13 August 2017

Principles of Patent-Eligibility Continue to Trouble Applicants and Examiners in Australia

BorderlineThe law in Australia regarding patent-eligibility of computer-implemented inventions was supposedly ‘settled’ in May 2016, when the High Court rejected an application for special leave to appeal against a decision of the Full Bench of the Federal Court of Australia, thus leaving Research Affiliates LLC v Commissioner of Patents [2014] FCAFC 150 and Commissioner of Patents v RPL Central Pty Ltd [2015] FCAFC 177 as – for the moment, at least – the last word on the subject.  Yet, clearly, patent applicants and the Australian Patent Office continue to disagree on what, exactly, these decisions mean in particular cases.  In July 2017 alone, there were four decisions issued by the Patent Office resulting from hearings relating to objections against alleged computer-implemented inventions on grounds that each was not for a ‘manner of manufacture’, i.e. patent-eligible subject-matter under Australian law.

In Todd Martin [2017] APO 33, claims directed to a system and method for tracking usage of athletic equipment of an athlete using a GPS-enabled device and a computer server were refused.  In Rokt Pte Ltd [2017] APO 34 claims directed to a computer implemented method for linking a computer user to an advertising message by way of an intermediate engagement offer were refused, following the applicant’s efforts to amend in view of an earlier decision in Rokt Pte Ltd [2016] APO 66.  The claims at issue, and which were also refused, in Discovery Life Limited [2017] APO 36 were directed to a computer implemented method for of managing a life insurance policy with a related medical scheme.  Finally, in Bio-Rad Laboratories, Inc. [2017] APO 38, an examiner’s objections against claims directed to a method of establishing a statistically valid assay mean and assay range for a particular lot of a quality control material were upheld by the Hearing Officer, although Bio-Rad has been provided with an opportunity to amend its application to try to overcome the objections.

I am not particularly surprised by the outcomes in Todd Martin, Rokt, or Discovery Life.  This is not to say that they are necessarily correct – I have not spent the time to review the patent specifications in the detail that would be necessary to form a considered view on the matter.  However, on its face each of the three decisions is consistent with the principles set out in Research Affiliates and RPL Central, i.e. the claims appear to recite the use of generic computer technology as a practical implementation tool, and if this generic technology is disregarded, what is left is a scheme, abstract idea, or business process of the kind that has not generally been regarded as patent-eligible.  The Full Court in RPL Central was very clear about the treatment of such cases: ‘Where the claimed invention is to a computerised business method, the invention must lie in that computerisation. It is not a patentable invention simply to “put” a business method “into” a computer to implement the business method using the computer for its well- known and understood functions’ (at [96]).

Bio-Rad, however, is a different matter.  While the invention in this case involves the development and use of a new statistical analysis procedure, which is conveniently implemented by using a computer for its well-known and understood ability to store information, and to perform computations more rapidly and accurately than a human calculator, it is not apparent that disregarding the computer-implementation leaves only patent-ineligible subject matter.  On the contrary, regardless of how the invention is implemented – and even if it were carried out by a person using a pen and paper for calculation – it would still provide significant economic and technical advantages in terms of reduced cost and resource consumption/wastage in the production of quality control material for use in medical testing.

This is not to say that there are no issues with Bio-Rad’s claims – it is certainly arguable that they are overly broad, and encompass subject matter that does not provide the stated advantages of the invention.  However, these are not conventionally eligibility issues, and it is therefore not clear how or why patent-eligibility came to be the central question in the Bio-Rad case.  This suggests to me that the Patent Office is continuing to grapple with the principles of patent-eligibility set out in Research Affiliates, RPL Central, and D’Arcy v Myriad Genetics Inc [2015] HCA 35, and that applicants therefore still cannot expect predictable or consistent outcomes as examiners endeavour to apply these principles on a case-by-case basis.

The Bio-Rad Invention

The Bio-Rad invention relates to the field of medical testing.  In particular, it is concerned with the characterisation of test samples that are used to confirm the correct operation of medical testing instruments.  The machines in question are used for measuring such things as a person’s cholesterol levels or blood sugar, testing for the presence of drugs in their blood or urine, or measuring other aspects of their blood chemistry.  Obviously, it is quite important to confirm regularly that such instruments are functioning correctly, and producing accurate results – a process known as ‘qualification’.  One common way to qualify these machines is to use manufactured test samples which have known characteristics, to check that the results produced by the machines match the known characteristics.  Manufacturers of the test samples (also known as ‘quality control materials’) are required to provide, for each batch of samples, tables indicating a mean and a range of test result values within which a result from a test of a sample from the batch is expected, with a specified probability, to fall when the quality control material is tested using a testing instrument that is operating properly.

