(588cy) A Comprehensive Database and Machine Learning Framework for Predicting Liquid Crystalline Behavior

tunable optical devices, biosensors, soft robotics an d drug delivery systems. The ability to predict LC behavior, particularly melting temperature and cl earing temperature, is essential for optimizing materials for these applications. While previous studi es have employed machine learning models to predict specific LC properties, such as the clearing temperatures of bent-core LCs or melting points of rod-like LCs, these models are often limited to specific molecular subsets and tasks.

This work presents a comprehensive machine learni ng framework capable of classifying LC molecules and predicting their melting and clearing temperatures.

By curating an extensive dataset of rod-like, discotic, and bent-core LCs, we develope d a state-of-the-art deep learning model trained on molecular structures and validated agains t experimental data.

Our freely accessible online platform enables researchers to predict LC pr operties efficiently, accelerating data-driven material discovery. This holistic approach surpasses existing models by integrating diverse LC systems into a unified predictive framework, ultimate ly facilitating the design of novel liquid crystalline materials for next-generation technologies.