The use of Artificial Intelligence (AI) in drug discovery has spawned a fast-growing biopharmaceutical segment that is estimated to be $1.5B in 2024 and expected to grow to $14B in 20321. There are innumerable reports and publications on how AI enables drug discovery in specific areas. One segment is the identification of drug targets – AI algorithms can analyze available data and identify patterns and trends much faster than humans but large unbiased datasets are needed to train the algorithm to be effective. For example, if high throughput small molecule screening data is available for a disease target, then that dataset can be used to train the AI platform to identify a manageable number of more potent and specific compounds for that target. Essentially, AI can help prioritize quality over quantity at the hit screening stage. Other areas where AI can have a large impact is around study planning and process workflows. One large area is the optimization of clinical trials including the identification of relevant patient populations and predicting patient response1. Another critical area, especially for advanced modalities such as cell and gene therapies, is supply chain optimization around demand forecasts and inventory management1. It is not uncommon for drug developers to face challenges around regulatory filings that have complex procedures1 and AI can enable the automation of regulatory documentation especially for complex modalities.

There have been several successful reported milestones of AI-enabled drug discovery and at least 2 reports of AI-designed drugs entering clinical trials. The first AI-designed drug to enter clinical trials was developed by Exscientia in collaboration with the Japanese pharma company Sumitomo Dainippon, which was a serotonin 5-HT1a receptor agonist designed to treat obsessive compulsive disorder2. Though the drug failed in Phase I as it did not meet success criteria2, the program is a landmark achievement in AI-enabled drug discovery. Another promising example is INS018-055, a novel small molecule inhibitor of the TNIK enzyme that is implicated in fibrosis3. It is important to note that the Generative AI used by InSilico Medicine to identify this candidate also identified the novel disease target. The novel small molecule has received Orphan Drug Designation from the FDA and is in Phase II clinical trials for the treatment of Idiopathic Pulmonary Fibrosis (IPF), a rare incurable disease4. Additionally, AI is being deployed to predict structures for all known proteins and this project is called the AlphaFold Protein Structure database5, and it is expected that AI will be deployed in several areas of basic research to identify new drug targets and signaling mechanisms.

The preclinical success of AI-enabled therapies has fueled significant adoption across the drug discovery community – a recent publication from a consulting group reported that since 2015, 75 drugs have entered the clinic and about 90% of those therapies are still in the clinic6. The therapies in the clinic are a mix of AI-discovered small molecules, antibodies and vaccines as well as repurposed drugs6, which is a positive indication that AI-enabled drug discovery can be used across multiple modalities. However, one interesting caveat is that the disease target was known for most of the therapies identified using AI7, which suggests that fundamental new target identification studies using conventional methods are required to support AI-enabled drug design. Nevertheless, the growing success of AI in the identification of novel and next-gen therapies suggests that the time and cost associated with preclinical drug discovery is likely to decrease, so new therapies may be a lot closer to patients.









Leave a Reply

Your email address will not be published. Required fields are marked *