Dipole moment prediction with graph neural networks and physics-informed models: a comparative evaluation on QM9 and traditional Chinese medicine molecules

Key Points

• Improved predictions of dipole moments were achieved using graph neural networks and physics-informed models.
• The models demonstrated a significant increase in accuracy, with the QM9 dataset showing a 20% improvement over traditional methods.
• Assessment of both quantum mechanical frameworks and traditional Chinese medicine molecules illustrated unique insights into molecular behavior.
• Advancing modeling techniques for dipole moment prediction may enhance applications in drug discovery and materials science.