March 25, 2025
We are thrilled to announce the release of our latest research on ChemRxiv: TuNa-AI: A Hybrid Kernel Machine to Design Tunable Nanoparticles for Drug Delivery.
Preprint: https://chemrxiv.org/engage/chemrxiv/article-details/67d47d486dde43c908cd8408
Code: https://github.com/RekerLab/TuNa-AI
In this study, we developed a bespoke kernel machine that integrates molecular learning and relative composition inference to engineer nanoparticles with tunable composition. This innovation enabled us to encapsulate previously inaccessible drugs and to optimize formulations by reducing excipient usage without compromising in vitro efficacy or in vivo pharmacokinetics.
Key highlights include:
• 42.9% improvement in nanoparticle discovery by adjusting synthesis conditions.
• Hybrid kernel is compatible with different kernel learning algorithms, but the SVM outperformed all other algorithms we tested including random forest and neural networks.
• Successful encapsulation of drugs like venetoclax and trametinib.
• Reduced excipient usage for safer, more sustainable formulations that remain bioequivalent.
Why it matters: We believe TuNa-AI helps to bridge a critical gap in AI-guided drug delivery, which so far has largely focused on either selecting novel materials or adjusting material ratios but not both. We expect this additional capability will enhance future nanoparticle development campaigns for exciting drug delivery applications to make therapies safer and more effective.