Machine Learning for Programmable Nanosystems: From Liposome Formulation to DNA Origami and Hybrid Vesicles

Liposomes are lipid-based nanocarriers widely used for the delivery of a broad range of therapeutic cargos. DNA nanostructures are emerging programmable platforms being explored for sensing, biomolecular engineering, and delivery applications. However, the development of both systems is challenged by large design spaces, complex structure-function relationships, and labor-intensive characterization. This mini review discusses how machine learning (ML) can accelerate liposome formulation, DNA origami design, and emerging hybrid liposome-DNA systems. Key examples include predicting liposome quality attributes, guiding microfluidic manufacturing, automating origami characterization, enabling model-guided shape and stability prediction, and optimizing DNA-loaded lipid nanoparticles. This review examines how predictive modeling, interpretable ML, and closed-loop optimization can reduce experimental burden, reveal design rules, and support more robust programmable nanosystems. Further progress will require standardized datasets and better integration of ML with high-throughput experimentation.