What problem does it solve? Two-ODE computational models for SCI papers rarely pass all quality gates on the first run: R² fits fail on step-response data, AUC collapses when computed with time-based labels, ablation ratios fall below 2.0x because coupling is not dominant, and MAPE is inflated by the wrong computation method. This Skill encodes a systematic tuning methodology distilled from papers 90–144+ to diagnose and fix these failures. ## Core Features & Use Cases - Five-Step Tuning Methodology: Establish baseline dynamics, tune R² with exponential-rise curve fitting, fix AUC via pre/post distribution comparison, enforce dominant coupling for ablation ≥2.0x, and verify all nine metrics simultaneously. - 19 Documented Pitfalls with Fixes: Covers positive-feedback coupling ceilings, missing baseline equilibration, multiplicative vs additive coupling forms, curve_fit boundary clamping, and empty parameter sweeps for novel domains. - Validated Parameter Baselines: Provides proven 8-parameter starting points for scleral, IOP, blood flow, corneal tension, corneoscleral shell, retinal shear, and macular degeneration systems, including biphasic secondary-variable patterns. - Use Case: Given a new 2-ODE biomechanical model whose ablation ratio is only 1.2x and whose primary variable hits the ceiling, the Skill diagnoses the multiplicative positive-feedback coupling term and prescribes an additive baseline-anchored form with a reduced kappa, bringing all nine metrics to passing values. ## Quick Start Ask the agent to tune my 2-ODE model parameters so that R², AUC, accuracy, MAPE, and ablation all pass their thresholds, starting from the P140 validated baseline.
