Ferenc Ender

Mehrab Mahdian, Ferenc Ender, Tamas Pardy

Electrospinning is a versatile nanofabrication technique whose outcomes emerge from a complex, high-dimensional interplay between solution properties, processing parameters, and environmental conditions. Optimizing this parameter space for targeted fiber morphology is inherently challenging, often driving extensive trial-and-error experimentation and generating vast experimental data across laboratories worldwide. Yet this knowledge remains fragmented and underutilized due to inconsistent reporting and a pervasive bias toward successful outcomes, limiting reproducibility and hindering data-driven research. Here we introduce Electrospinning-Data.org, a FAIR-aligned data aggregation infrastructure that organizes dispersed electrospinning experiments into structured, reusable, and failure-aware scientific records. The platform is built around a unified process-structure-property data model linking experimental inputs, environmental conditions, and nanofiber morphology, annotated through a controlled vocabulary, within a consistent, machine-readable schema. A two-stage moderation pipeline combining automated validation with expert review supports data quality and long-term interoperability. The resulting structured, failure-inclusive corpus provides a framework for data-driven research, including predictive modelling, inverse design of target morphologies, and systematic mapping of instability regimes that would otherwise require extensive trial-and-error experimentation.

Mehrab Mahdian, Ferenc Ender, Tamas Pardy

Electrospinning is a highly sensitive fabrication process in which small variations in operating parameters can significantly influence fiber morphology and material performance. Machine learning (ML) methods are increasingly employed to model these process-structure relationships and to identify the relative importance of processing variables. However, most existing studies rely on a single ML model, implicitly assuming that the resulting feature importance is robust and reproducible. In this study, the consistency of feature importance across multiple ML model families was systematically evaluated using a curated dataset of 96 polyvinyl alcohol (PVA) electrospinning experiments. Twenty-one ML models representing linear, tree-based, kernel-based, neural network, and instance-based approaches were trained and compared. To provide a unified interpretability framework, SHAP (SHapley Additive exPlanations) values were used to calculate feature importance consistently across all models. A rank-based statistical analysis was then performed to quantify inter-model agreement and assess the robustness of parameter rankings. The results demonstrate that predictive performance and interpretive reliability are fundamentally distinct properties. Although several models achieved comparable predictive accuracy, substantial differences were observed in their feature importance rankings. Solution concentration emerged as the most robust and consistently influential parameter (variability = 0), whereas flow rate and applied voltage exhibited high ranking variability (variability > 0.9), indicating strong model dependence. These findings suggest that feature importance derived from a single ML model may be unreliable, particularly for small experimental datasets, and highlight the importance of cross-model validation for achieving trustworthy interpretation in ML-assisted electrospinning research.