AI in medicine – ways not to get a patent and what we can learn from them

AI in medicine – ways not to get a patent and what we can learn from them

Generally, advances in the field of medicine and healthcare are considered technical by the EPO. Yet merely specifying a medical field of use is not enough to convince the EPO that the invention serves a technical purpose and has the technical effect critical for a finding of the presence of a inventive step (or non-obviousness), as illustrated by the two decisions discussed here. In T 2677/16 finding drug targets was found not to be a technical purpose. In the more recent decision T 1910/20 building a therapy plan was held not to be a technical task and therefore a display to help better understand glucose measurement data was just a non-technical display of information; nor did it help that physiological data was displayed. I will discuss these cases and contrast them with Board of Appeal decisions finding related technology to be patentable. 

The first of the two decisions, T 2677/16, is rather succinct. The claimed invention related to (in simplified terms) a method for identifying a drug discovery target, including an ontology for storing and accessing genomics information in a database, querying a database to generate a profile including a collection of nodes representing genes and including concepts that are biologically related to a disease state, identifying a disease related pathway in the profile including actor genes or proteins and identifying the actor genes or proteins as drug discovery targets. A drug target is a molecule in the body, usually a protein or a gene, that is associated with a particular disease process, and could theoretically be targeted by a drug to treat the disease by interrupting the disease-related metabolic pathway. The rationale of the contested decision was that "the identified genes and proteins represent potential targets for a drug intervention that have the likelihood to produce a therapeutic effect" and hence, by virtue of this therapeutic effect, the identification of such genes or proteins represents a technical purpose (although the application was still found to lack inventive step for different reasons). The Board held that this rationale unduly broadens the concept of a technical purpose to encompass any scientific endeavour in medicine and stressed that a drug target is not a therapy: it has no therapeutic effect but is merely a promising direction for future research. The Board concluded that the discovery of a drug target may well turn out to be a valuable scientific discovery, but neither discoveries nor science have technical character per se. And for the avoidance of doubt, the Board made it clear that this should not be confused with the invention of a drug, which would indeed represent a technical purpose.  

Since the claimed invention modelled the disease-related metabolic pathways in the body for the purpose of identifying a drug target and since such modelling itself is an abstract intellectual activity, the Board held it cannot contribute to the technical character of the invention unless it serves a technical purpose. This was not the case here, since identifying drug targets (as opposed to identifying therapies) was not a technical purpose but scientific discovery. Since there was no technical effect beyond implementing the invention on a computer, the first instance decision that the claims lacked an inventive step was confirmed.  

As the board has stressed, the decision is specific to the discovery of drug targets and does not affect the patentability of therapies. Equally, genetics per se remains a technical field and nothing here suggests that previous decisions finding patentable math-based genetic research techniques don’t remain valid, as cited in the Guidelines for Examination. See also, for example, T 2050/07

The second decision, T 1910/20, dealt with an invention that clustered measured ambulatory glucose data into clusters and displayed the glucose data in a clustered way. The details of the display algorithm are not material to the first instance and appeal decisions. At first instance, the examining division found that the distinguishing features related to no more than a presentation of information that did not credibly assist a user in performing a technical task by means of a continued and guided human-machine interaction process (one of the requirements for a technical effect being acknowledged for presentation of information in the Board of Appeal case law). Specifically, the appellants argued that the technical effect of the distinguishing features was enhancing patterns in measured glucose data which may otherwise not be recognised or may be overlooked, namely making said patterns visible or recognisable. They argued that in the case in hand the presented information, in particular the cluster centres, contributed to the technical character of the invention because the presented clusters would help an experienced and skilled doctor to better understand the measurement data and thus assist them in building a more appropriate diabetes therapy plan. 

The Board was not convinced. According to the Board, building a therapy plan, which should not be confused with practising a method of therapy on the human body, is a purely intellectual exercise, not a technical task. For want of any technical task, the argument that the information presented in this case allegedly assists a user in performing a technical task was dismissed. The applicant further argued that there was a technical effect in that a patient was a real object and their physiological state in terms of patterns in the blood glucose data was a physical state. Again, the Board was not convinced. It is not enough that the input to a mathematical method represent physical parameters. In the present case, the applicant was unable to demonstrate any technical effect of the presentation of cluster centres.  

The decision in this case can be contrasted with, for example, the decision in T2681/16, where the technical character of a mathematical method of calculating a diabetic risk score from glucose measurements was not even questioned. How can these decisions be reconciled? In the present decision, while the cluster centres may help a skilled clinician to devise a treatment plan, helping with an intellectual, non-technical activity was not acknowledged as a technical effect. By contrast, in T2681/16, the technical character of a risk score was never at issue.. Arguably the difference lies in the immediacy to a technical use. A risk score is immediately actionable by a patient or physician if a high risk is indicated, whereas displaying data in a way that may be easier to analyse is several (intellectual) steps removed from the technical outcome of a treatment plan.  

What, if anything, can we learn from these two recent decisions? They have in common that mathematical methods that facilitate endeavours the EPO considers non-technical (making a treatment plan, identifying drug targets) were held not to have technical character, irrespective of the inputs to the method representing physical quantities. In common to both decisions, a technical use down the line was apparent in both cases: finding a drug acting on a discovered target or implementing a treatment plan. However, that was not enough to acknowledge a technical character of the method in question. It seems that what would be needed is a more direct link between the output of the mathematical method and a technical use or purpose than was present here.  

Follow Alexander Korenberg’s AI musings

AI musings

Cookies improve the way our website works. By using this website you are agreeing to our use of cookies. For more information see our cookie policy I accept