Dual-Domain AUC-SHAP Feasibility Screening Framework: Design Notes and Reference Implementation

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

The dual-domain AUC-SHAP feasibility screening framework is a standardized pre-commitment procedure for determining whether a supervised machine learning task on tabular data is feasible, given the strength of the feature-outcome relationship ("signal"). The framework integrates a technical validation domain (discrimination, AUC) with an interpretability domain (SHAP-based pattern learning), and classifies candidate tasks into feasibility levels (GREEN / YELLOW / RED, with domain-specific refinements including YELLOW-A, YELLOW-B, and GREEN*). A core design principle is domain calibration: decision thresholds are adapted to the noise structure and conventions of the target domain rather than imported as universal constants. This record contains the framework specification, its version history, and reference implementation notes. Companion application manuscripts cover elite sport monitoring, respiratory medicine, educational data mining, autism screening, and cross-domain computational evaluation.