This creates a ‘chicken and egg’ problem of how to characterise the quality control materials in each batch.  Assuming that qualified test instruments are available, a difficulty still exists in that the testing process is destructive.  Therefore it is necessary to take just a sample of materials from each batch, measure the characteristics of the materials in the sample, and use the sample measurements to estimate the mean and variability of the relevant characteristics within the complete batch.  These estimates should be statistically valid, so that the resulting ranges can be relied upon with high probability when the quality control materials are used to qualify instruments used in testing real human samples.

To add to the complexities of this statistical problem, in addition to the statistical uncertainty arising from using sampling to estimate the batch characteristics, there are a number of other sources of uncertainty in the systems used to characterise the quality control materials:
  1. ‘within-instrument’ variability, whereby repeated tests on a single machine are expected to produce slightly different results;
  2. ‘between-instrument’ variability, whereby tests performed using different machines of the same type are expected to produce slightly different results; and
  3. ‘between-laboratory’ variability, whereby tests performed using the same machine, but in different laboratories, are expected to produce slightly different results.
Taking into account all of the sources of statistical variation, relatively large sample sizes are required in order to determine the ranges of expected test result values with a sufficiently high probability.  The invention is directed to reducing the required number of samples drawn from each batch, while retaining the statistical confidence and significance of the estimates of mean and range.

According to the specification, two realisations lay at the heart of the contribution made by the inventors leading to the development of the invention:
  1. a sample mean may be estimated using a relatively small number of samples, as compared with the number of samples required to estimate the sample variability with similar confidence; and
  2. while the mean test result expected from a particular quality control material varies between batches (resulting in the need to reassign means and ranges for each new batch), the variability of the test results tends to remain relatively stable between batches.
From these insights, a statistically-rigorous computational method was developed by the inventors that involves estimating the mean using samples drawn from the current batch of quality control materials, but which also stores historical results from earlier batches in a database, and uses these historical results in estimating the variability/range for the current batch.  As a result, the number of samples that must be drawn from each batch in order to determine statistically-valid mean and range values is significantly reduced.

Examiner’s Objection to Bio-Rad Claims

The essential steps in the invention as claimed are (I am paraphrasing somewhat here, except where text is placed in quotation marks):
  1. testing a number of samples from a batch of quality control material;
  2. computing the mean;
  3. retrieving historical test results from a database;
  4. computing ‘at least in part from the historical test results a variability estimate that is an estimate of the variability of test results obtained from tests performed on at least one prior lot of the quality control material’; and
  5. computing, ‘based at least in part on the mean and the variability estimate’ the range of qualification test result values expected, with a specified probability, for the batch of quality control material.
The examiner had rejected Bio-Rad’s claims, as not being for a patent-eligible ‘manner of manufacture’, stating that:

… the present alleged invention uses a standard computer system to carry out the abstract idea of testing various samples and determining the range result within which the result would fall with the target probability. Therefore, the substance of your application is just comparing the test results and determining the range within which the results of the test would fall. From reading your specification as a whole it is clear that this is merely implemented by using standard computer system i.e. the computer operates in the normal or expected way. Moreover, there is no part of the claimed method/system which results in improvement in what might broadly be called “computer technology. Merely implementing a non-patentable scheme into a computer does not make the application patentable. The computer is merely used for storing and comparing purposes.

The applicant requested a hearing in relation to the examiner’s objection. 

Bio-Rad Hearing Outcome

The hearing was held by written submissions, in which the applicant made the following arguments:
  1. the information which the invention extracts from the data – a mean and range – ‘is grounded in the context of a real-world need to be able to rely on medical tests’;
  2. the problem solved by the invention lies in reducing the amount of data required to be gathered, and thus the number of samples needing to be ‘characterised,’ to ensure the statistical validity of mean and range figures extracted from the data;
  3. that this is not a case, such as in recent court decisions relating to computer-implemented inventions, in which the ‘non-computer’ components of the claimed invention are inherently not patent-eligible;
  4. the effect of the invention may be observed in physical phenomena, namely the production of statistically valid means and ranges for quality control material with much reduced effort, cost, time, and resource consumption;
  5. the contribution of the claimed invention is thus technical in nature and solves a technical problem outside the computer implementation; and
  6. the invention does not lie in the generation, presentation, or arrangement of intellectual information.
The Hearing Officer appears to have essentially agreed with Bio-Rad on most of these points, yet still found that the claimed invention is not patent-eligible.  This may seem to be a harsh outcome, considering that the invention appears to have considerable technical merit and practical utility, and does not evidently fall within commonly-excluded subject matter, such as business methods, abstract ideas, or mere intellectual information.  Indeed, the claimed invention might be regarded as, in substance, a process or system in which the input is a batch of quality control material having unknown characteristics, and the output is the batch of quality control material (minus the tested samples) having known, statistically-validated, mean and range characteristics.

