Stonewise Debuts at AIBC2025, Focusing on Two Core Challenges of Molecular Generation Models Data and Evaluation
Stonewise Debuts at AIBC2025, Focusing on Two Core Challenges of Molecular Generation Models: Data and Evaluation
From June 12 to 13, 2025, the 2025 Artificial Intelligence and Biomedical Ecosystem Conference (AIBC2025) was successfully held in Shanghai. The conference gathered numerous renowned experts and scholars from the academic and industrial sectors at home and abroad to jointly discuss the cutting-edge progress and application practices of Artificial Intelligence (AI) technology in the biomedical field. Dr. Huang Bo, Vice President of R&D at Stonewise, was invited to attend and delivered a keynote speech on the topic of "AI-driven Design of Small Molecule Drugs".
In his speech on the afternoon of June 13, Dr. Huang Bo, with the title "Multimodal AI Generation Model Integrating Experimental Electron Density Assists Small Molecule Drug Design", deeply analyzed the innovative ideas and technical advantages of the Stonewise team in this field. He pointed out that the experimental electron density data generated during drug R&D contains enormous value but has not been fully explored. The Stonewise team innovatively applied mature quantum chemistry theories to analyze these data and extract in-depth information from them. This method can not only assist in annotating non-covalent interactions (NCI) during molecular generation, thereby enabling a more comprehensive understanding of the interaction mode between drug-like molecules and target pockets, but also effectively improve the efficiency of virtual screening.
Dr. Huang Bo emphasized the importance of the evaluation system for molecular generation models in his report. Addressing two major industry-wide pain points—"the phenomenon of non-drug-like molecules 'gaming the system for high scores'" and "computational indicators failing to truly evaluate molecular 'activity'"—Stonewise pioneered a dual strategy of "stepwise evaluation mechanism" and "active molecule reproduction verification". This innovative evaluation system fundamentally enhances the druggability of generated molecules and establishes a more scientific and reasonable evaluation standard for AI molecular generation models.
During the conference, Stonewise's booth also attracted many industry professionals to stop and communicate. The company comprehensively demonstrated its advanced technology platform, successful cooperation cases, and future development plans. Through the AIBC2025 conference, Stonewise not only fully showcased its technical strength and innovative achievements in the field of AI-driven drug R&D but also actively participated in industry exchanges and discussions, contributing to promoting the in-depth integration of AI and biomedicine. In the future, Stonewise will continue to deepen its efforts in the field of AI-driven drug R&D, committed to providing more innovative solutions for global new drug R&D, promoting the formation of a faster and better new paradigm for drug R&D, and supporting the upgrading and transformation of early-stage R&D in the pharmaceutical industry.
About Stonewise
Stonewise, founded in 2018, is a technology company driving new drug R&D through artificial intelligence technology.
Based on breakthroughs in the underlying theories of AI-driven drug R&D, drug R&D data governance, industry expertise in drug R&D, and strong software and engineering capabilities, Stonewise has built a core platform—a multimodal 3D molecular generation foundation model—that can accurately generate molecules/molecular skeletons matching the structure of target pockets.
This model is based on the GPT/Transformer framework and integrates algorithms such as geometric deep learning. Relying on this model, multiple pharmaceutical projects have achieved remarkable results, with some projects even entering the clinical stage. The model can also serve as a foundation model, allowing partners to fully integrate their own data, expertise, and models for customized iteration.
