This article is part 2 in our series on digital life sciences, following on from AI and software inventions in Biotech.

When speaking to our Biotech clients, we are very commonly asked to advise on whether or not to file new patent applications disclosing AI and other software inventions, whereas this question rarely seems to arise for other Biotech inventions. In addition to the usual considerations, (e.g. is the invention new and non-obvious over what is already public, is there enough supporting data to enable the desired claim scope), there is an added layer of complication when considering computer-implemented inventions: patenting these inventions can be especially challenging, yet as usual requires a detailed disclosure of the invention as part of the patenting process, and the publication of the patent specification usually occurs relatively early in the procedure before the allowable claim scope is identified. Additionally, the patentability of computer-implemented inventions can differ significantly between territories, and notably between the United States and Europe. Furthermore, in this particular technical area, there are other forms of protection available which do not require publicly disclosing the invention.

While each situation is different, below we set out some high-level issues to consider when deciding whether to file a new patent application concerning a computer-implemented invention in the Biotech field.

The ‘pros’ of patenting – reasons to file

Careful patenting strategies for AI and other computational inventions in Biotech can lead to broad granted claims of real commercial value: So long as care is taken when an application is being drafted, it is entirely possible to obtain broad claims in Europe covering many different implementations and iterations of an invention. The availability of Unitary patents also makes it possible to protect an invention across Europe for relatively low cost. This can give patentees a very broad monopoly lasting 20 years from the filing date without significant investment, and sometimes for a much lower overall cost compared to other Biotech inventions.

A robust patent portfolio adds value and helps secure funding for early-stage companies: A well-thought-out patenting strategy is a significant driver of value for companies. Indeed, intangible assets, including patents and other IP, make up an astonishing 90% of the value of companies in the S&P 500¹. Even at the other end of the scale, the importance of IP for startups and other early-stage companies should not be overlooked. According to a 2023 study from the EPO and EUIPO, startups with patent and trademark rights are around 10 times more likely to secure funding².

The difficulties of obtaining US patents are not a fundamental block: Patents are territorial rights and thus infringement of a patent necessarily involves territorial considerations. This can be a little complicated in the field of computer-implemented inventions, where a computational method may be carried out in country X but the information gleaned from that method is used in some further process in country Y (e.g. in the field of diagnostics, the computational analysis of a sample might be carried out overseas and the results communicated digitally to a clinician). In some situations, it may be instructive to consider with a patent attorney whether an EP-focussed drafting and filing strategy is more likely to generate commercial value than a global strategy, or one focussed on the US.

Disclosure of the invention might happen even without a patent application being published: Aside from the usual risks, (e.g. inventors inadvertently disclosing important details in industry conferences or scientific articles), detailed information on proprietary software and computational methods can sometimes be made publicly available by in-house developers using e.g. a GitHub repository to store and make changes to code. Additionally, for some inventions (e.g. in diagnostics), more detailed disclosure may be required by regulatory agencies, such as the FDA, or potential customers may expect a higher level of disclosure e.g. for technical understanding and/or troubleshooting purposes.

Competitors may invent similar – or the same – computational solutions: As discussed below, while copyright and trade secrets do provide some benefits, these rights may not allow a company to prevent third parties from using their invention (except in certain situations in which the code has been copied without permission, which might breach copyright in the code). This can mean that some competitive edge is lost. In the worst-case scenario, a competitor could independently invent and patent the same, or a similar, computational method, which may complicate a companies’ ability to use its own method (prior use rights are limited and country-specific, which can impede expansion). For this reason, investors can sometimes lack confidence in the ability of smaller companies or start-ups to compete and survive if they have no patent protection (e.g. if the company is relying only on trade secrets). Having a strong patent portfolio can help maintain investor confidence, especially in very competitive areas.

Software can sometimes be reverse engineered: In situations where customers are able to handle proprietary software themselves (e.g. a sequence analysis method where customers input sequence reads and the computation happens either on the customer’s own computer or using cloud computing), third parties might be able to reverse engineer the computational method to discover how the method works at least in general terms. If the method is not patented, third parties would potentially then be able to create their own software that performs the same sort of method.

Keep it secret, keep it safe – reasons not to file

Copyright and trade secrets can provide useful protection: Proprietary code may be protected by copyright, which can prevent third parties from copying and using the same code without authorisation. In a similar vein, trade secrets can be used to penalise unwanted disclosure of confidential information, e.g. concerning proprietary computational methods. However, neither of these rights is effective in preventing a third-party from independently developing similar computational methods, and there are certain limitations to trade secrets which make this right unsuitable for some companies. Nonetheless, there are some cases where companies may prefer to rely on e.g. trade secrets over patents, especially in cases where patenting an invention will be unusually challenging.

