One of the biggest tech trends of 2017 has seen pharma companies fully embracing AI in all areas of drug development
Simply put, artificial intelligence (AI) is the ability of machines to process information, analyse and make decisions, ie to think, learn and adapt.
Unbeknownst to most, AI is all around us, as exemplified by Amazon’s Alexa and iPhone’s Siri. Many online Live Chats are not manned by people but by ‘chat bots’. These exemplars employ behavioural algorithms acting as automated responders. Purists don’t consider these algorithm-dependent systems to be true AI, as they don’t possess the ability to improve their capabilities or knowledge over time.
Silicon Valley giants are not the only enterprises which are exploiting this new technology. The pharmaceutical industry has made embryonic forays into the field. British titans GlaxoSmithKline and AstraZeneca are amongst the vanguards, making significant AI investments to speed up drug discovery and clinical development.
Traditionally drug discovery is a laborious and capital-intensive process, with researchers testing a bank of molecular compounds against potential disease-related targets. It is mostly a ‘hit-and-miss’ endeavour. The process has been speeded up with the use of robotics, thus making high-throughput screening (HTS) possible. HTS allows millions of pharmacological tests to be performed rapidly to identify potential active compounds quicker.
Likewise, the FDA and EMA introduced ‘adaptive clinical trials’ to speed up studies and increase their success rate. This is done by modifying the protocol, statistical methodology, dosage or target population, in real time, using the knowledge gathered from pre-specified interim analysis.
However, HTS and adaptive trials have yet to lead to the breakthrough that many had hoped for. Will AI lead the industry to the ‘promised land’ by bringing new medicines to the market in a speedier and less costly manner?
Discovering new medicines
In July, GSK announced a preclinical collaboration with Exscientia, a Dundee-based company at the forefront of AI driven drug discovery. The Scottish company announced that it “will apply its AI enabled platform and combine this with the expertise of GSK, in order to discover novel and selective small molecules for up to 10 disease-related targets, nominated by GSK across multiple therapeutic areas”. Exscientia, spun-off by the University of Dundee in 2012, said that “AI driven systems actively learn best practice from vast repositories of discovery data and are further enhanced with knowledge acquired from seasoned drug hunters”.
Exscientia uses AI to design new molecules whilst automating and speeding up the entire drug discovery process, thus cutting costs simultaneously. Its CEO, Prof Andrew Hopkins, believed that it is possible to identify candidate molecules in a quarter of the usual time and at a quarter of the cost. Using big data analytics, it aims to examine existing large data sets from varied sources to understand correlations between candidates and targets, allowing it to ‘design millions of novel, project-specific compounds and pre-assess each for predicted potency, selectivity, pharmacokinetics and other key criteria’. At the experimentation phase, it will select ‘the best, information-rich compounds for synthesis and assay’. It then optimises the initial chemical designs using what it calls rapid ‘design-make-test’ cycles to develop novel compounds for their clients.
Developing new medicines
In contrast, AZ has decided to invest in AI at the clinical development stages. It intends to use a clinical informatics platform called iDecide, which will allow it to interpret efficacy results and safety signals rapidly.
iDecide is an innovative collaboration with iconic British institutions, namely Cancer Research UK, the world renowned oncology centre Christie in Manchester, and the University of Manchester.
The platform works by rapidly collating raw efficacy and safety data from phase I and II trials into easily understandable formats, including analysis of potential safety signals, thus allowing quicker progression to phase III studies. It also allows patients to express their trial experiences by sending video and audio clips anonymously online, thereby giving researchers additional insights into patient benefits and side effects. Finally, it informs scientists about current worldwide research, enabling them to incorporate the acquired information into existing studies in real time.
AZ hopes that these investments will enable it to shorten clinical development and bring new medicines to markets earlier while reducing costs.
Reinventing the Wheel?
Founded in 2013, London-based BenevolentAI advocates a different approach. It intends to mine the universe of biomedical data that is in the public domain and identify new uses for ‘old’ drugs, while designing new molecules as well. It aims to find new disease targets for molecules which were considered failures in another disease area.
“Many compounds go into clinic testing in volunteers for a particular disease and don’t work because the hypothesis was wrong,” said Jackie Hunter, CEO of BenevolentBio, its drug discovery arm. “Our system provides unbiased hypotheses. We can take that molecule and find a new disease to target and go straight to a clinical trial without repeating all the previous early testing that’s been done.”
Like the aforementioned collaborations, BenevolentAI is working with another British institution. In May, the University of Sheffield’s Institute for Translational Neurosciences announced that a molecule discovered by BenevolentAI prevented the death of motor neurons in patient cell models. The institute is exploring the suitability of taking the molecule into clinical development for motor neurone disease.
Are we there yet?
There have been numerous false dawns which promised speedier drug discovery and development. Will AI-identified molecules and AI-aided development perform at the crucial and costly phase III stage? AI may also lead to redundancies, replacing jobs which would have taken countless man-hours to perform.
However, Hunter of BenevolentBio highlighted that AI is more ‘augmented intelligence’ than artificial intelligence. AI will never replace the ‘hunch’ that successful business people often rely on nor the serendipity of scientists that led to revolutionary discoveries like penicillin – such abilities which would always elude that of a machine.
Stephen Huang is a pharmaceutical medicine consultant and James Huang is a policy researcher at SCP Medical