The process to bring a new drug to market usually takes over a decade and costs around $2 billion dollars. The lengthy process includes discovering new molecules, synthesizing the lead molecules by chemical or biological means, conducting experiments in rodents and finally in humans. Authority regulators then take in these results and decide if the drug is safe to public. With the aid of machining learning and generative AI, this process can be greatly shortened to 2 years. Therefore, academia and pharmaceutical companies are embracing artificial intelligence in the field of drug discovery.
Traditionally finding a new drug involves identifying proteins whose roles in a disease are clear. Researchers then study the structures of such proteins and find pockets in them that can be targeted. Then they find or design molecules that can modify the proteins' activities by fitting into the proteins' pockets. This process is laborious and expensive and usually involves tweaking hundreds or even thousands of compounds to test. With the help of machine learning models, this preclinical process is sped up and the optimized molecules can be found within a significantly shortened time frame.
Sources:
- https://cen.acs.org/physical-chemistry/computational-chemistry/AI-taking-over-step-drug/103/web/2025/10
