Share this article:

Drafting Chemical and Life Sciences Applications Involving AI: Considerations For Success At The EPO

By Joanne Addison, Senior Associate and Kimberley Bayliss, Senior Associate

The use of artificial intelligence (AI) in chemistry and life sciences is rapidly growing. AI can be useful across a range of technical fields in the chemistry and life sciences fields, such as: identification of new compositions having desirable properties; structure determination; medical diagnostic tools; identification of synthesis routes; data analysis; and the prediction of properties of materials.

The EPO’s two ways of conferring technicality

When considering the patentability of AI inventions, we must consider the EPO’s approach for assessing AI based inventions, that is the EPO considers AI inventions to be deemed technical and therefore not excluded from patentability in two ways:

  • By virtue of being adapted to a specific technical implementation.
  • By adaption to a field of technology

At first sight, following this approach, it may seem that patenting of AI inventions in chemistry and life sciences fields should be straightforward by virtue of the inventions being adapted to a field of technology. However, inventive step of the invention as a whole will still need to be considered, long gone are the days when the mere application of AI methods in a standard matter was seen as inventive. Therefore, it may be difficult (or even unlikely to be possible) to obtain a European patent for inventions that use AI methods to improve or automate known processes, such as using AI techniques to analyse large data sets more quickly than can be done by previous processes.

Before drafting chemistry and life sciences patent applications which use AI to some extent, an important point to consider is the role AI plays in the invention, by considering which category of AI inventions discussed below best fits the invention. For example, is the AI aspect of the invention incidental to the invention or is the invention only possible because of the advent of AI; or does the invention actually represent a contribution to the field of AI itself rather than just to a specific field within chemistry and life sciences?

The AI inventions we come across generally fall into one of the categories discussed below. Dependent on where an invention sits on the spectrum shown in the figure below, different types of claims will be suitable to define the invention. This spectrum can be used for quick orientation and claim drafting with the EPO’s patentable subject matter criteria in mind.

The x-axis of this figure might be thought of as “degree of abstraction from the processing of the computer”, with runtime inventions lying toward the left and hardware-inspired inventions to the right.

Drafting AI patent applications Figure provided by Kimberley Bayliss

Applied-AI inventions

At the far left of the figure sit the Applied-AI inventions. These are where the invention lies (as the name would suggest) in the manner in which a known AI algorithm is used, and are considered technical (and so not excluded from patentability) by the EPO due to being “adapted to a field of technology”.

The blue box represents inventions that amount to “a better classifier”. For these inventions, machine learning may be used to improve or automate known processes. Classifying medical images as containing lesions based on a corpus of annotated training images would be an example of this kind of invention. Before proceeding with drafting such an invention, it should be considered whether the invention produces an unexpected technical effect: are there any other merits to the invention? Without an unexpected technical effect if may be difficult to obtain a European patent for these types of inventions due to lack of inventive step.

The green box represents inventions where AI is incidental to the invention and merely one way that the invention might be realised. For example, the invention might involve a step of predicting the selectivity of a catalyst for a particular reaction product, in which the predictions may be made using a machine learning model (but might also be performed using other methods). For inventions of this type, AI is an implementation detail, but not the main invention. Therefore, when drafting a patent application for this type of invention, AI embodiments would be unlikely to be useful in the independent claims, possibly even the dependent claims.

The yellow box represents inventions that, whilst not representing improvements to fundamental AI algorithms, are only possible because of the advent of AI. Real-time camera effects and interactive filters are an example of inventions of this type. In chemistry and life sciences fields, an example of such an invention is the application of AI in imaging diagnostics to identify nodules that indicate early lung cancer much earlier in terms of disease progression than can be done with existing methods. For these types of inventions, where the technical effect can’t reasonably be obtained without AI, the AI will likely feature in the independent claims of a patent application.

Core-AI

Moving further towards the right, the group of purple inventions represent Core-AI inventions. These represent contributions to the field of AI itself, for example: better models; improved pre-processing of data; improved methods of training. These inventions can generally be applied to a wide range of problems across a wide range of fields.

