360 Knee Systems Pty Ltd

[2024] APO 39

IP AUSTRALIA

AUSTRALIAN PATENT OFFICE

360 Knee Systems Pty Ltd [2024] APO 39

Patent Application:                2022202026

Title:Graphical representation of dynamic knee score for a knee surgery

Patent Applicant:                   360 Knee Systems Pty Ltd

Delegate:  Dr N. R. Madsen – Deputy Commissioner of Patents

Decision Date:  11 September 2024

Hearing Date:  by written submissions filed 26 June 2024

Catchwords:  PATENTS – section 45 – examiner’s objection – modelling of possible outcomes for patients for knee surgery under varying surgical parameters – clarity – novelty – inventive step – manner of manufacture – examiner’s novelty objection is unsustainable - inventive step objection is sustainable – invention in substance appears to be directed to a mere scheme/algorithm lacking technical effect – opportunity to amend – time provided for further examination

Representation:  Patent attorneys for the applicant:  FB Rice Pty Ltd

IP AUSTRALIA

AUSTRALIAN PATENT OFFICE

Patent Application:                2022202026

Title:Graphical representation of dynamic knee score for a knee surgery

Patent Applicant:                   360 Knee Systems Pty Ltd

Date of Decision:                   11 September 2024

DECISION

The examiner’s objections that the claims lack an inventive step and are not for a manner of manufacture is sustainable.  I provide the applicant an opportunity to amend and provide six (6) months from the date of this decision to amend and gain acceptance. 

REASONS FOR DECISION

BACKGROUND

1. 360 Knee Systems Pty Ltd (“the applicant”) filed patent application 2022202026 on 23 March 2022 as a divisional application of application AU 2016411801.  This parent application is the national phase entry of application PCT/AU2016/050483 filed under the Patent Cooperation Treaty.  It has an earliest priority date of 14 June 2016.

2. A first examination report issued on 11 May 2023 including objections under the grounds of manner of manufacture, clarity, novelty, and inventive step.  The applicant filed a response with amendments and a second adverse report issued including these same grounds.  Following this report and a telephone discussion with the examiner, the applicant filed a request to be heard.  With their written submissions for the hearing of the matter the applicant also filed amendments addressing the clarity issue in the second report.    

3. The examination of the present application is governed by the Patents Act 1990 (“the Act”) as amended by the Intellectual Property Laws Amendment (Raising the Bar) Act 2012 (“the Raising the Bar Act”) as the application was filed after 15 April 2013.  Thus, I must accept the application if satisfied on the balance of probabilities that the application complies with the Act.  If I am not so satisfied, then I can refuse the application.  Furthermore, the final date for acceptance of the application was 11 May 2024, however paragraph 13.4(1)(g) of the Patent Regulations 1991 is available to extend the time for gaining acceptance to 3 months (or longer if appropriate under sub-regulation 13.4(3)) from the date of the present decision.

   SPECIFICATION

4. The specification[1] begins by discussing the condition of osteoarthritis noting that it is the degenerative loss of cartilage tissue in a joint and is the most common joint disease in Australia.  Note is made of the knee as being a common site for sufferers of the disease particularly in older people, and that a form of treatment for the condition is a total knee replacement with goals being pain relief and improved functional ability[2].  The specification then discusses measures for success as follows[3]: 

   The primary objective measure for success is survivor analysis with regards to revision rate, which sits at 6.5% over a 12 year window.  Interestingly, this figure is vulnerable to underestimation as the conventional tracking of a patient endpoint when they undergo revision surgery implies two other success conditions: either the patient dies before undergoing a revision surgery they may require in the future or a patient’s health deteriorates with age to the point where it is deemed safer not to operate even if a revision surgery is required.  Nevertheless, this statistic masks a greater problem: as many as 20% of patients report dissatisfaction with the pain relief and functional outcomes of their surgery after 1 year.  Due to the relative ease of data collection and hence wider adoption in joint registries of survivorship based data, as well as the relatively greater exposure of the practicing surgeon to a smaller number of highly dissatisfied patients affected by outcomes such as implant loosening than a larger number of less dissatisfied patients, there exists the potential for a bias in favour of mechanically ‘safer’ but not necessarily patient outcome optimal surgical decision making.

   [1] Specification at [0002]

   [2] Specification at [0003]

   [3] Specification at [0003]

5. The specification[4] then notes that in order to increase the success of patient outcomes, surgeons might make small changes to surgery parameters, but they rarely have the tools available to allow them to investigate how parameter changes might have a positive impact for a particular patient. 

   [4] Specification at [0004]

6. After some generic text, a set of consistory statements and a listing of the various figures, the specification turns to a detailed description of embodiments.  At this point it is noted that the disclosed invention provides a system that generates a graphical display for a surgeon to show the surgeon in which direction the surgery parameters might be changed to improve patient outcome[5].  Involved in the invention is the use of a computer tomography (CT) scan of a patent’s knee and machine learning that links mechanically simulated priorities of historical patients with reported outcomes.  This data assists in identifying predicted outcomes for the current patient. 

   [5] Specification at [0031]

7. After canvassing some rather generic computer technology for implementing the described embodiments, Fig. 2 is discussed.  This figure is a good general depiction of the invention presented in the independent claims and is straight forward to understand.   

