AI technologies are catalysing the initial and most crucial step in the biopharmaceutical value chain.
The process of drug discovery has been historically slow, labour-intensive, failure-prone, and costly. Its four main stages, as shown below, typically take around five to six years to attain completion.
This is a huge amount of time, especially during crisis situations such as the COVID-19 pandemic and considering the fact that drug research and discovery is only the first step in the biopharmaceutical value chain — all in all it would take about a decade to finish the entirety of this process.
Looking back in the past, discoveries were made mostly due to accidents and unexpected observations, like that of penicillin. In recent years, advancements in science and technology made drug discovery an actual paradigm instead of merely a stroke of luck.
Now, AI-enabled solutions are emerging to further improve drug discovery — reducing timelines, increasing accuracy, and improving value.
With so much medical information present in databases and other sources, it is easy to get lost. AI can help in making the research process easier.
BenevolentAI is a platform which takes in as output various sources such as research papers, patient records, and patents in order to generate knowledge graphs. Using natural language processing, the AI system can identify information links that researchers might have missed out.
“AI can put all this data in context and surface the most salient information for drug-discovery scientists,” according to Jackie Hunter, chief executive of BenevolentBio, the company responsible for BenevolentAI.
Recurrent Neural Networks (RNN) and other deep learning methods are increasingly employed in drug discovery in order to generate drug molecules subject to specific rules or restrictions. This way, adverse effects of drug interactions — a common problem among the elderly who are prescribed multiple medications — can be avoided from the get-go.
Apart from avoiding unwanted drug interactions, deep learning methods can also be used to create drug molecules with custom properties. If developed further, this may open the doors to personalised medicine in the future.
British start-up Exscientia and Japanese pharmaceutical firm Sumitomo Dainippon Pharma recently created with the assistance of AI a drug that can potentially treat patients diagnosed with obsessive-compulsive disorder (OCD).
To do this, the AI algorithm considered a massive number of potential compounds, checking each one against a set of specified parameters. The development of this drug only took a year, way faster than the typical development span of five years or more.
The AI system can also be used to develop drugs for other diseases. This agnostic nature of AI algorithms makes them extremely versatile in finding new drug molecules.
It may only be a matter of time until we see AI take over the process of drug discovery in the biopharmaceutical industry, if not the entire value chain. Who knows, in the years to come, potentially the majority of the drugs that will be released in the market are created with the assistance of AI tools.