M3DISEEN

M3DISEEN predicts fabrication parameters for fused deposition modeling (FDM) three-dimensional printing (3DP) of oral drug-loaded formulations using machine learning to optimize printability, filament characteristics, and processing temperatures.

Key Features:

• Predictive Modeling: Machine learning and AI models predict printability and filament characteristics, achieving 76% accuracy for printability and 67% accuracy for filament characteristics.
• Temperature Prediction: Models predict hot melt extrusion (HME) and FDM processing temperatures with mean absolute errors of 8.9°C (HME) and 8.3°C (FDM).
• Data-Driven Approach: Models were developed from a dataset of 614 drug-loaded formulations comprising 145 pharmaceutical excipients and evaluated using a 75:25 training:test split.

Scientific Applications:

• Accelerating Development: Reduces empirical trial-and-error in optimizing FDM 3DP parameters to shorten formulation development timelines.
• Enhancing Personalization: Supports design of personalized drug-loaded filaments and oral formulations by predicting formulation-specific fabrication parameters.
• Facilitating Innovation: Provides a predictive framework to enable exploration of novel excipient combinations and FDM processing conditions.

Methodology:

Model development used a dataset of 614 formulations with 145 excipients; AI/ML models were trained on 75% of the data and tested on 25%, and focused on predicting printability, filament characteristics, and HME and FDM processing temperatures.