8. As a first step, a CT scan of a patient’s knee is received and a model of the knee joint in 3D is built.  A surgeon then provides input being information about a knee implant that may involve an identifier.  The specification[6] notes that a database of knee implants may be accessed to retrieve the geometries and recommended surgery parameters of the chosen knee implant.  It is noted that most surgery parameters are defined in the database but the particular surgery parameters of rotation and slope of the tibial component of a knee implant are particularly relevant.  These parameters have a relationship with reported patient outcomes as the output of machine learning models, and this relationship is represented by the retrieval of multiple machine learning model parameters in the figure. 

   [6] Specification at [0033]

9. Fig. 2 then enters into a looping procedure whereby a computer virtually performs a surgery.  The specification notes regarding the looping procedure that involves repeated action for values of tibial rotation and slope[7]: 

   …processor 102 changes the 3D shape of the bones by introducing cut surfaces and adding the shape of the selected implants according to the surgery parameters from the database. Processor 102 can derive a kinematic model, which is a simplified representation that disregards the 3D details that are not needed when considering the movement of the knee joint. In other words, processor 102 configures a post-operative kinematic model by virtually performing the operation and simplifying the 3D model to the kinematic model. The kinematic model comprises joints, bearing/contact surfaces, tension elements, members etc. Processor 102 can then use the kinematic model to perform a simulation of the post-operative knee joint as described in more detail below. In particular, processor 102 may perform a movement of the knee joint, that is processor 102 iteratively changes the angle between tibia and femur and calculates for each angle the multiple kinematic parameters. Processor 102 may also aggregate parameters from the multiple angles into one parameter, such as maximum or average.

   [7] Specification at [0034]

10. In relation to Fig. 2 the specification[8] then discusses that the kinematic simulations are used to estimate current patient outcomes by applying machine leaning model parameters to the simulated parameters, calculating one estimated or predicted outcome for each combination of rotation and slope parameters.  As discussed in the specification:

    Processor 102 then estimates 212 a current patient outcome by applying the multiple machine learning model parameters to the multiple simulated kinematic parameters of the current patient. For example, processor 102 selects the parameters as identified by the feature selection step of the machine learning method. Processor 102 may then calculate the weighted sum of those selected parameters in the example of a linear model. This way, processor 102 calculates multiple estimated or predicted patient outcomes. That is, processor 102 calculates one estimated or predicted patient outcome for each combination of rotation and slope parameters.

    [8] Specification at [0043]

11. With these outcomes calculated, a plot can be made to graphically depict the relationship between rotation and slope of the tibial components of the knee implant and outcomes.  Fig. 4a and Fig. 4b show example shaded surfaces[9], or in other words 2D data plots, with optimised patient outcome being in white colour.   In Fig. 4a the surgeon can see that the slope (Tibial Component IE) has little effect on outcome while rotation should be left at a minimum while Fig. 4b seeks maximum slope and minimum rotation to achieve optimum patient outcome. 

    [9] Specification at [0044]

12. Amongst the description above as laid out in the specification is discussion of the fact that after a knee surgery is performed, a patient can fill in an electronic questionnaire to assist in understanding patient outcomes.  It is noted that[10]:

    The questions relate to the patient outcome of the operation, such as pain levels, flexibility

    and other objective or subjective measures as described in more detail below.  As the patient answers the questions, processor 102 stores the answers to create a reported patient outcome stored on data memory 106.  In one example, processor 102 calculates a single score, that is, a value that represents the patient outcome.  For example, processor 102 calculates a single pain score where a lower pain score is preferable.  As a result, processor 102 can store the reported patient outcome, such as the pain score, associated with the determined kinematic parameters.  This data can serve as a sample for a supervised learning algorithm where the kinematic parameters are the inputs and the reported patient outcome is the output variable.

    [10] Specification at [0036]

13. Thus, it is clear that this questionnaire can serve as training data, to assist in the generation of machine learning models that link parameters of surgery with some objective or subjective measure of patient outcome.  I will touch upon the nature of the described machine learning a little later however for present purposes I consider it important to note that key to understanding the nature of the invention is obtaining an understanding of the meaning of the term “patient outcome”.  I note that patient outcome may refer to measured patient outcomes that are obtained via a questionnaire and used to train a machine learning model, or estimated or predicted patient outcomes that are generated as a result of simulated surgery and plotted on the 2D shaded graphs shown above.

    Patient Outcome

14. There are a number of locations in the specification that refer to the idea of a patient outcome.  I will refer to some significant locations in order to seek to understand the intended meaning of the term as used in the specification.  For example, the questionnaire is discussed as follows[11]:

    The historical patient records may further comprise historical demographic            parameters and patient questionnaire data capture parameters, the multiple machine learning model parameters may be indicative of a relationship between (i) the historical demographic
    parameters and patient questionnaire data capture parameters and (ii) the reported patient outcome, and estimating the current patient outcome may comprise applying the multiple
    machine learning model parameters to a current patient’s demographic and patient questionnaire data capture parameters.

    Estimating the current patient outcome may be based on kinematic expert knowledge to either modify or reweight penalty factors from the kinematic simulation or is based on new penalty factors from the kinematic simulation.

    [11] Specification at [0020] to [0021]

15. Nothing in this text identifies the particular data gathered regarding patient outcome.  It simply suggests that survey data will be used to generate relationships between parameters and outcome and that expert knowledge may be applied to optimise the data.

16. Paragraph [0056] provides some detail regarding the questionnaire and the nature of patent outcome data as follows:

    In another example, the historical patient records further comprise historical demographic and patient questionnaire data capture parameters. In this example, the machine learning model parameters are indicative of a relationship between the historical demographic and patient questionnaire data capture parameters and the reported patient outcome. Estimating the current patient outcome then comprises applying the multiple machine learning model parameters to a current patient’s demographic and patient questionnaire data capture parameters. The demographic and patient questionnaire data capture parameters may comprise age, gender, occupation, current knee pain state, current knee related activity impairment state, other musculoskeletal impairment and pain and subjectively measured anxiety and depression scores.

