Droplet microfluidics enables the generation of highly monodisperse picoliter droplets and underpins a wide range of applications, from single-cell analysis to materials synthesis. Despite the maturity of the field, the design and operation of droplet microfluidic systems still largely rely on trial-and-error approaches, driven by the absence of universal predictive models and exacerbated by device-to-device and lab-to-lab variability. In recent years, machine learning (ML) and artificial intelligence (AI) have emerged as powerful tools to address these challenges by extracting structure–property–process relationships from experimental and numerical data. This Perspective provides a critical overview of recent advances in AI-enabled droplet microfluidics, with a particular focus on three interconnected themes of crucial interest in microfluidic applications: prediction of droplet properties, read out of droplet characteristics from experimental images, and closed-loop optimization of device operation and design. We discuss how data-driven models have enabled accurate prediction of droplet size, frequency, and regime across different geometries, how computer vision approaches have transformed high-throughput droplet analysis, and how Bayesian optimization and autonomous frameworks are moving the field toward automated microfluidic platforms. Finally, we highlight current limitations, including data sparsity, generalizability, and the treatment of complex fluids, and outline key opportunities for future research aimed at establishing robust, interpretable, and broadly applicable ML and AI tools for droplet microfluidics.
Francesco Del Giudice (Sun,) studied this question.