T 803/17 - Planning a surgery using a statistical shape model - Is providing a computer model a technical task?

Decision T 803/17 from Board 3.2.02 is remarkable, in my opinion, not so much for what it considered important for the assessment of inventive step, but more for what the decision did not consider important.

The invention in this case relates to a computer implemented method of planning at least a part of a surgical procedure to be carried out on a body part of a patient. The method provides a "statistical shape model" of a body part and then "instantiat[es] the statistical shape model of the body part using data derived from the patient's real body part". When the method instantiates the statistical shape model, it "adapts the part of the planned surgical procedure to reflect the anatomy of the patient's real body part to automatically plan the part of the planned surgical procedure."

Claim 1 reads as follows:

"1. A computer implemented method (110) of planning at least a part of a surgical procedure to be carried out on a body part of a patient, comprising:
providing (112) a statistical shape model of the body part; and
instantiating (116) the statistical shape model of the body part using data derived from the patient's real body part, characterised in that, the statistical shape model incorporates data representing at least a part of a planned surgical procedure to be carried out on a corresponding real body part of the patient and wherein instantiating the statistical shape model of the body part also adapts the part of the planned surgical procedure to reflect the anatomy of the patient's real body part to automatically plan (118) the part of the planned surgical procedure."

While the official catchwords of the decision deal with the concept of reformatio in pejus in the case of an inadmissible opposition (if you are interested in this, please read here), I find more remarkable the manner in which the Board dealt with the question of inventive step, namely, the manner in which they dealt with "model"-related features and the question of whether such computer models have technical character.

G 1/19 was pretty clear in that the simulation of a technical system is per se not technical. Only in exceptional situations (e.g., if the simulation serves to make an "indirect measurement", G 1/19, point 99; or when the simulation is adapted to the the physical hardware on which it is run, G 1/19, point 115) is the simulation considered a technical activity.

In the present case (T 803/17) we are looking at a computer-implemented method which uses statistical shape model and adapts it to the anatomy of a specific patient. To me, such a method has much in common with a simulation. Both is and remains on a data level. The results of both are not used to control a technical process, or to modify an object in the physical reality. What is produced is just data.

Therefore, I think it is justified to say that the technical character of the "statistical shape model" in T 803/17 is at least questionable.

Even when considering the purpose of the method mentioned in the preamble of claim 1 ("planning at least a part of a surgical procedure") the technical character of this purpose is not immediately apparent. It may be argued that planning a surgical procedure is first and foremost a mental act (Art. 52(2)(c) EPC).

The Board in T 803/17, however, did not address these questions at all. Without any commenting on the technical nature of "statistical shape models", or methods of planning a surgical procedure, they seem to have accepted that a (technical) problem is solved. They defined the problem to be solved as "to provide an alternative model to the model used in [the closest prior art]" (r. 8.4). 

One may wonder whether this can be taken as an indication that Technical Boards, after G 1/19, regard "statistical shape models", or computer-implemented models in general, to have technical character.

I think not. From the manner in which the Board in T 803/17 did not address the question, I suspect that the Board has simply overlooked the issue. (To their defence: the parties do also not seem to have argued in this direction.) In the absence of further post-G 1/19 decisions going in that direction I would assume that providing a computer model - unless there are exceptional circumstances (see above) - is a non-technical task.

In any event, and as always, it will be interesting to see whether other Boards follow, or in which direction the case law will evolve.

The full text of T 803/17 can be accessed here.