17. Clearly there are elements of objective and subjective measurements present in the understanding of patient outcome.  In other words, some patient outcomes appear to potentially relate to physical states of objects and others are purely subjective indications.   Greater detail regarding patient outcomes is provided a little later in the specification under the heading Reported Patient Outcomes[12].

    [12] Specification at [0067]

18. Extensive discussion in the specification focuses on machine learning approaches to identifying the relationship between kinematic parameters such as tibial rotation and slope and reported patient outcomes.  Initial discussion[13] mentions a single pain score or standardised Patient Reported Outcomes Measures (PROMS).  The specification discusses that in some examples[14] patient input data is generated by a patient sensor and uploaded to a computer system.  For example, the patient may wear a step counter and step counter data could be used in a manner similarly to questionnaire data. 

    [13] Specification at [0068]

    [14] Specification at [0074]

19. Fig. 10 is a presented as a statistical model that ultimately arrives at a point of analysis representing the satisfaction of the patient with a particular surgery. 

20. The specification discusses[15] that:

    Each node is associated with a probability function that takes, as input, a particular set of values for the node's parent variables, and gives (as output) the probability (or probability distribution, if applicable) of the variable represented by the node. For example, if m parent nodes represent m Boolean variables then the probability function could be represented by a table of 2m entries, one entry for each of the 2m possible combinations of its parents being true or false. For example, there may be multiple age related nodes where each node is indicative of whether or not the patient age is within a predefined age bracket. Such as a true or false value for the statement “age is below 40”. Other representations could include multiple sub-tables for groups of parent nodes with dependence upon each other. The edge weights may be considered machine learning model parameters.

    In other examples, statistical model 1000 is an undirected, and possibly cyclic, graph;

    such as a Markov network.

    The satisfaction node 1003 may also be a Boolean node representing whether or not the patient is satisfied with the operation. This way, the probabilities given the actual patient input data as described above can be propagated through the statistical model 1000 to calculate a final probability for the patient being satisfied, that is, a probability for a value of ‘1’ or ‘True’ at the final satisfaction node 1003. This probability can then serve as the predicted satisfaction value. An example of this is that a BMI over 40 and age under 55 years would return a predicted satisfaction value of ‘64%’, indicating that there is a 64% chance of the patient being satisfied after the operation, and that the relatively young age and high BMI has negatively impacted her chance for a successful outcome.

    [15] Specification at [0079] to [0081]

21. After discussing further details with respect to data modelling and the relationship between operative parameters and outcomes the specification continues by discussing that the methods described may be used for operating a healthcare system.  Paragraphs [0112] and [0113] note that:

    In particular, processor 102 may perform the method 200 in Fig. 2 to determine a predicted satisfaction value for each of multiple patients enrolled in the healthcare system. Processor 102 may then determine a patient care item for each of the multiple patients by maximising utility of healthcare spent in the healthcare system. For example, processor 102 may determine whether the cost for additional physiotherapy treatment outweighs the expected gain in predicted satisfaction value.

    By combining the multiple patients at the same time, processor 102 can minimise the global overall cost with a collective satisfaction or patient outcome target and perform a per patient cost minimisation with a per patient satisfaction or patient outcome target. Processor 102 may also perform fixed global cost allocation with a satisfaction of patient outcome target maximisation.

22. The specification adds[16] that a range of scoring metrics may be used in order to strike a balance between incorporating objective, directly measurable data such as range of motion (ROM) measurements and subjective, questionnaire-based data.  Discussed are examples of patient focused scores such as the Knee Osteoarthritis & Injury Outcome Score (KOOS) and the Western Ontario and McMaster Osteoarthritis index (WOMAC).  It is clear however, that subjective self-assessment is contemplated by the specification as a measure of patient outcome.  For example, the specification discusses[17] that:

    Subjective self reported measures of activity and mobility level may vary greatly from objective measures in non predictable ways, however, with sub population trends and variable subject level bias both skewing results. End stage knee osteoarthritis patients may have reduced steps/day counts over healthy comparable age subjects, dropping from about 8800 steps at peak to 6600. These figures are variable within and between patient population groups, however, with deliminations (sic) such as public vs. privately treated patients, age and gender all creating enormous variance. Seemingly at odds with this variation is the observation that only 3 days of active measurement are required to elicit a patients activity level profile when assessing step count, which would seem to dispute the idea that patients will change behaviour on weekends vs weekdays and other distinguishing factors.

    [16] Specification at [116]

    [17] Specification at [0153]

    Technology described in the specification

1. At present the claimed invention does not focus upon any particular details of technology involved in the generation of CT scans of a patient’s knee, the inputting and outputting of data, the operation of a machine learning model, or the generation of post-operative kinematic models.  There is detail in the specification regarding computer architecture employed to implement the invention, however this architecture appears to be a description of standard computer technology[18].  There is also detail provided of the kinematic simulation system that can be employed to implement the present invention which focusses on particular modelling approaches using X-ray images[19].  I will not discuss any of these elements further at this stage and will address this technical detail as necessary throughout the decision. 

