COREX: Causal Origin Resolution and Empirical eXamination — An Autonomous Multi-Stage Framework for Robust Causal Discrimination in Data-Driven AI Systems

COREX (Causal Origin Resolution and Empirical eXamination) is a deterministic, graph-free, model-agnostic computational framework that treats causality as an empirically testable robustness property rather than an assumed structural characteristic. The framework implements a four-axis evaluation pipeline — Statistical Stability (S), Representation Invariance (R), Intervention Consistency (I), and Domain Robustness (D) — and fuses their outputs through a weighted scoring function to produce a calibrated causal classification: CAUSAL (≥ 0.80), SPURIOUS (0.50–0.79), or REPRESENTATION ARTIFACT (< 0.50). Unlike prior approaches dependent on pre-specified causal graphs or structural equation models, COREX requires no prior causal knowledge and operates as a meta-evaluation layer atop any learned relationship. An optional learnable MLP meta-scoring layer provides adaptive calibration for high-dimensional settings. Validated on 1,500 synthetic labeled datasets and real-world vagus nerve electrophysiology data: C-Precision 0.91 · C-Recall 0.88 · A-Recall 0.93 · False Causal Rate 0.07. Released as open-source Python package (PyPI: pip install corex) under MIT License. OSF Preregistration: https://doi.org/10.17605/OSF.IO/3ABZF | BIO-MED-02 | Ronin Institute / Rite of Renaissance.