National Science Foundation News | AI Enters Biomedicine #news #science #biomedical @NSFScience | Uploaded October 2024 | Updated October 2024, 1 hour ago.
From user facing language models to real-time, in-depth analysis of massive amounts of information, Artificial Intelligence is transforming technologies and the ways researchers are approaching societal challenges. We’ll explore a new AI training technique with biomedical applications in the U.S. National Science Foundation’s “Discovery Files”.

Artificial Intelligence holds tremendous potential for addressing major global challenges. In biomedical research, for example, AI can improve the time and resources needed to conduct the research necessary for new drugs and treatment approaches.

NSF-supported researchers from the University of Texas at Austin have developed a new system for training AI models called Evolutionary Ranking or EvoRank that seeks to accelerate the development of less toxic treatments and new preventative strategies in medicine.

Proteins have evolved over billions of years, mutating and swapping out amino acids in beneficial ways. EvoRank training distills protein evolution into AI models that use those principles to inform what changes are most plausible in nature during biotechnology development.

This new training technique seizes on the ability of AI models to synthesize large databases and generate new designs, suggesting how to alter amino acids to create new proteins and new biotechnologies.

This approach could be commercially adapted to save time, some of the massive expenses of drug development, and could enable new biotechnology markets.