High-throughput screening accelerated by machine learning for the morphology of silica nanoparticles with high cell permeability

Abstract

In recent years, silica nanoparticles have garnered tremendous attention as drug-delivery carriers. However, the cell permeability of nanoparticles remains a major obstacle that limits the drug-delivery efficiency of drug carriers. It is a common practice to regulate the low-dimensional characteristics of nanoparticles, such as charge and size, to promote their permeability. In addition, the morphology of nanoparticles has also been proven to be a significant influencing factor. However, it is still a challenge to screen the high-dimensional characteristics of nanoparticles to optimize cell permeability. Herein, we introduce a high-throughput screening framework accelerated by machine learning for determining the morphology of silica nanoparticles with high cell permeability. We found that icosahedral morphology, the common morphology of a virus, exhibits the best performance. The superior permeability of the icosahedral nanoparticles is attributed to their larger rotational freedom during the internalization process, which promotes the formation of pores in the cell membrane. These findings are expected to inspire the design of efficient drug-delivery nanocarriers.