August 19, 2025Nature CommunicationsOpen Access

Data-driven organic solubility prediction at the limit of aleatoric uncertainty

View Full Paper

Ask AI

Bookmark

Authors

Discussion

Models demonstrate improved accuracy in predicting organic solubility of small molecules, suggesting they approach aleatoric limits.

The models predict organic solubility with 2-3 times greater accuracy than existing alternatives.
Both models are trained on the BigSolDB dataset, addressing time-intensive experimental methods.
Utilizing FASTPROP and CHEMPROP architectures, the models predict solubility at various temperatures effectively.
The open-source models show significant speed improvements and reproducibility over current state-of-the-art methods.

Attia et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark

August 19, 2025Nature CommunicationsOpen Access

View Full Paper

Ask AI

Bookmark

Authors

Models demonstrate improved accuracy in predicting organic solubility of small molecules, suggesting they approach aleatoric limits.

The models predict organic solubility with 2-3 times greater accuracy than existing alternatives.
Both models are trained on the BigSolDB dataset, addressing time-intensive experimental methods.
Utilizing FASTPROP and CHEMPROP architectures, the models predict solubility at various temperatures effectively.
The open-source models show significant speed improvements and reproducibility over current state-of-the-art methods.

Attia et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark