Predictive modeling approach using machine learning-integrated design of experiments in quality by design for...

Area of Science:

• Pharmaceutical Sciences
• Nanotechnology
• Computational Chemistry

Background:

• Resveratrol (RES) possesses therapeutic potential but requires effective delivery systems.
• Polymeric nanoparticles (PNPs) offer a promising platform for drug delivery.
• Quality by Design (QbD) is crucial for reproducible pharmaceutical manufacturing.

Purpose of the Study:

• To develop and optimize resveratrol-loaded polymeric nanoparticles (RES-PNPs) using a Machine Learning (ML)-integrated QbD approach.
• To compare the predictive performance of ML-integrated QbD with traditional QbD for nanoparticle formulation.
• To evaluate the characteristics and performance of optimized RES-PNPs for transdermal delivery.

Main Methods:

• RES-PNPs were prepared via nanoprecipitation.
• Critical Material Attributes (CMAs) and Critical Process Parameters (CPPs) were identified.
• ML-integrated QbD models were developed and compared against traditional QbD regression models.
• Optimal formulation parameters were determined from the design space.

Main Results:

• ML-integrated QbD demonstrated higher validation accuracy (lower RMSE, higher R²) compared to traditional QbD.
• Optimal RES-PNPs exhibited desirable characteristics: particle size (72.85 nm), PDI (0.30), ZP (-18.11 mV), and %EE (79.25%).
• Optimized RES-PNPs showed excellent skin permeation, cell viability, antioxidant activity, and stability.

Conclusions:

• ML-integrated QbD is a superior approach for developing robust nanoparticle formulations.
• The optimized RES-PNPs are suitable for transdermal drug delivery applications.
• ML-integrated QbD represents a promising advancement for smart pharmaceutical manufacturing.