Intelligence prediction of microfluidically prepared nanoparticles

Area of Science:

• Biomaterials Science
• Nanotechnology
• Pharmaceutical Sciences

Background:

• Developing poly(lactic-co-glycolic) acid (PLGA) nanoparticles is vital for drug delivery.
• Controlling nanoparticle physicochemical properties for optimal drug encapsulation and loading remains a significant challenge.

Purpose of the Study:

• To compile a comprehensive dataset of PLGA nanoparticle formulations.
• To apply machine learning algorithms for predicting drug loading (DL) and encapsulation efficiency (EE).

Main Methods:

• Compiled a dataset of over 300 PLGA nanoparticle formulations from literature.
• Included 25 key features related to microfluidic preparation.
• Utilized various machine learning algorithms, including random forest, to predict DL and EE.

Main Results:

• The random forest model demonstrated high predictive performance.
• Achieved R² values of 0.93 for DL and 0.96 for EE predictions.
• Found minimal impact between EE and DL predictions, indicating distinct formulation insights.

Conclusions:

• Machine learning effectively predicts key properties of PLGA nanoparticles.
• This approach can guide the rational design of drug delivery systems.
• Accelerates the development of nanoparticles with tailored drug loading and encapsulation efficiency.