Rapid screening of the stability of polyacrylamide-based hydrogel coatings via droplet microarray analysis and interpretable machine learning

Abstract

The unsatisfactory stability of hydrogel coatings hinders their functional and service performance. Until now, the development of high-performance hydrogel coatings largely relies on the intuition and prior experience of researchers. Machine learning, as a powerful engine for material design, was demonstrated to accelerate the development of hydrogels with desired properties. However, the scarcity of labeled data of the target property is a fundamental challenge. Herein, we develop a miniaturized high-throughput evaluation method of hydrogel coatings. This method achieved a rapid and parallel investigation of the stability of a large number of unique acrylamide-based hydrogel coatings. Moreover, a list of main feature descriptors was screened and their quantitative contributions to coating stability were analyzed via interpretable machine learning technology. A new ternary hydrogel coating was prepared to validate the accuracy of the machine learning strategy. This advanced methodology facilitated the rational design of high-performance hydrogel coatings.