As discussed further below, however, the Hearing Officer appears to be concerned that while the claims may encompass subject matter having the qualities identified in the applicant’s submissions, they also encompass subject matter than does not have those qualities.  At [36] the decision cites Grant v Commissioner of Patents [2006] FCAFC 120, in which the court stated that: ‘It has long been accepted that “intellectual information”, a mathematical algorithm, mere working directions and a scheme without effect are not patentable. …. It is necessary that there be some “useful product”, some physical phenomenon or effect resulting from the working of a method for it to be properly the subject of letters patent.’  A claim that includes within its scope implementations that do not produce the requisite ‘effect’ or ‘useful product’ is therefore not patentable.

A New Hope?

However, the Hearing Officer did not completely close the door on the application, suggesting that there may be a problem with the current form of the claims, rather than with the underlying substance of the invention.  As such, he has indicated that the objection may yet be addressed by an appropriate amendment, and has allowed a period of two months for the applicant to file amendments and any supporting submissions.  A strong suggestion as to where the Hearing Officer considers the problem to lie may be found at paragraph [35]:

The applicant then goes on to assert that “reduced effort, cost, time and resource consumption” are “physical phenomena in which the effect ... may be observed” - once again employing phraseology from NRDC. And once again, I agree with this assertion, or at least its broad thrust. In any given circumstance one would expect these physical phenomena to manifest themselves in proportion to the extent to which one avails oneself of historical test results. However, the claim suffers from vagueness in this regard, allowing a variability estimate to be obtained from historical test results in full or in part. As a result, I am not satisfied that the described effect is necessarily produced by the claim. The claim, in its breadth, appears not to be limited to less measurements being made for determination of variability. Such an issue also appears to create a section 40 issue in that the claims do not appear fully supported by the disclosure of the specification. In this regard the specification only discusses determination of variability in totality from historical data. No measurements are made for the purpose of working out variability beyond those taken to calculate the mean.

In other words, it appears that the Hearing Officer considers the current claims to be overly broad, and that they do not sufficiently tie the method to the asserted technical benefits of reducing effort, cost, time, and resource consumption.

I can see the Hearing Officer’s point here, but at the same time I am inclined to think that he was closer to the mark in suggesting that this is an issue under section 40 of the Patents Act 1990, which relates to the form and content of a complete specification.  It is almost as if the Hearing Officer is suggesting that the substance of the invention, as discerned from reading the specification as a whole, is potentially patent-eligible, but not in the particular form in which it is claimed.  This is the opposite of findings in recent Australian court decisions, in which the form of the claims may give the appearance of patent-eligibility, but the substance of the invention derived from a reading of the specification as a whole lies within a field of ineligible subject matter.

I also wonder whether the problem here might be better regarded as one of utility.  Embodiments within the scope of the claims that do not achieve the promised benefits of the invention are not ‘useful’, as required by section 18(1)(c) of the Patents Act, and an objection might be raised to the entire claim on this basis.

Conclusion – Is Computer-Implementation the Issue Here?

Bio-Rad does not appear to be a ‘classic’ case of a patent-ineligible computer-implemented invention.  The field of the invention – production of quality control materials for use in qualifying medical test equipment – is not one that you would generally expect to fall foul of eligibility problems.  It appears that the invention is able to provide significant economic and technical benefits, at least by reducing resource consumption (waste) and cost of production of quality control materials.  If the only objective and result of the invention were to label batches of materials with mean and range values, without employing any new form of measurement process or technology, then that might be a different matter.  But here there is a new measurement process, which takes advantage of historical data to reduce the overall sample size required for estimation of the mean and range values.

I am not sure that it even matters whether or not the computations are computer-implemented.  While it is obviously advantageous to use a computer, in order to speed up the calculations, and minimise opportunities for error, this is not necessary to solve the problem addressed by the invention, i.e. reducing the amount of data required to be gathered, and thus the number of samples needing to be measured, in order to determine the mean and range values.  These benefits would still be achieved if the calculations were performed by hand with a pen and paper.

I therefore find it odd that the application has been treated as being for a computer-implemented invention at all.  There may be issues with the scope of the claims, i.e. they may be overly-broad so as to fall foul of section 40 of the Act, and/or to encompass embodiments that do not solve the identified problem so as to be useful as required by section 18(1)(c).  However, if these problems are addressed then I think that the claims should be patentable even if they do not recite the use of a computer to retrieve the historical data and perform the calculations.

The examiner and the Hearing Officer were both therefore correct to identify that the invention does not result in any improvement to computer technology, and that a ‘generic’ computer is used as a tool, for its known properties of improving the speed and accuracy of computations.  But that is because the invention does not reside in the use of the computer.  And when the computer-implementation is disregarded, then what is left is not a patent-ineligible scheme, business process, or abstract idea.  It is a specific method that solves a practical problem with tangible benefits in a particular field of technology and, as such, is patent-eligible.  Hopefully, then, the applicant will be able to propose amendments to the claims that will make this clear, resulting in the existing objections being withdrawn.


Post a Comment

Copyright © 2014
Creative Commons License
The Patentology Blog by Dr Mark A Summerfield is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Australia License.