   [18] Specification at [0060] to [0066]

   [19] Specification at [0181] to [0220]

   Claimed Invention

2. In response to the examiner’s first report, and also with their written submissions for the hearing the applicant filed statements of proposed amendments.  The examiner took no issue with the first statement of proposed amendments in terms of allowability.  Looking at both requests, I similarly see no issue of allowability of amendments in accordance with section 102.  Thus, the claims before me for consideration are those as proposed on 26 June 2024.  Independent claim 1 is presented below with an amended feature (additions in underline and deletions in strike-through) addressing the examiner’s clarity objection from his second examination report.  There is a further independent claim (claim 18) reflecting the same subject matter. 

   1.A method for assisting a surgeon with a knee surgery, the method comprising:

   receiving computer tomography data of a knee of a patient;
        building a 3D model of the knee of the patient using the computer tomography data;
        receiving user input from the surgeon, the user input comprising an identifier of a knee implant, the knee implant comprising a tibial component;
        retrieving multiple machine learning model parameters indicative of machine learning performed on historical patient records, the historical patient records comprising multiple historical kinematic parameters of each of multiple historical patients as inputs and a reported patient outcome for each historical patient as output, the machine learning model parameters being indicative of a relationship between the multiple historical kinematic parameters and the reported patient outcome;
        for each of multiple values of rotation of the tibial component and slope of the tibial component:
                    configuring a post-operative kinematic model of the knee of the patient using the 3D model, the user input and the value of the rotation and the value of the slope, wherein configuring the post-operative kinematic model comprises:
                    virtually performing a surgery on the 3D model by introducing cut surfaces to change a shape of bones in the 3D model, adding a shape of the knee implant, and reducing one or more 3D details of the 3D model to simplifying the 3D model to the postoperative kinematic model;
                    performing a kinematic simulation on the post-operative kinematic model to determine multiple simulated kinematic parameters; and
                    estimating a current patient outcome by applying the multiple machine learning model parameters to the determined multiple simulated kinematic parameters; and
        generating a shaded surface on portions of a user interface, the portions spanning the multiple values of rotation of the tibial component and slope of the tibial component on the user interface to graphically represent the estimated current patient outcome for each of the multiple values of rotation of the tibial component and the slope of the tibial component.

   CLAIM CONSTRUCTION

3. While the rules of construction for an Australian patent specification are well summarized in Decor Corp v Dart Industries[20], the correct application of these rules to the construction of claims was discussed by Bennett J in H Lundbeck A/S v Alphapharm Pty Ltd[21]:

   "the words in a claim should be read through the eyes of the skilled addressee in the context in which they appear ... while the claims define the monopoly claimed in the words of the patentee's choosing, the specification should be read as a whole ... it is not permissible to read into a claim an additional integer or limitation to vary or qualify the claim by reference to the body of the specification ... terms in the claim which are unclear may be defined or clarified by reference to the body of the specification."

   [20] [1988] FCA 399; 13 IPR 385

   [21] [2009] FCAFC 70; 81 IPR 228 at [118] – [120]

4. The examiner’s objection to independent claims 1 and 18 prior to the above amendment was as follows:

   “It is unclear from the claim language what is intended by the term ‘simplifying’.  Exactly what the simplifying process might entail in this context is not clear, and therefore the reader is left in doubt as to the precise scope of the monopoly being claimed.”

5. The amendment above now includes the feature that simplifying is done by reducing one or more 3D details of the 3D model.  In the context of the examiner’s observation, there may still be a question as to how this reduction in 3D details moves the 3D model “to the post operative kinematic model”.  On plain reading it would seem to me that the post operative kinematic model is simply one that is by definition, lower in detail than the 3D model and somehow derived therefrom.  Turning to the specification this understanding is reinforced by paragraph [0034] which notes:

   That is, processor 102 changes the 3D shape of the bones by introducing cut surfaces and adding the shape of the selected implants according to the surgery parameters from the database. Processor 102 can derive a kinematic model, which is a simplified representation that disregards the 3D details that are not needed when considering the movement of the knee joint. In other words, processor 102 configures a post-operative kinematic model by virtually performing the operation and simplifying the 3D model to the kinematic model.

6. I am satisfied that the relevant feature as proposed to be amended is clear.  While the nature of the simplification in reduction of 3D details in not outlined in the claim, it is clear that there is a reduction in details of the 3D model to move that data towards a different representation being a postoperative kinematic model.  I do not see reason why a skilled addressee would be uncertain of the bounds of this feature. 

7. Moving to the whole of the claim I do not see any particular difficulty in understanding the nature and operation of the invention of the independent claims.  The claimed invention is well reflected by Fig. 2 discussed earlier.  I note that a key term warranting close scrutiny is the idea of a patient outcome, be it a reported patient outcome or an estimated current patient outcome.  I have already focussed upon this element when considering the specification.  Plainly construed, the term “patient outcome” is broad, and would appear to include within its scope outcomes that are reflective of objective measurements of physical variables related to a post-operative knee joint, and subjective variables that may, for example, be elucidated from a patient in a post operative survey.  This is entirely consistent with the discussion of the invention provided in the description. 

8. In their submissions for the hearing the applicant noted that:

   “In light of the current specification, “patient outcome” (either historical patient outcome or current patient outcome) is a value that represents the ‘outcome’ of the knee surgery. For example, as discussed in paragraph [0026], the patient outcome may be a single pain score where a lower pain score is preferable. In some examples, the reported patient outcome may be based on Patient Reported Outcome Measures (PROMS). In one example, the patient outcome may be a desired patient outcome. So instead of simply trying to achieve the highest (or lowest) outcome value, the surgeon needs to find the surgical parameters that achieve the desired outcome.”

