Self-regulating microfluidic system for lipid nanoparticle production

Lipid nanoparticles have emerged as valuable gene delivery systems paving the way for next-generation vaccine and cancer therapeutics. Inevitably, this evolution is carried by dissecting and rationalizing the vehicles’ complex formulation process. Given the vast design space, in silico methods resemble an elegant and cost-effective optimization approach. Here, we provide a proof-of-concept study on how data-driven automatization leverages rapid formulation parameterization, using readily obtainable, low-cost microfluidic hardware. Insights gained from both computational fluid dynamics simulations and microfluidic screenings are harnessed to derive and complement machine learning algorithms that predict critical quality attributes, such as size and encapsulation efficiency. Subsequently, these models are used to deploy a self-regulating microfluidic device, thereby bridging the gap between our computational and experimental work and enabling fully automated lipid nanoparticle formulation optimization on the fly, with minimal human intervention required. We envision our approach to accelerate the discovery of optimized nanoparticles in future designs. • Simulated fluid flow dynamics of lipid nanoparticle formulations. • Experimentally formulated particles using an automated microfluidic system. • Built machine learning algorithms to complement simulation-based interpretations. • Implemented a self-regulating optimization mechanism without human intervention.