Autonomous Droplet Microfluidic Design Framework with Large Language Models

Droplet-based microfluidic devices have substantial promise as cost-effective alternatives to current assessment tools in biological research. Moreover, machine learning models that leverage tabular data, including input design parameters and their corresponding efficiency outputs, are increasingly utilized to automate the design process of these devices and to predict their performance. However, these models fail to fully leverage the data presented in the tables, neglecting crucial contextual information, including column headings and their associated descriptions. This study presents μ-Fluidic-LLMs, a framework designed for processing and feature extraction, which effectively captures contextual information from tabular data formats. μ-Fluidic-LLMs overcomes processing challenges by transforming the content into a linguistic format and leveraging pretrained large language models (LLMs) for analysis. We evaluate our μ-Fluidic-LLMs framework on prediction tasks utilizing publicly available data sets on droplet microfluidics. We demonstrate that our μ-Fluidic-LLMs framework can empower deep neural network models to be highly effective and straightforward while minimizing the need for extensive data preprocessing. When combined with LLMs like LLAMA3.1 and DEEPSEEK-R1, deep neural networks achieve marked improvements, lowering the mean absolute error in generation rate by nearly 40%, reducing the root mean squared error in droplet diameter by around 26%, and enhancing regime classification accuracy by over 3% in comparison with prior results. This study lays the foundation for the huge potential applications of LLMs and machine learning in a wider spectrum of microfluidic applications.