9. I do not see this submission as contradicting my construction above in any way.  The term “outcome” is identified in the broad sense with some specific examples discussed.  The patient outcome is merely some value representing the outcome of a knee surgery.  It could merely be a binary representation of subjective patient satisfaction.

10. I will first turn to the examiner’s objections to the claimed invention under s18(1)(a) of the Act.

    MANNER OF MANUFACTURE

11. I refer to summaries of principles present in recent Patent Office decisions regarding patentability computer implemented inventions[22].  Suffice to say, one must consider the substance of the invention[23]:

    [22] Mastercard International Incorporated [2023] APO 42, Block, Inc [2023] APO 34

    [23] D’Arcy v Myriad Genetic Inc. [2015] HCA 35 at [144]

    “Whatever words have been used, the matter must be looked at as one of substance and effect must be given to the true nature of the claim.”

    I also note that relevant principles to apply are well articulated in Rokt 1[24].   

    [24] Rokt Pte Ltd v Commissioner of Patents [2018] FCA 1988 (“Rokt 1”) at [189]

    “17.1 The Court must decide, as matter of substance not form, whether the claimed invention is the proper subject-matter for a patent: RPL Central at [99]; Research Affiliates at [106], [117].

    17.2 This requires consideration of both the claims of the Application and the invention described in the body of the specification: RPL Central at [114].

    17.3 The assessment is not done mechanically. There are no precise guidelines or mathematical formula. It is ‘a question of understanding what has been the work of, the output of, and the result of, human ingenuity’and then applying the developed principles: Research Affiliates at [116]. See further RPL Central at [112]:

    Recognising that the claims are to a method and system comprising a combination of integers, it is necessary to understand where the inventiveness or ingenuity is said to lie ...

    17.4 One well-settled principle is that a distinction exists between a technological innovation and a business innovation. A technological innovation is patentable. A business innovation is not: Research Affiliates at [94]; RPL Central at [100]. Consequently, a business method or scheme is not, per se, a proper subject for letters patent: RPL Central at [96]. Nor are abstract ideas, mere intellectual information or mere directions for use patentable: Research Affiliates at [101]; RPL Central at [100].

    17.5 A computerised business method or scheme can, in some cases, be patentable. However, ‘[w]here the claimed invention is to a computerised business method, the invention must lie in that computerisation’: RPL Central at [96] (emphasis added). This requires ‘some ingenuity in the way in which the computer is used’: RPL Central at [104]. It is not a patentable invention ‘to simply “put” a business method “into” a computer to implement the business method using the computer for its well-known and understood functions’: RPL Central at [96]. In other words, if the ingenuity lies in the business method or scheme alone, the invention will not be patentable despite the computer-implementation.

    17.6 Thus, a claimed invention must be examined to ascertain whether it is in substance a scheme or plan, or whether it can broadly be described as an improvement in computer technology: RPL Central at [96]. Contrary to [the applicant’s submissions at [49]], this is a binary distinction: the invention is either an unpatentable scheme or plan, or it is a patentable improvement in computer technology. In conducting the analysis, it is useful to:

    17.6.1 ascertain whether the contribution to the claimed invention is technical in nature: RPL Central at [99], Research Affiliates at [114];

    17.6.2 consider whether the invention solves a ‘technical’ problem within the computer or outside the computer: RPL Central at [99], Research Affiliates at [103];

    17.6.3 consider whether the invention results in an improvement in the functioning of the computer, irrespective of the data being processed: RPL Central at [99], Research Affiliates at [118];

    17.6.4 consider whether the invention requires merely ‘generic computer implementation’, as distinct from steps which are ‘foreign’ to the normal use of computers: RPL Central at [99], [102]; Research Affiliates at [101]; and

    17.6.5 consider whether the computer is merely the intermediary, configured to carry out the method using program code for performing the method, but adding nothing to the substance of the idea: RPL Central at [99].”

12. It is also useful to turn to patentability principles regarding mathematical algorithms, and methods of testing, observation and measurement.  There is little court guidance in this specific space however I note that, while a mathematical algorithm per se may not be a manner of manufacture, the presence of such an algorithm within the steps in an otherwise patentable method does not exclude a claim from patentability.  For example, in the matter International Business Machines Corporation v Commissioner of Patents[25] the Court, in considering a method for producing an improved (that is, smoother) visual representation of a curve, found that:

    “In the present case, it seems to me that the use of the algorithm is not different conceptually from the use of the compounds involved in National Research and Development Council. Just as those compounds were previously known, so here, it is not suggested there is anything new about the mathematics of the invention. What is new is the application of the selected mathematical methods to computers, and in particular, to the production of the desired curve by computer. This is said to involve steps which are foreign to the normal use of computers ....”

    [25] (1991) FCA 625 at paragraph 16

13. Consistent with the Full Court principles articulated above it appears reasonable to conclude that the distinction to be drawn is between a claim to an algorithm (or scientific principle or natural phenomenon) implemented on a computer in the abstract sense, and the application of the formula to a process giving a result that is technical in nature in that it produces, as posited in National Research Development Corporation v Commissioner of Patents [26], “some advantage which is material, in the sense that the process belongs to a useful art as distinct from a fine art”.  Such ideas would appear to apply similarly to methods of testing, observation and measurement to the extent that what is claimed must be more than a mere scheme or algorithm.  Where such a method addresses some technical limitation or problem such as the improvement to a measurement of a physical characteristic, patentability will generally be found.  Such ideas are reflected in the Patent Examiner’s Manual of Practice and Procedure[27]. 

