Interpretable Machine Learning for Feature-Based Classification of Platelet Activation in Rotary Blood Pumps

Thrombosis in rotary blood pumps arises from complex flow conditions that remain difficult to translate into reliable and interpretable risk predictions using existing computational models. This limitation reflects an incomplete understanding of how specific flow features contribute to thrombus initiation and growth. This study introduces an feature-based supervised machine learning framework for spatial assessment of activation-based thrombogenic risk based directly on computational fluid dynamics-derived flow features.

A logistic regression model combined with a structured feature-selection pipeline is used to derive a compact and physically interpretable feature set, including nonlinear feature combinations. The framework is trained using spatial risk patterns from a validated, macro-scale platelet-activation-based thrombosis model for two representative scenarios.

The model reproduces the labeled risk distributions and identifies distinct sets of flow features associated with increased thrombosis risk. When applied to a centrifugal pump, despite training on a single axial pump operating point, the model predicts plausible thrombosis-prone regions. These results indicate that interpretable machine learning can link local flow features to activation-based thrombogenic risk while remaining computationally efficient and mechanistically transparent. The low computational cost enables rapid thrombogenicity screening without repeated or costly physics-based simulations.

The proposed framework complements physics-based thrombosis and platelet-activation modeling and provides a methodological basis for integrating interpretable machine learning into CFD-driven thrombogenicity analysis and device design workflows.