Professor Jackie Hunter looks at why the pharmaceutical industry has been so slow to adopt artificial intelligence-based technologies and makes a prediction for the industry’s future, including a significantly increased amount of interplay between large technology and pharma companies.
The past two decades have seen an explosion of biomedical data, making it impossible for even the most learned researchers to process and garner real insight from this wealth of information
There is no doubt that the pharmaceutical industry is ripe for disruption. Costs over the past two decades have soared – it currently costs USD 2.6bn to develop a drug – but the industry has not delivered high levels of innovation in return. True, important new medicines have been delivered, especially in biological areas such as immune-oncology and multiple sclerosis, but other areas such as dementia and brain cancer have not seen therapeutic advances.
The past two decades have also seen an explosion of biomedical data, making it impossible for even the most learned researchers to process and garner real insight from this wealth of information. It is now an industry imperative to find ways to harness and use that data to create new knowledge. Similarly, CROs have failed to significantly improve operational efficiency in clinical trials, with many trials failing to recruit the required numbers of patients within the original timeframe.
However, with the advent of electronic health records and patient registries, the ability to mine this data for appropriate trial participants should lead to significantly quicker recruitment and lower screen failure rates. Artificial intelligence, from machine learning to deep neural networks, is a technology that is perfectly suited to this task, and its adoption by the industry should enable companies to reduce costs and improve success rates in drug discovery and development.
So why is it taking so long for companies to embrace this technology wholeheartedly? Technology companies like Alphabet, Microsoft, Apple and Amazon have developed a strong presence in healthcare, primarily from a healthcare delivery and management perspective. However, there is clear evidence that they are beginning to turn their attention to drug discovery and development. For instance, Google’s DeepMind has developed AlphaFold which uses vast amounts of genomic data to predict protein structure solely from its genetic sequence. Their AI-derived 3D predictions were far more accurate than anything that had been previously developed and DeepMind have emphasised the impact of this technology on drug discovery and development. Clearly large pharmaceutical companies will either have to build these capabilities and platforms internally or develop partnerships and acquisitions to access them – a good examples being Roche’s recent acquisition of Flatiron Health and GSK’s investment and partnership with 23andMe.
What have been the barriers to the adoption of AI by traditional companies? Well one barrier is the fact that in many companies, data is not stored in one place or in one format. The huge numbers of mergers and acquisitions that have occurred in the industry over the past 20 years have meant that many companies do not have the systems and infrastructure in place to make all their data available even internally, let alone to collaborators. Implementing these systems is costly – one mid-sized pharmaceutical company spent USD 200m federating all its clinical data. But this federating is essential to reap the full benefits of AI.
A second barrier is partly organisational, and partly cultural. Although most companies have a chief digital officer and data scientists, these positions are not necessarily embedded alongside the drug discovery and development experts. To really reap the benefit of AI and other digital technologies, data scientists, engineers and user interface designers ideally should be firmly embedded with the end users, working in cross-functional teams.
Until recently, people were also sceptical that this technology had been over-hyped, but there are now many examples of where AI is really impacting drug discovery. BenevolentAI, and, more recently this September, the Canadian company Deep Genomics, have both shown that AI can predict novel therapeutic targets which have then been validated. Companies like Ex-Scientia and In Silico Medicine are focussed on drug design and have demonstrated that AI can significantly shorten the time to find candidate molecules. In terms of clinical development, the benefits are also being seen – the patient data that Flatiron Health provided to Roche allowed for earlier approval of their drug in a number of countries.
So where will the industry be in 5 years’ time? I think we will see even more interesting new business models and partnerships. There will be significant interplay between the large technology and pharma companies, for example, Emma Walmsley – the CEO of GSK – has recently joined the Board of Microsoft. I am sure we will see the first AI-enabled drugs in clinical trials and hopefully clinical validation. But to ensure that this technology delivers, a close relationship between data scientists and biologists, chemists and clinicians is essential so that the right data is used, the right questions are asked and the right patients selected. Ultimately, I believe this will fuel innovation in the healthcare industry, whether from traditional or non-traditional players, and that has got to be good for society in general and patients in particular.