    [26] (1959) 102 CLR 252

    [27] >

    The claimed invention clearly relates to a method whereby data is gathered and used to simulate a potential surgery such that parameters of a surgery can be mapped against potential patient outcomes.  The invention clearly involves a mathematic/algorithmic exercise that fundamentally simulates and predicts the outcome of a surgery.  It is a method that is akin to testing, observation and measurement in that it takes in, and manipulates data in order to generate a useful final output/data representation.  Fundamentally, to find patentability for the present invention, it appears appropriate to seek to determine whether the claimed invention as a matter of substance is technical in nature. 

    Examiner’s Objection

14. I summarise the examiner’s key points supporting his objection as to the patentability of the claimed invention as follows:

    ·     “[T]he claimed invention simply results in the generation of the shaded surface. The invention does not dictate performance of the surgery within the ‘safe zone’ as the Applicant submits. Rather, the invention as claimed fails to define any action which results from the provision of this information. Accordingly, the invention simply results in the production of intellectual information, with no relevant physical effect.”

    ·     “[T]he claimed solution merely represents the computer-implementation of an abstract idea wherein the computer is merely exploited for its well-known and well-understood capabilities.”

    ·     “As was the case in Research Affiliates, the ingenuity of the invention lies in the abstract idea, rather than the computer implementation… no part of the claimed invention relates to an improvement in ‘computer technology’. Therefore the invention does not provide a technical contribution to the art.”

    ·     “The claimed use of the computer to create an in-silico model simply represents the exploitation of the conventional data processing capabilities of the computer, rather than a technical advance in the use of the computer.”

    ·     “Although it is accepted that the result of the claimed method, the shaded surface, may be ‘useful’ in the broadest sense of the phrase, this is insufficient to found a conclusion that the computer is not operating as an intermediary. The substance of the invention lies in the content of the data, and the particular intellectual concepts implemented to provide the shaded surface, rather than any specific effect generated by the computer. That is, the computer is in essence operating as an intermediary to perform the requisite data processing in the process of generating the shaded surface.”

15. Ultimately the examiner concluded that the substance of the claimed method is a mere scheme for simulating a patient’s knee surgery and therefore not a manner of manufacture. 

    Applicant’s Submissions

16. The applicant’s submissions usefully dive deeply into considering the various factors that the Federal Court has made clear are applicable to considering patentability of computer implemented inventions in Australia. These submissions align closely with the summary of the factors I identify above at paragraph [33]. I will consider them in detail. As a preamble to my consideration, I note that the applicant points out in their submissions the problem that is addressed is consistent with my understanding of the invention. They argue that the invention works by performing the following steps for multiple values of rotation and slope of a tibial component wherein an estimated patient outcome for values is graphically represented on a shaded surface:

    ·“building a 3D model of the patient's knee to represent the computer tomography data;

    ·virtually performing the surgery by introducing cut surfaces to change a shape of bones in the 3D model;

    ·simplifying the 3D model to the post-operative kinematic model;

    ·performing a kinematic simulation based on the post-operative kinematic model to determine multiple simulated kinematic parameters; and

    ·estimating a current patient outcome by applying the multiple machine learning model parameters to the determined multiple simulated kinematic parameters.”

17. I agree with this, and I also agree that there is prima facie ingenuity in providing an improved representation of patient outcomes using these steps.  However, it remains for me to consider the nature of this ingenuity and the nature of the claimed invention as a matter of substance.  On this the applicant suggests that “the substance of the invention defined by the claims is a method to determine optimal surgical parameters to improve outcome of a knee surgery”. 

    Consideration

18. The applicant firstly takes a general approach to their assessment of the substance of the invention fundamentally arguing that:

    ·“The contribution is optimal and accurate surgical parameters as a result of incorporating a 3D model of the knee of the patient and virtually performing the knee surgery on the 3D model.. [and] does not lie simply in it being a computer program [or] a scheme.”

    ·“The claimed invention provides an improvement to surgical technology by increasing the accuracy and efficiency of knee surgery, enabling a reduction in post-operative knee pain, and generally providing diagnostic functionality previously unavailable… limitation of the 3D model, as well as the post-operative kinematic model, means that the claim relates to mechanical properties of a knee joint and not some abstract mental process.” 

    ·“The claims are directed towards using computer tomography data in conjunction with machine learning models and computer simulations to assist surgeons performing knee surgery.  These are purely technical features and are not abstract… the substance of the invention is not in the display of information that was previously available in a new way or form.  The ingenuity and innovation provided by the claimed invention happens before the display of the shaded surface.”

1. A key aspect of understanding the nature of the invention on my reckoning relates to the result of the claim.  Clearly, the claimed invention serves to calculate/estimate/predict an outcome for a patient that is being considered for surgery by the surgeon.  There is no doubt that real world data is being used in the form of a 3D model, and that there is a post-operative modelling of the knee of the current patient involving the selection of possible values of tibial implant rotation and slope properties.  Ultimately this data is used with machined learned relationships between historical kinematic parameters and reported patient outcomes to arrive at predicted patient outcomes.  The data that relates to patient outcomes is discussed in the specification at [0056] as I have identified above in my discussion of the invention at [16] – [22].  There I made clear that the concept of a “patient outcome” appeared to encompass not only measurements of physical properties and parameters of a surgery but also some kind of subjective self-assessment as a measure of patient outcome. 

