncertainty

pathway_switch_velocity

Recovery signals

recovery_distance

recovery_gradient

return_feasibility

Perturbation signals

perturbation_radius

collapse_trigger

Delta signals

delta_stress_load

delta_physiologic_buffer

delta_inflammatory_lag

delta_organ_coupling

Outcome fields

trajectory_shift

minimal_intervention_path

stabilization_success

Label

label_icu_collapse_stabilization

Files
data/train.csv

Training rows including:

stabilization_success

label column

data/tester.csv

Evaluation rows excluding:

stabilization_success

label column

scorer.py

The scorer evaluates both classification accuracy and intervention reasoning.

Binary metrics

accuracy

precision

recall

f1

confusion_matrix

Primary v0.9 metrics

recall_correct_intervention_selection

false_effective_intervention_rate

Intervention competition diagnostics

primary_intervention_path_accuracy

secondary_intervention_path_accuracy

intervention_alignment_accuracy

high_uncertainty_intervention_miss_rate

rescue_window_miss_rate

intervention_competition_error

pathway_switch_misread_rate

benchmark_spec.json

Machine-readable dataset specification including:

schema

thresholds

signal groups

evaluation interface

Dataset construction

Dataset rows are generated in stages.

Stage 1 — baseline system state

Quad variables are sampled to represent plausible system states.

Stage 2 — trajectory dynamics

Drift signals describe movement toward or away from instability.

Stage 3 — regime detection

Boundary and transition signals describe proximity to instability basins.

Stage 4 — intervention competition modelling

Multiple intervention paths are generated.

For each candidate path:

intervention_alignment_score measures regime compatibility

intervention_leverage_score estimates trajectory impact

pathway_divergence_margin measures separation between paths

intervention_competition_ratio captures relative leverage

rescue_window_width represents remaining recoverability

The best path becomes:

primary_intervention_path

The second-best becomes:

secondary_intervention_path

Dataset limitations

This dataset models structural intervention geometry, not detailed intervention pharmacology.

Limitations include:

intervention_alignment_score represents structural compatibility rather than exact treatment efficacy

rescue_window_width represents geometric recoverability rather than clock time

intervention paths represent simplified bundles of actions

These abstractions allow intervention competition geometry to generalize across domains.

Intended use

This dataset is intended for:

research on intervention selection under instability

evaluating regime-aware reasoning systems

testing models under narrowing rescue windows

benchmarking control-style decision models

This dataset is not intended for:

clinical decision support in patient care

treatment recommendation systems

medical device deployment

regulatory evaluation

It is a research benchmark.

Clarus ladder context

Clarus datasets follow an instability-geometry ladder.

v0.1 cascade detection
v0.2 trajectory awareness
v0.3 cascade forecasting
v0.4 boundary discovery
v0.5 recovery geometry
v0.6 intervention reasoning
v0.7 uncertainty-aware intervention
v0.8 regime transition geometry
v0.9 intervention competition geometry

v0.9 marks the transition toward control-layer reasoning benchmarks.

Structural note

Clarus benchmarks represent instability geometry directly.

This dataset frames collapse as competition between:

failure pressure

intervention leverage

rescue window narrowing

The benchmark therefore behaves closer to a control problem than a simple classifier.

Production deployment

This benchmark structure helps models distinguish between:

a plausible intervention

the best intervention

a nearly tied but weaker intervention

an intervention that becomes obsolete as the rescue window closes

This geometry appears in many domains including:

clinical deterioration monitoring

infrastructure failure mitigation

supply chain disruption recovery

AI system stabilization protocols

Enterprise and research collaboration

The v0.9 intervention competition structure generalizes across domains where multiple rescue paths compete under instability pressure.

Possible extensions include:

sepsis escalation monitoring

respiratory deterioration modelling

infrastructure recovery planning

financial crisis intervention sequencing

AI system resilience benchmarking

Downloads last month: 16

Number of rows:

Total file size:

33 kB