AI as a tool for drug development
Artificial intelligence is about to become a transformative tool in pharmaceutical research, helping to streamline the journey from the lab to the pharmacy shelves. With AI, the possibilities seem endless. Applications today span from tailoring personalised medicine and forecasting the success rates of drugs in clinical trials to identifying the most effective drug combinations. BioStock has taken a closer look at what opportunities the technology offers.
During the 2023 Stanford Drug Discovery Symposium, Najat Khan, Chief Data Science Officer at Janssen, a subsidiary of Johnson & Johnson, concluded that we are at a tipping point with artificial intelligence (AI) adoption, particularly in drug development.
Echoing this sentiment, UK-based biotech firm Exscientia delivered a closing keynote at the 2023 Society for Laboratory Automation and Screening Conference in Brussels, predicting that the design of all new drug candidates will be augmented with AI within this decade.
The emergence of AI in life science
The enthusiasm is justified, not least because it addresses a well-known problem in the sector. According to a 2018 MIT study, 40 per cent of phase III drugs fail, contributing to 75 per cent of R&D costs. And this is where AI could play a key role. Read more in BioStock’s article series on drug development, here.
Morgan Stanley Research suggests that already modest improvements in early-stage drug development success rates, facilitated by AI and machine learning, might result in an additional 50 treatments and a 50 billion USD market over a decade. Statistics in the coming years will reveal if AI has succeeded in cutting cross-sector clinical trial failures and lowering R&D costs. According to Markets and Markets, AI in the health market is projected to grow from 15.6 Billion USD in 2023 to 102.7 Billion by 2028.
Optimise clinical studies
AI can accelerate drug discovery by analysing extensive and complex data sets such as biological measuring points and data from clinical studies. This can increase the chances of identifying promising candidates, optimizing its properties and predicting the drug’s potential ability to achieve the desired therapeutic effect.
Another possible use is to identify candidates early on who are unlikely to receive from government agencies. AI can also facilitate the development of precision medicine by tailoring treatments based on patient data.
Companies active in AI
As seen in the examples above, AI has transcended its buzzword status to become a seismic force reshaping the contours of life science.
Take Atomwise, a US-based pioneer, for instance. With a library of over three trillion synthesizable compounds, their AtomNet technology can screen molecules for therapeutic potential at an unprecedented speed. This is not just theoretical – in on day, Atomwise identified two drugs that could combat Ebola, a feat that would traditionally take years.
Another great example is DeepMind, a Google affiliate, which has developed AlphaFold, an AI system that has accurately predicted the 3D structures of over 200 million proteins. While this may not directly involve drug discovery, understanding these protein structures is a cornerstone for drug development. Insilico Medicine, based in Hong Kong, leveraged this very database to design a potential drug for hepatocellular carcinoma in just 30 days. They even employ AI chatbots to engage with researchers, adding another layer of efficiency to the drug discovery process.
In the UK, Exscientia uses AI to emulate human creativity in analysing large datasets and designing unique disease-treating compounds. They claim AI reduces the typical 2,500-compound test list to around 250.
Similarly, Healx focuses on repurposing existing drugs to treat rare diseases. Their AI platform, HealNet, sifts through a myriad of disease insights to predict the most promising drugs and combination treatments.
BenevolentAI is another key player in the UK, partnering with pharmaceutical giants like AstraZeneca, GSK, and Novartis. They are using AI for better drug target selection—a critical step in drug discovery.
Across the Channel, Paris-based Iktos specialises in AI for de novo drug design. They focus on generative modelling with built-in synthetic accessibility for drug discovery and have formed partnerships with Japanese Ono Pharma and British Sygnature Discovery.
Not to be outdone, Janssen, is applying AI across the entire drug development value chain. From drug discovery and clinical trial design to patient identification and manufacturing optimisation, Janssen has over 100 ongoing AI projects.
Even Scandinavian companies are making their mark. Evaxion Biotech uses AI to decode the human immune system for personalised cancer treatments, while Cline Scientific’s AI tool, CellRACE, predicts cancer metastasis potential. 2cureX employs AI for drug sensitivity tests, and medtech firm Elekta uses AI in oncology.
To be clear, the number of life science companies using AI around the world is in the hundreds if not thousands, ranging from pharmaceutical giants to specialised startups. This is only a snapshot and not an extensive overview.
The future and ethical considerations
As AI continues to evolve and integrate with other cutting-edge technologies, the potential to develop safer, more effective, and personalised medicines increases. Exactly how far the technology will reach is not easy to guess.
However, it is worth noting that alongside all optimism, there are also valid concerns about the ethical and regulatory implications of AI, which the industry must also address while in its current upward trajectory.
This will require collaboration between researchers, regulators, and policymakers to set guidelines and ensure that the technology is used ethically. This will certainly be a challenge, just like in all other areas where AI is now making rapid progress.
Yet, progress is constantly being made, both in terms of technological development and in regulating the forms of use of its pioneering tools.