2. It does not appear to me controversial to suggest that the invention clearly involves a mathematic/algorithmic exercise that fundamentally simulates and predicts the outcome of a surgery.  The invention uses data which one may characterise as “real world” or even “technical”, however it seems that consideration of patentability in this context is not conclusive upon the mere presence of such data.  If that were so it would be the case that a business method that focussed upon for example, something like a financial cost benefit analysis that used data about a mechanical system, may be patentable merely because such an analysis was performed using such real world or technical data.  Instead, as I note above and as reflected in the Examiner’s Manual of Practice and Procedure, it appears that there is a distinction to be drawn between a claim to a mere algorithm implemented on a computer in the abstract sense, and the application of a mathematical process giving a result that is technical in nature in that it produces some advantage which is material, in the sense that the process belongs to a useful art as distinct from a fine art.  Where such a method addresses some technical limitation such as the improvement to a measurement of a physical characteristic, patentability will generally be found. 

3. The specification does not suggest that there is any particular improvement present in the process of 3D model generation, post-operative modelling, and virtual surgery, and the claim only presents these steps in a broad and functional manner.  As suggested by the applicant, the claimed invention focusses on providing a method that enables selection of optimal surgical parameters to improve outcome of a knee surgery.  With this in mind, the result of the mathematical algorithm currently claimed does not appear to me to lie in providing a particularly identifiable material advantage.  The estimation of a current patient outcome for various values of modelled tibial slope and rotation appears to me to be claimed at such as level as to include subjective/abstract data that is not representative of some physical property of a patient outcome that could definitively provide for patentability.  I consider the principles regarding this type of invention, where one is performing some mathematical simulation and prediction in a computer, generally requires a technical effect represented by a calculation of prediction of a physical state of an object in the real world.  At present, I do not see how the broad concept of a “patient outcome” satisfies this requirement.  Similar ideas are present in European Patent Office Guidelines for Examination where simulations, design and modelling are discussed[28] in the following terms regarding the output of a simulation:

   Simulations interacting with external physical reality 

   Computer-implemented simulations that comprise features representing an interaction with an external physical reality at the level of their input or output may provide a technical effect related to this interaction. A computer-implemented simulation that uses measurements as input may form part of an indirect measurement method that calculates or predicts the physical state of an existing real object and thus make a technical contribution regardless of what use is made of the results.

   [28] >

   The data that is the result of the presently claimed invention is used to assist the decision making of a surgeon, and on the admission of the application is potentially subjective, and thereby may not necessary possess material or objective character.  With this in mind, I find it difficult to agree that the present invention, as a matter of substance, is technical in character. 

4. Turning to the applicant’s specific submissions as summarised above I note that I do not see how the contribution lies in some kind of technical or material optimisation of surgical parameters.  The claimed method uses mathematical modelling techniques to arrive at an estimated patient outcome and I do not see how that patient outcome is inherently technical in nature.  While the applicant suggests an improvement to surgical technology, I cannot see how the claimed invention necessarily improves the efficiency of a knee surgery or improves post operative pain.  At present the claim only uses real world and historical outcome data in an algorithmic process to arrive at a potentially subjective patient outcome and I do not see how the mere present of “technical” data can change this.  I agree that the substance of the invention is not mere display of information and that the alleged ingenuity may be said to occur before display of a shaded surface.

5. Given the above, I consider that the present invention is directed towards a mere mathematical algorithm having an abstract result.  I pause at this point to make clear that it may well be the case that within the scope of the claimed invention and the concept of a “patient outcome” that there may be non-abstract subject matter present.  For example, the claim may be used to predict a particular patient outcome that could be characterised as a prediction of a particular measurable physical state or result of a surgery and in this context, may provide for patentable subject matter.  However, I would defer such a consideration to an amendment that may be made by the applicant subsequent this decision.      

6. For completeness I will continue to address the entirety of the applicant’s submissions.

   Practical and useful result

7. Clearly, whether a result is practical and useful is relevant to considering patentability.  I consider that delegate Kolev in his decision in Visa International Service Association[29] addresses this consideration rather well noting that in his case, a practical and useful result was insufficient to confer patentability on its own.  He pointed to paragraph [4] of Repipe[30], where Perram J (Nicholas and Burley JJ agreeing at [13] and [14], respectively) had “no difficulty in accepting that the invention is useful and represents a clever use of mobile devices and a server”, however His Honour still concluded that the invention “does not constitute patentable subject matter”.  To this extent it appears reasonable to accept that there must be a practical and useful result for there to be at least the potential for patentable subject matter.  However, ultimately, considerations other than broad ideas of practicality and usefulness become determinative. 

   [29] [2024] APO 34 at [138]

   [30] Repipe Pty Ltd v Commissioner of Patents [2021] FCAFC 223

8. In the present case the applicant points to the claimed method in noting that “[a]nother way of thinking about the values of rotation and slope is to consider it as an optimisation problem, where the values of rotation and slope can span a solution space”.  They point to a scientific paper[31] that shows: 

   “…that the Knee Injury and Osteoartritis Outcome Score (KOOS) pain score (sic) is improved when the surgeries are planned to use the parameters with the ‘safe zone’ as dictated by the claimed invention.  Therefore, the claimed invention provides a practical and useful result by making the entire solution space available to the surgeon and establishing a “safe zone” of optimal surgical parameters.

   [31] Patient-Specific Simulated Dynamics After Total Knee Anthroplasty Correlate with Patent-Reported Outcomes, The Journal of Anthroplasty, Vol. 33,9 (2018): 2843-2850.

