benchmark_spec.json

{
  "repo_name": "clinical-five-node-mof-cascade-boundary-v1.0",
  "repo_path": "ClarusC64/clinical-five-node-mof-cascade-boundary-v1.0",
  "version": "v1.0",
  "conceptual_upgrade": "Closed-Loop Control Geometry",
  "concept_summary": "v1.0 evaluates whether a model can select and execute a complete closed-loop control sequence across a five-node multi-organ-failure cascade under live feedback, adaptive timing, and durable recovery constraints.",
  "domain": "clinical",
  "structure": "five-node-cascade",
  "variables": [
    "metabolic_stress",
    "perfusion_deficit",
    "renal_strain",
    "hepatic_drift",
    "buffer_capacity"
  ],
  "failure_type": "mof_boundary",
  "label_column": "label_mof_boundary",
  "primary_metric": "recall_correct_control_sequence_selection",
  "secondary_metric": "false_effective_control_rate",
  "default_label_rule": "label = 1 if stabilization_success = 1 AND trajectory_shift < -0.10 AND intervention_alignment_score >= 0.60 AND control_sequence_alignment_score >= 0.60 AND recovery_consistency_score >= 0.60",
  "mid_strength_label_rule": "label = 1 if stabilization_success = 1 AND trajectory_shift < -0.10 AND control_sequence_alignment_score >= 0.60",
  "relaxed_label_rule": "label = 1 if stabilization_success = 1",
  "trajectory_signals": [
    "drift_gradient",
    "drift_velocity",
    "drift_acceleration",
    "trajectory_shift"
  ],
  "boundary_signals": [
    "boundary_distance",
    "secondary_boundary_distance",
    "boundary_competition_ratio"
  ],
  "uncertainty_signals": [
    "boundary_uncertainty",
    "trajectory_uncertainty",
    "regime_confidence",
    "transition_uncertainty",
    "intervention_uncertainty",
    "controller_confidence",
    "feedback_noise_ratio"
  ],
  "regime_transition_signals": [
    "regime_transition_score",
    "transition_direction",
    "regime_separation_margin",
    "transition_velocity"
  ],
  "intervention_signals": [
    "intervention_leverage_score",
    "intervention_alignment_score",
    "rescue_window_width",
    "pathway_divergence_margin",
    "intervention_competition_ratio",
    "primary_intervention_path",
    "secondary_intervention_path",
    "pathway_switch_velocity",
    "minimal_intervention_path"
  ],
  "control_signals": [
    "control_sequence_alignment_score",
    "control_horizon",
    "feedback_response_score",
    "intervention_timing_score",
    "adaptation_latency",
    "control_stability_margin",
    "sequence_divergence_margin",
    "controller_confidence",
    "recovery_consistency_score",
    "control_recalibration_count",
    "terminal_pathway_state",
    "feedback_noise_ratio",
    "controller_oscillation_score",
    "rollback_trigger_count"
  ],
  "recovery_signals": [
    "recovery_distance",
    "recovery_gradient",
    "return_feasibility"
  ],
  "perturbation_signals": [
    "perturbation_radius",
    "collapse_trigger"
  ],
  "delta_signals": [
    "delta_metabolic_stress",
    "delta_perfusion_deficit",
    "delta_renal_strain",
    "delta_hepatic_drift",
    "delta_buffer_capacity"
  ],
  "outcome_fields": [
    "stabilization_success",
    "terminal_pathway_state",
    "label_mof_boundary"
  ],
  "withheld_in_tester": [
    "stabilization_success"
  ],
  "thresholds": {
    "control_alignment_threshold": 0.6,
    "intervention_alignment_threshold": 0.6,
    "recovery_consistency_threshold": 0.6,
    "high_uncertainty_threshold": 0.25,
    "narrow_window_threshold": 0.3,
    "oscillation_threshold": 0.25,
    "stable_margin_threshold": 0.6,
    "trajectory_shift_threshold": -0.1
  },
  "scorer_entrypoint": "scorer.py",
  "scorer_mode": "binary_classification_with_closed_loop_control_diagnostics",
  "reference_metrics": [
    "accuracy",
    "precision",
    "recall",
    "f1",
    "confusion_matrix",
    "recall_correct_control_sequence_selection",
    "false_effective_control_rate",
    "primary_intervention_path_accuracy",
    "secondary_intervention_path_accuracy",
    "control_sequence_alignment_accuracy",
    "control_horizon_error",
    "feedback_response_accuracy",
    "intervention_timing_accuracy",
    "high_uncertainty_control_miss_rate",
    "narrow_window_control_miss_rate",
    "adaptation_latency_error",
    "control_stability_error",
    "recovery_consistency_error",
    "recalibration_overuse_rate",
    "controller_oscillation_misread_rate",
    "terminal_pathway_state_accuracy"
  ],
  "evaluation_diagnostics": [
    "primary_intervention_path_accuracy",
    "secondary_intervention_path_accuracy",
    "control_sequence_alignment_accuracy",
    "control_horizon_error",
    "feedback_response_accuracy",
    "intervention_timing_accuracy",
    "high_uncertainty_control_miss_rate",
    "narrow_window_control_miss_rate",
    "adaptation_latency_error",
    "control_stability_error",
    "recovery_consistency_error",
    "recalibration_overuse_rate",
    "controller_oscillation_misread_rate",
    "terminal_pathway_state_accuracy"
  ],
  "files": [
    "data/train.csv",
    "data/tester.csv",
    "scorer.py",
    "benchmark_spec.json",
    "dataset_schema.json",
    "README.md"
  ]
}