Automated Causal Disentanglement for Efficient Lipid Emulsion Formulation Optimization in Riboflavin Delivery Systems

**Abstract:** Developing stable and bioavailable lipid emulsion formulations for riboflavin (Vitamin B2) delivery presents significant challenges due to the inherent complexity of interfacial phenomena and the sensitivity of riboflavin to oxidation. This research introduces a novel framework leveraging Automated Causal Disentanglement (ACD) to optimize lipid emulsion formulation parameters – specifically, oil phase ratio, surfactant type, and homogenization energy – by accurately identifying and isolating the causal factors influencing emulsion stability, particle size distribution, and riboflavin encapsulation efficiency. The ACD framework combines advanced microscopy image analysis, machine learning-driven causal inference, and simulations to drastically reduce formulation development time and improve product quality. We demonstrate a 10-fold reduction in experimental iterations required to achieve target formulation characteristics, achieving 95% accuracy in predicting emulsion stability and encapsulation efficiency.