Beyond misconceptions: Exploring the patentability and challenges associated with AI-driven bioinformatics innovations in Europe

An article by Technical Assistant Rebecca Bennett and Partners Anna Gregson and Andrew White outlining the available protection for AI-driven bioinformatics has been featured in the latest edition of The Patent Lawyer.

Read the feature below.

The bioinformatic landscape has undergone a profound shift with the infusion of artificial intelligence (AI) technologies. The integration of AI, and specifically its subsets like machine learning and deep learning, has catalysed a revolution in bioinformatics. This transformative impact is most evident in the enhanced analysis and interpretation of biological data, unlocking invaluable insights from expansive datasets. The synergy between AI and bioinformatics is not only expediting drug discovery but also enriching our understanding of complex biological systems. Today, AI stands as an indispensable instrument within the bioinformatics toolkit, propelling the field into new realms of discovery and innovation.

However, a prevalent misconception has persisted, particularly in Europe, that bioinformatic inventions are inherently unpatentable due to their classification as mathematical methods or computer programs by the European Patent Office (EPO) and as such fall within excluded subject matter (Article 52 European Patent Convention (EPC)). This has led to the belief among inventors that their innovations may be ineligible for patent protection. While it is true that mathematical methods and computer programs, in isolation, are excluded from patentability, it is essential to recognise that many AI-driven inventions, particularly those in the dynamic field of bioinformatics, do not fit this restrictive narrative and are not classed as excluded subject matter.

This misconception is evidently mirrored in the discernible contrast in the number of patent applications across various jurisdictions. In 2022, both the United States and China saw a substantial surge in bioinformatics-related patents, with over 4000 published patent applications each. In stark contrast, Europe recorded roughly half that number, totalling just over 2000 published patent applications in the same year. This marked divergence in patent activity may be attributed, in part, to another prevalent misperception — the belief that the United States is more permissive when it comes to software-related inventions, such as those at the intersection of bioinformatics and AI implementation. It’s essential to note that while the United States Patent and Trademark Office (USPTO) may have exhibited a historically more permissive stance toward software-related innovations, including those in bioinformatics and AI, the landscape has significantly evolved in the past decade. A discernible shift towards a more rigorous approach has emerged.

The way that many AI-related bioinformatics inventions are assessed for patentability in Europe is often under the guise of inventive step.  In performing this assessment, when considering whether any features contribute to the presence of an inventive step, the EPO will assess whether those features are “technical” or provide a technical contribution in the sense that they provide a technical solution to a technical problem.  There is no explicit definition of what is considered “technical” or not, rather this is framed by case law of the EPO Boards of Appeal built up over the last few decades.

It is often in this domain where we find that the argument on patentability can be won or lost.  To support arguments on inventive step, it is helpful to be able to refer to statements in the application as originally filed of technical advantage associated with any features that are being argued over.  Therefore, care should be taken at the time of drafting to gain input from a European practitioner, experienced and knowledgeable on EPO case law, as to what may help support the presence of an inventive step and therefore the patentability of the invention in Europe.  Adding in statements in a letter of response to the EPO is often not as effective as including such statements in the application as originally filed.

Helpfully, the EPO offers illustrative instances of AI applications across various technology fields to help explain what may be considered “technical” or not. For instance, utilizing a neural network in a heart monitoring device to identify irregular heartbeats constitutes a clear technical contribution. Similarly, the classification of digital images based on low-level features, such as edges or pixel attributes for images, exemplifies typical technical applications of classification algorithms. Therefore, although this a hurdle that an invention has to overcome, there are clear circumstances and approaches to meet this requirement.

Additional challenges arise when assessing whether a patent application provides an enabling disclosure of an invention. In particular, according to the EPC, a fundamental requirement is that “The European patent application shall disclose the invention in a manner sufficiently clear and complete for it to be carried out by a person skilled in the art.”. This sufficiency criterion gains heightened significance in the context of patents that encompass machine learning technologies. In some cases, this may require the description and or disclosure of training data, if these cannot be generalised in some way that is reproduceable by the skilled person.

As European practitioners working with bioinformatics inventions, we witness first hand compelling instances of successful bioinformatic applications implementing AI culminating in granted patents. A prime illustration of this lies in the field of genetics. The EPO has rendered affirmative judgments on inventions pertaining to computational methodologies for detecting mutations, including Single Nucleotide Polymorphisms (SNPs) and Single Nucleotide Variations (SNVs). Furthermore, the EPO has extended its positive rulings to encompass predictive technologies for phenotypic traits—ranging from symptoms and their severity to disease prognosis—based on viral genotypes. These instances underscore the EPO’s recognition of the technical merit and innovative impact of AI-driven bioinformatic solutions, offering encouragement and assurance to companies operating in these cutting-edge domains.

In conclusion, the prevailing misconceptions surrounding bioinformatics inventions, especially those implementing AI, are often unfounded, as these innovations are indeed patentable in Europe. The notion that they face insurmountable obstacles is often unsupported.