It may be initially assumed that inventions in chemistry and life sciences fields that use AI will usually fall into the category of “Applied-AI” and are unlikely to ever make a Core-AI invention. However, we would caution against this assumption, as many inventions we come across arise from real-world problems that inventors encounter when trying to apply known techniques to their particular data.

As an example, an inventor might say: “My dataset was too small, so I just reformatted the data to improve the learning of the model…”. If the same technique can be used on other small datasets, then this is a Core-AI invention.

Another practical example is advancements in federated learning which arise due to organisations needing to pool their datasets in order to train a model, without wanting to actually share the underlying data directly.

While inventions relating to Core-AI are generally more likely to encounter patentability issues at the EPO, in particular mathematical method objections, when these inventions are developed based on problems encountered in chemistry and life sciences fields it will be possible to clearly describe how the invention is adapted to a field of technology, by providing use-cases. These use-cases may be useful if it is necessary to limit the claims in Europe to ensure that they are considered technical (and so not excluded from patentability) by the EPO due to being “adapted to a field of technology”. Nevertheless, even in chemistry and life sciences fields, these inventions should be treated with care.

In the first instance, inventions relating to Core-AI may be defined usefully by independent claims directed to the mathematical method itself, irrespective of any field restrictions. This ensures that the application offers the applicant the widest number of options in prosecution, particularly if the application will also be prosecuted in other jurisdictions having different patent eligibility requirements to the EPO.

With the EPO in mind, however, the dependent claims should contain specific use-cases of different, and preferably graded, scopes. For example, in an analogous manner to providing examples of diseases to which a new drug may have a therapeutic effect in applications for which second medical use claims may be useful. Importantly a conversation should be had about the realistic scope that may be obtained for Core-AI inventions in Europe and the commercial usefulness of such claims.

With this in mind, use-cases might be chosen that detail how the Core-AI invention could be applied to the applicant’s most commercially important AI models and products. They should therefore be chosen in a strategic manner, as opposed to merely relying on the use-cases provided by the inventors themselves. Building up a claim set in this way provides the best opportunity to obtain granted European patents that are commercially relevant for Core-AI applications.

Hardware

Finally, to the far right, the red box represents hardware inventions. These include AI algorithms specifically adapted to particular hardware configurations. In this sense, the design of the algorithm is motivated by technical considerations of the internal functioning of the computer.

While inventions of this type may occur in the chemistry and life sciences fields, it may not be necessary to limit claims in Europe to use-cases. This is because the EPO considers this type of invention to be technical by virtue of being adapted to a specific technical implementation. As the hardware is an integral part of the invention, the independent claims are likely to make some reference to a hardware configuration and the manner in which the algorithm is adapted for that particular configuration, rather than directed to specific use-cases.

Sufficiency and Inventive Step Considerations

In chemistry and life sciences fields, providing experimental data to ensure that patent applications meet the requirements of sufficiency at the EPO and demonstrate an inventive step has always been at the forefront of our minds. However, it should be noted that for inventions in chemistry and life sciences fields in which AI is a key aspect (e.g., particularly inventions falling into the yellow and purple boxes), the provision of examples is also important in terms of sufficiency and inventive step of the AI aspects.

Researchers in the field of AI understand that the design of a training data set can be critical to the success of an algorithm, as well as the possible effects of model assumptions and design. Therefore, if AI embodiments are to be included in the claims, the patent application should provide information in relation to the training data set (e.g., size, how outliers are handled, selection), the model used to derive the AI (including the type of model, e.g., neural network, genetic algorithm, a decision tree, etc. and how the model is structured), as well as any assumptions made by the model.

Contact Jo Addison and Kimberly Bayliss at Haseltine Lake Kempner if you have any questions or would like to discuss these issues further.

This is for general information only and does not constitute legal advice. Should you require advice on this or any other topic then please contact hlk@hlk-ip.com or your usual Haseltine Lake Kempner advisor.

HLK bubble graphic HLK bubble graphic

Stay connected with HLK

Keep up-to-date with the latest IP insights and updates as well as upcoming webinars and seminars via HLK’s social media.

LinkedIn

Twitter