First published with Lawtext Publishing's Bio-Science Law Review - Volume 18 - Issue 6
AI technology has been around for about 80 years and has gone through several cycles of optimism followed by disillusionment. Yet recent years have seen unprecedent growth of this technology in terms of commercial applications. Global private investment in AI was over USD 93 billion in 2021, over double that of the previous year. In line with this, patent filings have skyrocketed, with patent filings in 2021 30 times higher than in 2015. While ‘typical’ applications of AI in areas like image processing and speech recognition still dominate the field in general and patenting activity in particular, the potential of AI in the life sciences is being recognised. For example, there is a huge boom in the use of AI technology aiming to shorten development cycles in drug discovery, with corresponding investment activity. What is more, the rise of AI in the life sciences is not limited to drug discovery, with huge strides being made in other areas, for example protein structure prediction. Yet this trend is not visible in patent filings, at least not yet.
While life science AI has not yet hit patenting statistics in a big way, there is increasing interest in this cross-disciplinary area, with innovators not only looking for help with patenting per se but also strategic guidance.
This article explores a number of important questions, including:
● Is AI-based life science innovation patentable?
● When should AI-based life science innovation be patented?
● What about patenting the innovations made using AI?
These questions will be approached from a perspective based on practice before the European Patent Office. However, since these questions are clearly important beyond Europe, the article will also consider international perspectives where appropriate and in particular with a view to what is happening in the United States.
Is AI-based Life Science Innovation Patentable?
The short answer is yes. As ever, it is a little more complicated than that. AI methods are considered mathematical methods by the EPO. Mathematical methods per se, like computer programs per se, are excluded from patentability. However, this exclusion applies only narrowly, so as soon as a computer or other technical means is involved, patenting becomes possible. This is not quite as liberal as it sounds, because the AI or other mathematical method per se is considered by the EPO to be non-technical and is not taken into account when assessing the non-obviousness (inventive step) of subject matter to be protected in a patent application. However, if the AI method makes a contribution to the technical character of the subject matter, then this is not the case and the AI method can contribute to the patentability of the subject matter.
Considering the field of life sciences, the most common way in which an AI method can contribute to the technical character is by contributing to a technical effect that serves a technical purpose by its application to a field of technology.
Much therefore turns on whether an application of an AI method in the life science context is considered to provide an application to a field of technology. The development of drugs and treatments, or new methods of diagnosis are all considered fields of technology by the EPO. In all of these there is a tangible outcome, such as a drug, treatment or diagnosis. However, the EPO goes further and considers some forms of data analysis to define technological fields. The best example of this is genetics, which is very clearly considered a technical field, with the EPO Boards of Appeal having ruled positively on inventions relating to improved genotyping methods, where the innovation lay in the statistical analysis providing confidence bounds or in methods for correcting the genetic input data. Other examples from the case law of the EPO Boards of Appeal include monitoring physiological parameters (here heart time volume) based on measurements (here distal blood pressure) and even calculating a risk score for diabetics based on blood sugar measurements. Albeit in a different context, the EPO’s Guidelines for examination make it clear that microbiological processes and biotechnology more generally are considered fields of technology by the EPO. While there is no specific case law for these observations, it is clear that activities like analysing electronic health records in specific ways to stratify or triage patients are considered to be technical activities that enable an AI method to contribute to the technical character of an invention.
In short, applications of AI in different fields of the life sciences can indeed be patented at the EPO. Of course, the general rules of patenting apply in that the patent application must disclose the invention in sufficient detail to enable it to be practised, and the invention must be new and involve an inventive step over the prior art. As will be clear from the above, an inventive step can arise in numerous fields of application of AI technology in the life sciences.
What about other jurisdictions?
Even within Europe, there are subtle differences between different jurisdictions, although generally the courts of the Member States of the EPO try to keep a degree of consistency in their approach, with decisions by other courts or the EPO’s Boards of Appeal generally taken into account. Further afield, again there are both commonalities and differences. For example, the US standard for an invention to be eligible for patenting is to ask whether the invention relates to an abstract concept and, if it does, whether there is significantly more than the abstract concept per se, in particular an application to a field of technology. Although there are differences in the way these concepts are applied in Europe and the United States, in most cases a similar outcome can be expected. There are, however, some notable exceptions. For example, diagnostic methods are typically not patentable in the United States (at least not per se), whereas they are considered technical and patentable at the EPO and many other jurisdictions, for example Canada and Australia. Another example is methods of genetic analysis, like genotyping, which are considered technical at the EPO. In contrast to this, a recent case at the Federal Court of Appeals in the United States looked at a method of haplotype analysis and concluded that this is merely an abstract statistical analysis. While there is no case law in Europe specifically on haplotype analysis, it seems highly likely that such a method would be patentable, being a specific form of genotyping.
In conclusion, while there are variations between specific technological fields across global jurisdictions, the question of whether AI-based innovation in the life sciences can be patented can be answered in the positive. The next question is whether and when it should be.
Please contact Alexander Korenberg or Sarah Lau if you have any questions or would like to discuss this topic further.
Subscribe to AI musings
If you want to find out more and to be kept up to date with the latest developments in the field of AI and Machine Learning, view Alexander’s monthly newsletter on patenting AI here.