Machine learning in healthcare is an extremely hot topic, with the ability to accelerate innovation. A common problem for AI/ML approaches is the quality of data. Many scientific papers in this field contain methodological errors, lack detail and use low quality data; commentators have suggested many are destined for the “digital scrap heap”? Arguably, these problems arise when studies are published prematurely before all details are worked out.
Since the publication of an invention represents a hard deadline by which a patent application needs to be filed, premature publication means premature patent filings, if patent protection is sought. So, will we see a corresponding number of patent applications destined for the “digital scrap heap”? This could become a serious problem for applicants who will not achieve the protection their innovation warrants and the commercial context required, as well as for patent offices who may have to sort through this “digital scrap heap”.
While a race to the patent office and to publish may sometimes make sense to “get there first”, it goes without saying that innovators are, in fact, not well served if their inventions can only be protected by, at best, weak patents based on poor quality methods and results. There are three foreseeable outcomes:
-
No patent is granted.
-
A patent is granted, but it fails to protect the invention in a commercially relevant way.
-
A patent is granted which does have commercial relevance but is more vulnerable to challenges by third parties than would otherwise have been the case.
In many instances delaying the filing of a patent application could have translated to a better outcome. For example, the inventors could have added details to their methodology or confirmed their findings in higher quality datasets. The problem of premature filing is, of course, not restricted to the AI/ML field. For example, in life sciences, data generation can be slow and expensive, tempting applicants to file too soon. However, given the concern the scientific community has expressed with respect to poor-quality, premature publication in healthcare technology based on AI, it seems fair to be concerned that a similar trend will be seen in the filing of patent applications.
When do problems typically surface if a patent application is filed too soon? And how might those problems spell doom for the patent or patent application?
When do issues surface?
As highlighted above, issues can surface at both the pre-grant and post-grant stage in Europe.
Once an application is filed at the patent office, for example the European Patent Office (EPO), a search is conducted to identify the state of the art and to determine if the application meets the requirements for patentability. The next stage in prosecution is examination, an ex parte procedure in which the Examining Division will decide whether to grant a patent or not. If the patent has been filed too soon, this could lead to objections during examination, which may prevent the patent from being granted.
After grant, the validity of a European patent can be challenged by third parties during a nine-month opposition period. Oppositions are filed at the EPO, so the outcome of the proceedings will affect the status of the patent all over Europe. This means that despite a patent being granted, if it does not withstand attacks during this opposition period, it could ultimately be revoked or limited on a European-wide basis.
Following expiry of the opposition period, the validity of a European patent can still be challenged, but this must be done on a national basis. This is because after grant a European patent is validated in countries of interest to the applicant, resulting in a bundle of national patents registered at each of the national patent offices. At this point, third parties can bring invalidity proceedings before national courts or patent offices alleging the patent is invalid. These invalidity proceedings can rely on newly identified prior art and may include objections that the Examining Division did not raise. Also, a defendant in infringement proceedings can counterclaim that the patent is invalid. A future change to this system will be the introduction of the Unified Patent Court (UPC), which will open a central forum for litigation effective across all European states participating in the UPC. At the time of writing, the UPC is expected to begin operations sometime in the latter half of 2022, or possibly in early 2023. You can read more about the UPC here.
Applicants can also face a strategic problem where early studies have methodological limitations. A publication can be cited against a later filed patent application. This applies whether the publication is a scientific paper or an earlier published patent application. The publication of a study with methodological limitations may cause problems for higher quality patent applications filed later.
The above problems can arise if a patent application has been filed too soon. But why does filing a patent application prematurely impact its chance of going to grant or surviving challenge?
What issues surface?
Prematurely filed patent applications could turn out to be commercially irrelevant; perhaps with further repetition or interrogation, different conclusions would have been drawn from the early data on which the patent application is based.
Insufficiency and obviousness are two barriers to patentability that may arise when a patent application has been filed too soon.
Sufficiency of disclosure requires that a person skilled in the art is able to practice the invention using the information disclosed in the application and their common general knowledge. A detailed description of at least one way of carrying out the invention is required. If there are serious doubts, substantiated by verifiable facts that the invention cannot be put into practice across the entire scope of the claims, then the application may lack sufficiency of disclosure. A fundamental flaw of insufficiency cannot be rectified by later filed data. If a patent application has been filed hastily and later development shows the invention just doesn’t work, then the patent application may suffer from a fundamental flaw of insufficiency.
If an invention proves not to be reproducible (for example because too few data were available when the application was filed), this can also become relevant for inventive step. The EPO Guidelines for Examination (Part F, Chapter III.12) explain how the EPO decides if lack of reproducibility is a problem for sufficiency or for inventive step: it depends on whether the technical effect that has turned out not to be reproducible is expressed in the claims or not. If the technical effect is part of the claims then the problem is one of insufficiency; if the technical effect is not expressed in the claims then the problem is one of inventive step.
A key question for many innovators is whether data that wasn’t included in the patent application can later be used to support an argument for inventive step. Where possible, this allows early filing of a patent application while limitations to the inventive step position are addressed by data generated after filing. The answer to whether post-filing data can be used to support inventive step in Europe has often been connected to the concept of “plausibility”. In a seminal Board of Appeal decision Factor-9/John Hopkins T1329/04, it was held that post-published evidence could only be relied on to support inventive step if the effect relied on for inventive step was “at least plausible” at the filing date of the patent application. But EPO case law has not settled how high the plausibility threshold should be, or even whether plausibility should be considered at all. The EPO Enlarged Board of Appeal has therefore recently been asked to consider in what circumstances post-filed data can be used to support inventive step. You can read more about the referral here. The outcome of the referral could have wide-reaching implications for applicants, particularly in the fields of chemistry and life sciences where generating data can be expensive and slow and in fast-moving fields such as AI/ML where the pressure to publish early data may be difficult to resist.
How soon is too soon for filing with limited data? Whilst filing an application early may help to avoid anticipatory prior art, filing too early can also risk objections arising from a lack of good quality data.
Conclusion
Intellectual property can be crucial to commercial success. To build solid IP foundations for future success innovators need both technical expertise and trusted legal advisors. Your patent attorneys will be able to assist in optimising your filing strategy, so you know when the right time is to prepare and file a patent application.
If you have any concerns or queries about patent filing strategies, including those related to AI, ML or healthcare inventions, please contact Alexander Korenberg, Michael Newton, Beth Ormrod or your usual Kilburn & Strode advisor.