LLM-Retrieval Augmented Generation for Drug Shortages

2024 Large Language Models have improved over previous software in generation quality, making them more applicable for use in drug shortage cases. This includes issues of reproducibility, cost efficiency, and green chemistry that can now be better addressed using advanced natural language processing techniques. Through the utilization of Retrieval Augmented Generation, literature analysis of several organic syntheses was performed with a high degree of accuracy. Stemming from the original literature, additional insight regarding next additional synthetic steps was provided by both ChatGPT 4o and Llama 3.1 405B. The ChatGPT 4o RAG workflow was particularly easy to use and generated the most effective generations, whether a single document was used, or multiple documents were used simultaneously with scores of 9.5/10 and 9.4/10 respectively, shown in Figure 3.

Due to the inherent complexity of chemical reactions, supply chain of raw materials, and demands for process optimization and automation, additional LLM-RAG workflows will likely be implemented to solve issues related to the availability of FDA approved drugs. Further improvements in generative ai drug discovery to provide more specific excerpts of text and line numbers guiding ai decision processes will likely also extend to the vast chemical space that requires more exploration. Thus, the Human-AI team will be further strengthened through multiple types of language interactions on a larger scale to further promote drug safety and effectiveness.

July 23, 2024

New AI Drug Discovery  

LLM Drug Discovery Applications