Infringement can be difficult to detect: Given the nature of computational inventions, third parties and competitors might not disclose detailed information on their own software and computational methods, making infringement difficult to detect. This can make enforcement of granted patents challenging. The territorial nature of patents adds a further layer of complication – though, as discussed above, this can be leveraged to identify territories of most importance for patent protection.

Computational methods evolve quickly: Unlike traditional biotechnology inventions, e.g. an antibody therapeutic, computational methods tend to evolve relatively quickly and indeed can change very quickly over the lifetime of a product. Engineers will regularly make updates, perhaps to make an AI model more efficient or to improve the quality of the output, and so over time the software product might not resemble the method that was claimed in the original patent application, and might not even be encompassed by the claims anymore. Sometimes this means that the useful ‘lifetime’ of a patent in the AI and computational areas of Biotech is much shorter than in the wider Biotech space, and patentees might abandon granted patents before the full 20-year term has elapsed. Additionally, repeated filings may be necessary to keep up with the pace of innovation and updates to the software product, increasing costs.

Different jurisdictions apply different standards to computer-implemented inventions: As seasoned users of the patent system will be aware, there is a lack of harmonisation across patent offices in different jurisdictions. As a consequence, especially when considering computer-implemented inventions, it can be challenging to draft a patent application that will meet the requirements of, e.g., the EPO and USPTO; what is patentable in Europe may not overcome Alice rejections in the US, and vice versa. Sometimes, this can mean that the downsides of disclosing an invention are not outweighed by the potential scope of the monopoly that could be afforded. Applicants should consider consulting patent attorneys or agents in key countries (e.g. at least EP and US, and perhaps also CN and/or JP depending on commercial goals) before an application is filed, to discuss whether the invention is patentable and to ensure a potential application is in good standing for the differing requirements of the EPO, USPTO, CNIPO, JPO, and any other jurisdictions of interest. Though the patentability challenges differ by territory, the inevitable (and usually early) publication of the invention as part of the quid pro quo of the patenting process is a constant and of course has a global prior art effect.

Inappropriate patenting strategies for computational inventions in Biotech can lead to narrow claims of very little real commercial value: Those who have looked at a number of granted European patents for inventions in this field will have seen many with very narrow independent claims, these often reciting limiting features (such as complex mathematical formulae) unlikely to present any meaningful FTO barrier to competitors. This can be the result of poor drafting, and applicants subsequently being hamstrung by the EPO’s restrictive added matter and clarity requirements. Of course, these suboptimal claim scope results should not dissuade companies from filing patent applications for AI and software inventions in Biotech, and should instead motivate them – if they decide to file a patent application – to ensure that they get the best possible commercial value from the application.

Making the decision

Before disclosing proprietary information to third parties (including to investors), companies need to decide whether or not they are going to file a new patent application or applications. Below, we have set out a few basic points that can be used to help companies investigate and balance the positive and negative factors identified above, to facilitate their decision on whether or not to file a patent application.

1. Understand the nature of the likely non-patent publications of the invention and their timeline.

• How much time do you have before the potentially patentable ‘secret sauce’ is revealed?
• Do you have robust confidentiality agreements in place with all third parties who are privy to your commercially confidential information?
• How much information will be – or could be – revealed by e.g. conference disclosures, investor disclosures, articles, online repositories, regulatory agencies?

2. Consider strategies that can be adopted (at the USPTO, EPO and elsewhere) to give early insights into the likely scope of allowable patent claims.

• Consider prior art searches to assess the likely scope of allowable patent claims.
• Consider using Track One at the USPTO or PACE/acceleration for direct EPO/UK-IPO filings to get an indication of patentability before publication.

3. Conduct an in-house assessment of the ease of designing workarounds.

• How easy would it be to achieve the same result with a slightly different implementation? E.g. can different statistical analyses be used to achieve the same output? Could a different training method lead to a similar model?
• Consult scientists and/or engineers to get a picture of the risk of reverse-engineering.

4. Assess the value of the invention.

• Especially for startups, what are the key differences between you and your competitors – and can those differences be translated into patentable inventions?
• Is the computational invention going to be commercialised (e.g. as part of a diagnostic method), or does commercial value lie in the output of the computational invention (e.g. an AI-designed drug)?
• Is there a potential for licencing or cross-licencing with other companies?

How to get the best commercial value from a patent application

There are simple steps that companies can take, if they decide to file a patent application, to ensure that they get the best possible commercial value from the application. These steps will be highlighted in an upcoming article in this series.

Contact the Carpmaels team

If you have any questions about patenting inventions that cross the boundaries of biology and computer technology, please get in touch with our highly experienced Digital Life Sciences team.

Footnotes

[1] https://www.wipo.int/en/web/intangible-assets

[2] https://www.epo.org/en/news-events/press-centre/press-release/2023/945253#:~:text=Startups%20with%20patents%20and%20trade,new%20study%20finds%20%7C%20epo.org