Predicting Cold Atmospheric Plasma‐Induced Wettability Transformations in Biopolymers: A Stacked Ensemble Machine Learning Framework

Cold atmospheric plasma (CAP) is a highly effective technology for modifying the surface wettability of biopolymers; however, the complex, non‐linear nature of plasma‐surface interactions makes outcomes notably difficult to predict, frequently necessitating exhaustive trial‐and‐error optimizations. This study proposes a robust, explainable machine learning (ML) framework to accurately predict CAP‐induced wettability transformations and decode their underlying mechanisms. Utilizing a comprehensive dataset of 534 experimental observations compiled from existing literature, a sophisticated stacked ensemble architecture, integrating an Adaptive Boosting (AdaBoost) base learner with a Random Forest (RF) meta‐learner, was developed following the synthetic minority over‐sampling technique for regression with Gaussian noise (SMOGN)‐based distribution balancing. To address the inherent baseline noise of untreated materials, the modeling strategy compared the prediction of the absolute final water contact angle (WCA) against derived relative metrics (retention and reduction). The results demonstrated a pivotal methodological shift, while predicting absolute WCA yielded a moderate predictive accuracy (), predicting the relative wettability reduction effectively isolated the genuine modification power of the plasma from the initial material variance, achieving an exceptional independent test of 0.9079 with minimized prediction errors. Furthermore, SHapley Additive exPlanations (SHAP)‐based feature importance analysis revealed that the magnitude of wettability reduction is fundamentally governed by the treated polymer type, applied voltage, and treatment time, effectively recontextualizing established literature conventions. Ultimately, this explainable predictive framework provides a reliable in silico compass, enabling researchers to bypass empirical iterations and accelerating the rational design of plasma‐activated biomaterials for advanced biomedical applications.