AI technologies are being rapidly implemented into drug development pathways, and large pharmaceutical companies are collaborating with smaller technology groups and new start-ups to do so. Whilst these technologies are reducing the cost and time-frame associated with developing a medicine, does AI have the potential to limit a company’s monopoly right over drug discoveries?
There is no denying that drug development is time consuming and expensive. According to several industry studies on the drug approval process, it takes on average twelve and a half years from the discovery of a drug to it being put on the market, and this process is estimated to cost £1.15 billion.
Considering the time and expense associated with producing a medicine, it is understandable that companies are looking for ways to streamline the research and development process. The drug development industry has started to use AI and machine learning for a range of functions, often through collaborations with small research groups and technology companies. Examples include:
1. Reducing development timelines. Exscientia is an AI-driven drug development spin-out from Dundee University. In March this year, Exscientia entered into a multi-million pound partnership with Celgene, which will use an AI platform to identify therapeutic targets in the areas of oncology and autoimmunity. Exscientia’s platform has been shown to reduce timelines for the development of drug candidates in advance of clinical trials by 75%.
2. Profiling molecules. In August 2018, bioinformatics company Cyclica announced a collaboration with WuXi to investigate molecular targets using an AI screening platform. This collaboration will also focus on identifying any secondary interactions of the molecules within the body, known as “polypharmacology”. The partnership hopes to develop a model to assist with the design and screening of medicines by identifying molecule properties such as metabolism rates and levels of toxicity.
3. Developing personalised treatments. In January this year, Novartis and the University of Oxford’s Big Data Institute announced that they would establish a five-year research alliance. The project will interpret data from around five million people to identify methods which predict individual responses to therapies for inflammatory diseases such as psoriasis and rheumatoid arthritis. It is hoped that this research will help with the development of targeted treatments for patients.
Whilst AI is improving many aspects of drug development, it may also affect the traditional methods of protecting these discoveries. Once a drug has been developed, a company may look to protect it with a patent. A drug patent prevents other companies from manufacturing and selling the medicine for a period of time. Patents have traditionally played an important role in incentivising research and development, as this exclusivity allows companies to earn back the money spent during the research process. However, the implementation of AI may force drug development companies to change their patenting strategies.
At present, there is no legislation which deals specifically with the patenting of AI-enabled inventions. Under current UK law, the owner of a patent will be the actual deviser of the inventive concept.However, research collaborations may face the competing interests of different individuals involved in the development process alongside an AI system. The question for the courts applying the existing law is this: who should own the right to a monopoly on discoveries made by an AI system? Should the inventor of an AI-enabled drug discovery be the developer of the AI; the individual who identifies a use for the AI; or the company that uses the AI to select a medicine? Rather than leave these challenging questions to the courts to resolve, a much better solution would be for collaborators to agree up front who should own any rights in inventions arising out of a particular project.
Questions remain about whether there could be scenarios in which the role of the AI is so dominant that there is no actual deviser of the invention and therefore no individual is entitled to the patent. For an industry incentivised by the ability to have an exclusive right to a pharmaceutical product, this could be incredibly problematic.
At present, the answer to this question is unclear. Whilst this issue is unlikely to be a substantial problem within the context of current AI technologies which require substantial levels of human interaction, as AI technologies become more advanced, this is likely to become a more pressing issue in the future.
The growing implementation of AI has the potential to improve drug discovery processes to ensure faster, cheaper and safer treatments. Nevertheless, we must wait to see how the UK courts rule on these inventions, to ensure that we continue to incentivise the development of drug therapies and the use of AI in this sector.