Artificial Intelligence or AI has been touted as the groundbreaking approach to more efficient and cost-effective drug discovery. At its core, AI is a combination of several computational techniques that require programming and training, that can rapidly analyze enormous datasets. However, it is important to note that output from an AI platform will only be good as the algorithms and size and quality of input datasets1. Given that the drug development consists of several steps where each step generates large amounts of data, AI applications will help streamline the process and potentially cut time and costs. An added benefit is that AI will help minimize human inefficiency and errors that will help standardize the drug design, screening and validation process2 and AI can also help weed out drug candidates that are likely going to fail in downstream validation. This allows drug developers to focus on viable candidates that have a higher likelihood of success.
Recently, several startup companies have been developing cutting-edge AI methods and the pharmaceutical industry has been quick to leverage the available expertise via large dollar collaborations. One example is Sanofi who has announced a couple of large AI collaborations – in January 2022, Sanofi expanded a partnership with Exscientia to deliver up to 15 new targets in oncology and immunology for an upfront payment of $100 million3. If the candidate search shows clinical and commercial success, the deal could net Exscientia up to $5.2 billion. Additionally, Sanofi recently signed a deal with Atomwise, an AI company that has a proprietary platform for structure-based drug design, for $20 million upfront and up to $1.2 billion if the program shows success4. Not be outdone, Merck has teamed up with Absci in a deal that is valued up to $610 million to use their Integrated Drug Creation platform to identify 3 disease targets along with therapies for those targets5. Amgen has teamed up with Generate Biomedicines, an AI company that is generating a lot of interest, to identify multi-specific drugs across various disease indications6. Similar to other AI deals, Amgen has committed to paying $50 million upfront and up to $1.9 billion in milestones if the targets achieve success6.
These collaborations seem to follow a similar pattern where pharma companies essentially fund small AI companies to refine and test their programs upfront with the promise of huge payouts if the AI platform generates viable candidates that show clinical and commercial success. This suggests that AI driven drug discovery is considered to be in its early days, especially since there are no data as yet to show that AI methods do result in more effective and cheaper drugs. Indeed, a poll by a pharma trade magazine showed that about a third of respondents believe that AI will peak after about a decade7.
One area where AI seems to have a more widespread effect is diagnostics. The most commonly used diagnostic method is pathology based where tissue samples are histologically analyzed manually. Manual diagnosis is time consuming and introduces human error due to subjective analysis of specific tissue sections. AI based methods have the potential to speed up accurate diagnosis, reduce human error and provide insights into disease biology8. Digital pathology has made significant strides in recent years and complete digital pathology workflow systems that have been approved by the FDA are available9. Advances in digital pathology-based diagnoses have been seen in the cancer space and this has helped pathologists provide more accurate diagnoses as well as assess biomarker expression for targeted therapies9. It is evident that AI will continue to advance precise diagnostics in order to support targeted therapies and precision medicine.