Protection for Artificial Intelligence in Personalised Medicine – The Patent/Trade Secret Trade Off

Abstract

Personalised medicine is a new and developing field that carries a great potential for future healthcare applications, for prediction, diagnostic and treatment. Personalised medicine allows targeted treatment of different subgroups of patients, making it possible to tailor the treatment to the best-responding patients, while avoiding non-responders or patients that are likely to suffer adverse effects. Data is crucial in the research for personalised medicine as the interesting potential for the clinical use of personalised medicine may become hampered by absence of reliable large data sets. Conclusive evidence on relevant information of molecular mechanisms is not readily available, as most data often comes from inconclusive and relatively small studies of insufficient quality. Once collected, it is claimed that biomedical data becomes often fragmented, as there are numerous IP claims including both patents and trade secret, as well as the reluctance among the researchers in sharing data. Fragmentation in biomedical data is further exasperated by complex legal, moral and ethical questions surrounding biomedical data. While the need to share personal and private genetic information to advance research and industrial development is acknowledged, there are fundamental moral and ethical questions stemming from the general discomfort against exclusive control of an individual rights’ holder over genetic information. As rights overlap, restriction in one right affects access in other right in personalised medicine. Arguably, the recent move by the US courts to strictly regulate the patentability of diagnostic tools in biomedicine to address the problem of the individual rights holder pushed the shift in protection as trade secrets. Moral and ethical questions become even more controversial where machine learning algorithm (AI artificial intelligence) is used in the personalised medicine. AIs and machine learning algorithms are used to discover genes and variants related to specific decisions, to handle next generation sequencing so that they can be analysed in the full context of other genomics and clinical information to drive personalised medicine research. Not only the data sets, the selection of the correct machine algorithms is crucial in their use in personalised medicines. When AIs routinely processes private medical data, to make sure such algorithm based decision making is morally unbiased and ethically correct, the disclosure of the algorithm may be necessary. However, restricting patenting of algorithms for fear of depriving basic research tools from the public, simultaneously creates incentive to protect these algorithms as trade secret. This chapter explores this conundrum of patent and trade secret trade off in the context of AIs used in personalised medicine. Among others, the chapter discusses how these questions are addressed in the current EU law at the interface of GDPR, trade secret directive and biotechnology Directive. The combination of development in the technological sectors – use of AIs in biomedicines and legal changes – restrictive patent eligibility, genetic privacy and expansive trade secret protection could lead to an under-use of valuable biomedical data, which may result in a new type of tragedy of anti-commons. The paper concludes by revisiting the function of patent right and suggests that it may be necessary to introduce additional regulatory measures to mandate ethical disclosure of algorithm for public policy.