Is Canada on the Brink of AI and Diagnostics Patent Rush?
Diagnostics and some types of artificial intelligence (AI) inventions (as well as other computer-implemented inventions) have often found themselves in the “grey area” of patentability in many jurisdictions, leading to uncertainty and an uneven playing field for diagnostics and AI companies. Here in Canada, the Canadian Intellectual Property Office’s (CIPO) approach to the question of patentable subject matter – what can be patented – has been much-criticized in recent years as departing from that which is required in law.
In upholding an appeal by Amazon.com in 2011, the Federal Court of Appeal ruled that CIPO’s attempts to use a “substance of the invention” test of its own creation was “incorrect in law.” Rather, the Court made clear that “determination of subject matter must be based on a purposive construction of the patent claims.” Nevertheless, CIPO in the intervening years has directed its patent examiners to use yet another, novel “problem-solution” approach to claim construction as a required step when considering the issue of patentable subject matter.
This year, the Federal Court of Canada gave hope to prospective patentees – particularly those in the life sciences and computer industry – that CIPO may finally adopt a consistent claim construction approach in accordance with established case law. On August 21, 2020, the Federal Court of Canada handed down its ruling1 in the appeal of Yves Choueifaty against the rejection of his application by the Commissioner of Patents. Choueifaty had argued that CIPO was wrong to use the problem-solution approach in construing patent claims. Ruling in Choueifaty’s favour, the Court found that the Commissioner of Patents had indeed “erred in determining the essential elements of [Choueifaty’s] claimed invention by using the problem-solution approach,” rather than the approach the Supreme Court has directed2 to be used, in particular, one that considers the inventor’s intention and the words of the claim.
CIPO’s problem-solution approach has created great difficulty for computer-based and medical diagnostic inventions. For example, in computer-related cases, patent examiners will often characterize the alleged “problem” in such a manner that any computer elements recited in a claim are non-essential and can be disregarded. Once the computer elements are disregarded, what remains is easily classified as an abstract idea or mere scheme and thus the entire invention is excluded from patentability. Similarly, in medical diagnostic cases, patent examiners often determine that the innovation is directed to the discovery of a correlation between a genetic alteration, or a level of an analyte, and a disease or medical prognosis and identify the correlation as the solution to the Examiner-identified problem. Once the elements of detecting the genetic alteration or measuring the analyte are disregarded (as these elements are not considered essential to solving the Examiner-identified problem), the claim is then alleged to be directed to a disembodied idea (i.e. a mental process or having no practical application) and, again, excluded from patentability.
In response to the Choueifaty decision, CIPO has now published a practice notice3 – a guidance document for patent examiners – to aid in correctly applying purposive construction. While Choueifaty did not specifically delve into what constitutes patentable subject matter, the practice notice does provide some guidance in this regard.
In particular, the practice notice outlines that CIPO examiners must now apply a new “actual invention” approach, which appears to comprise three steps. The first step of the new approach is to apply the purposive construction test, as set out by the Supreme Court4, to determine the essential and non-essential elements of a claim.
While the practice notice explicitly states that any claim construction that looks only at the so-called “substance of the invention” is incorrect, the second step of the new approach appears to be an attempt to distinguish between elements deemed essential for establishing the boundaries of a patent monopoly, and those involved in the patentable subject matter analysis. In so doing, the practice notice goes on to describe an “actual invention” step in which even those elements deemed essential using purposive construction “would not form part of the actual invention because the fact that it has no material effect on the working of the invention means it does not cooperate with other elements of the claimed invention” (emphasis added). Thus, to form part of the actual invention, an element “must cooperate with other elements of the claimed invention so as to become part of a combination of elements making up an actual invention.”
The final step of the new approach is to determine whether the actual invention “has physical existence” or “manifests a discernible physical effect or change”, and “relates to the manual or productive arts.” While this step could undoubtedly benefit from clarification, there seems to be two distinct ways of satisfying this final step of the test. The first is to demonstrate that physical elements (e.g. hardware) form part of the actual invention (e.g. a method that includes a step of interacting with a particular component of a computer in a specific way). The second is to demonstrate that the actual invention affects and/or is affected by a physical effect or change (e.g. a method that includes measuring a physical phenomenon and/or causes a physical phenomenon). For diagnostic inventions, the distinction between physical existence and physical effect is less clear. What is clear however is that elements relating to physically measuring a level of analyte meets the physicality requirement.
From an artificial intelligence perspective, application of AI to automated methods (e.g. expert systems) are likely to see the most benefit from the new approach. Such AI-implemented methods include processes like analyzing ATM security camera footage to determine the likelihood of criminal activity and diagnosing heart disease by analyzing echocardiogram data. From a diagnostics perspective, the new approach suggests that, in addition to data acquisition and correlation steps, measuring and detecting steps will also be considered when assessing patentable subject matter, which should make patenting diagnostic methods easier in Canada.
Taken at face value, the new guidance is a welcome development. For examiners, it provides a reasonably clear roadmap for assessing patentable subject matter. For applicants, the new guidance gives hope that AI and diagnostic inventions will be considered more holistically. While the new approach may fall short of affording patentees the certainty and level playing field that could have resulted from the Federal Court’s decision, there appears to be little doubt that the new approach will lead to increased patent allowances, and even to the granting of broader claims. This trend is likely to make Canada a more favourable jurisdiction for the patenting of AI-implemented methods and diagnostics in particular.
On December 2, 2020, Denis Keseris and Carmela De Luca presented a webinar on this topic. To view this webinar click here.
