SpinCastML an Open Decision-Making Application for Inverse Design of Electrospinning Manufacturing: A Machine Learning, Optimal Sampling and Inverse Monte Carlo Approach

Electrospinning is a powerful technique for producing micro to nanoscale fibers with application specific architectures. Small variations in solution or operating conditions can shift the jet regime, generating non Gaussian fiber diameter distributions. Despite substantial progress, no existing framework enables inverse design toward desired fiber outcomes while integrating polymer solvent chemical constraints or predicting full distributions. SpinCastML is an open source, distribution aware, chemically informed machine learning and Inverse Monte Carlo (IMC) software for inverse electrospinning design. Built on a rigorously curated dataset of 68,480 fiber diameters from 1,778 datasets across 16 polymers, SpinCastML integrates three structured sampling methods, a suite of 11 high-performance learners, and chemistry aware constraints to predict not only mean diameter but the entire distribution. Cubist model with a polymer balanced Sobol D optimal sampling provides the highest global performance (R2 > 0.92). IMC accurately captures the fiber distributions, achieving R2 > 0.90 and <1% error between predicted and experimental success rates. The IMC engine supports both retrospective analysis and forward-looking inverse design, generating physically and chemically feasible polymer solvent parameter combinations with quantified success probabilities for user-defined targets. SpinCastML reframes electrospinning from trial and error to a reproducible, data driven design process. As an open source executable, it enables laboratories to analyze their own datasets and co create an expanding community software. SpinCastML reduces experimental waste, accelerates discovery, and democratizes access to advanced modeling, establishing distribution aware inverse design as a new standard for sustainable nanofiber manufacturing across biomedical, filtration, and energy applications.

💡 Research Summary

SpinCastML is an open‑source decision‑making platform that enables inverse design of electrospinning processes by predicting full fiber‑diameter distributions rather than single mean values and by generating feasible formulation and operating conditions for user‑defined target distributions. The authors curated a comprehensive literature‑derived dataset comprising 68,480 individual fiber‑diameter measurements from 1,778 studies covering 16 polymers and a wide range of solvents, concentrations, voltages, flow rates, tip‑to‑collector distances, and ambient conditions. Recognizing the severe class imbalance (some polymers dominate the data), they evaluated three structured sampling strategies: simple random sampling, Sobol low‑discrepancy sequences combined with D‑optimal experimental design, and a polymer‑balanced hybrid that preserves minority‑polymer representation while still maximizing information content. Benchmarking showed that a sample size of about 10 000 points provides a sweet spot where predictive performance plateaus (R² ≈ 0.94) while computational cost remains manageable.

Eleven machine‑learning algorithms—including linear regression, random forests, XGBoost, and Cubist rule‑based regression—were cross‑validated (k = 3, 5, 10) using RMSE, MAE, MAPE, and R² as metrics. The Cubist model, when trained on the Sobol + D‑optimal polymer‑balanced sample, achieved the highest global performance (R² > 0.92) and, crucially, was capable of predicting the entire probability distribution of fiber diameters. This distribution‑aware capability allows the model to capture the heavy‑tailed, often multimodal nature of electrospun fibers that traditional mean‑only approaches miss.

The inverse design engine builds on the trained global model. Users specify a desired diameter distribution (e.g., target mean, variance, or even a bimodal shape) and a tolerance band. The engine then generates chemically feasible candidate formulations using quasi‑random Sobol/LHS sampling, enforcing constraints such as polymer‑solvent solubility, solvent‑solvent miscibility, and optional “green” solvent preferences. Each candidate is evaluated by the Cubist model to predict its full diameter distribution; the predicted success probability is calculated by comparing the simulated distribution to the user’s target. Validation on seven polymer‑solvent systems showed that the IMC predictions achieved R² > 0.90 against experimental data and less than 1 % error in success‑rate estimation.

SpinCastML is distributed as a standalone executable with a graphical user interface, allowing researchers without machine‑learning expertise to upload their own curated datasets, run automated preprocessing, train the selected model, and perform inverse design in a reproducible, auditable workflow. By integrating data‑driven modeling, optimal experimental design, and chemistry‑aware constraints, the platform reduces the number of wet‑lab iterations, cuts material waste, and accelerates discovery across biomedical scaffolds, filtration membranes, energy devices, and other nanofiber applications. The open‑source nature invites community contributions, fostering an expanding database and continuous improvement of the modeling pipeline. In sum, the work establishes a new standard for sustainable, distribution‑aware inverse design in electrospinning, shifting the field from trial‑and‑error experimentation to a reproducible, probabilistic, and chemically realistic design paradigm.