AI is revolutionising drug discovery - so why are approval rates still falling?

Artificial intelligence is transforming our understanding of countless sectors at a remarkable rate. This is particularly true of drug discovery, where AI is revolutionising a usually drawn-out operation. Traditionally, the discovery process is a costly and time-consuming one with uneven success rates. Each step is manually undertaken and slow to come to fruition.

Now, AI is altering the landscape and its integration is set to grow. The global market for AI in drug discovery is expected to grow at a compound annual growth rate (CAGR) of 41% from 2022 to 2030, according to a report by Vantage Market Research. In tandem, it’s projected to be worth $30 billion by 2030.

Despite consistent progress in discovery spaces, the number of drugs approved by the FDA has been declining since 1995. In 2022, 37 novel drugs were greenlit by the board - a decrease compared to 2021’s 50 and 53 in 2020 - and the lowest annual approval rate since 2016. 

Currently, it feels like you read more articles about a new advancement in AI drug discovery, rather than stories focused on actual drugs that were discovered through AI. Clearly, the tools are becoming more and more advanced yet, comparatively, there is little to show for it. The spike in approvals we may have expected has not yet materialised. It’s worth pondering why this is happening - and when approvals may start to catch up.

New meets old

For many years, drug discovery has been an arduous and complex operation. This is, of course, not surprising. Scientists must carry out rigorous testing at each stage to ensure a drug is safe and fit for approval by the FDA. However, this does mean that the pathway to market is extremely slow.

AI is changing that. Swathes of data can be analysed rapidly, much quicker than what can be achieved manually, so that each stage of the process can be completed with more time saved. Not only that, but AI offers higher quality assurance, meaning that less mistakes, and therefore delays, occur. As a result, when the time comes for clinical trials, they are set up more efficiently and have greater chance of success, to a large extent due to AI’s thoroughness earlier in the process.

Questions to answer

To gain the trust of regulatory bodies such as the FDA, concerns must be addressed before AI can be utilised to its full potential in drug discovery. Like all technology, ethical considerations - such as privacy, data sharing and transparency - must be examined at each step to ensure legitimacy.

Unsurprisingly, AI models are often challenging to interpret, which can cause confusion when pinning down how a model arrived at a specific conclusion. A similar issue arises with data reliability; datasets in drug discovery must be complete and unbiased to provide accurate predictions. With AI still very much an unknown quantity for many, regulators may also be cautious. Officials may require more detailed validation to greenlight certain testing; this can add time and strain to an already busy workload.

The future of drug discovery

The vast potential of AI certainly points to a promising new frontier for drug discovery. But to advance drug approvals, and not just develop increasingly sophisticated tech for its detection, those at the forefront must commit to sharing data with their peers - and regulators. Greater transparency of research allows challenges to be resolved much more swiftly, as well as improving the overall understanding of complex methods at play.

As is often the case, collaboration will help us arrive at an end goal in quicker, more effective fashion. With officials kept informed, and concerns around AI addressed, trust will begin to grow in these evolving processes, with more approvals surely to follow.

The bottom line

The possible benefits of AI-supported drug discovery remain innumerable. But for regulators - and FDA approval rates - to catch up with such rapid progress, more work must be done to instil trust in the discovery procedure. Only then will we see medicines in the hands of those who critically need them.