9. Firstly, the concept of a “safe zone” is not defined in the claims let alone the specification.  I have no doubt that the data and shaded graphical representation generated by the claimed method is useful in practical affairs of surgery in that a picture can be generated of the dependency of “patient outcome” on tibial slope and rotation such that a surgeon can make a good strategic decision about how to proceed with the surgery.  However, this is insufficient.  I must still assess the nature of the practical and useful result to arrive at a conclusion as to patentability.  I have made such an assessment above under the previous heading and further elaborated below. 

   The nature of the computer implementation

10. Following their addressing of a practical and useful result requirement the applicant submits:

    “As such, there is ingenuity in the way that the computer is used by providing more than simply an abstract idea.  The computer is further utilised to provide a practical and useful result.  Moreover, the computer is used more than simply automating a manual process of ‘speeding up’ an otherwise manual process.”

11. I deal with this submission rather briefly in that I can see nothing in the claim that goes to the way that a computer is used as being more than automation.  The claim is to a method that:

    ·Receives data and builds a 3D model

    ·Retrieves machine learning model parameters

    ·Virtually performs surgery on the 3D model to produce a post-operative kinematic model

    ·Performs kinematic simulation on the post-operative model

    ·Estimates patient outcomes by applying machine learning parameters

    ·Generates a shaded surface

12. There is simply nothing in the claim that goes to the manner in which a computer is used to perform specific tasks.  The claimed invention is instead directed towards the mathematical/algorithmic data input and analysis that a non-descript computer is simply being used to process.  The claimed invention has no regard to how the machinery of a computer might be arranged to perform the relevant functions.  The computer is the tool which is used in its normal way to perform the algorithm.  It may even be that the algorithm could be performed manually, albeit with significant difficulty and in a time-consuming manner, and that a computer is merely used for automation of something that could otherwise be manual, but this is beside the point.  Fundamentally, there is clearly no ingenuity present in the claim which focuses upon the way in which a computer, as a physical device, is arranged to work. 

13. Following this submission the applicant takes some time to discuss the machine learning aspect of the claim.  Inviting me to consider the claimed invention as a whole, they argue that:

    “In the claimed invention, the machine learning model (more specifically, the multiple machine learning parameters indicative of machine learning performed on historical patient records) are used to estimate a current patient outcome form (sic) the multiple simulated kinematic parameters provided by the post-operative kinematic model.  As such, the machine learning model in the claimed invention is not directed towards any business innovation, but rather the machine learning model is used to quantify the effect of changes to slope and rotation of the tibial component during a knee surgery.”

14. I have already considered all the features of the claim above and noted that nothing in the claimed invention suggests to me that there is any ingenuity in the manner of implementation of the claimed method in a computer.  In focusing upon the machine learning aspect of the claims the applicant appears to seek to distinguish the present invention from business innovation, and while it may be reasonable to suggest that “business” is not an optimal characterisation of the presently claimed invention, this does not conclude the consideration.   It is always necessary to consider patentability with the understanding that mere schemes and mere mathematical algorithms are not patentable as fundamentally lacking material effect, not being technical, or even, being abstract.

15. The applicant’s own submission above points to the manner of claiming of the machine learning aspect of the invention as suggestive of the relevant features not conferring patentability.  In this regard, the claimed invention does not seek to identify a particular process of machine learning as such, but instead claims the receiving of parameters that are merely indicative of some machine learning that has been performed to identify trends between historical kinematic parameters and patient outcomes.   Here the machine learning itself is, so to speak, a black box.  It is a tool for data analysis that is characterised in the claim by the data analysed and produced, and nothing else.  With this in mind I consider that the machine learning aspect of the claim must represent the use of computer technology for its well-known and well understood purposes. 

16. To address the position that the applicant seeks to identify machine learning as a decisive identifiable element of technology in the invention, I refer to the Federal Court judgement in Repipe Pty Ltd v Commissioner of Patents[32] which addresses an invention where GPS and geofencing technology is incorporated into a claimed invention to facilitate functions of device tracking, location-based alerting, and data stamping.  McKerracher J commented on amended claims involving such geo-locating technology:

    “There is a category of amendments which relates to the addition of two claims or features concerning GPS tracking, for example, claim 4 in the 943 Patent. But the hardware sought to be added is entirely generic and the functions to be performed are functions of computer technology as it presently exists. All that is disclosed is a scheme. For example, the amended claims disclose the use of GPS technology to gather location data to be used for cross-referencing and geofencing. The fact that the existing GPS technology may be put to use in a new scheme is not sufficient to constitute patentable subject matter. The generic technology is used in the same way it is normally used, albeit that it is applied in a new scheme.”

    What I draw from this is that it is appropriate to view the machine learning aspect of present claim 1 as the use of technology for its “generic” or “well-known and understood functions”.

    [32] (No 3) [2021] FCA 31

17. The applicant also focuses upon the idea of genericness in the following submission:

    “The kinematic simulation of a post-operative kinematic model is beyond the functioning of a generic computer.  More specifically, building a 3D model of the patient’s knee, multiple machine learning model parameters indicative of machine learning performed on historical patient records, a post-operative kinematic model and perform (sic) a kinematic simulation of a post-operative kinematic model are not well-known functions of a generic computer.”

    “Further, we submit it is not routine for a knee surgeon to configure a post-operative kinematic model based on a 3D model of the patient’s knee, perform a simulation based on the post-operative kinematic model and estimate patient outcome by applying a machine learning model to the simulation results.  In the claimed invention, the CT data is used in a non-conventional manner, as it is used to build a 3D model of the patient’s knee and subsequently configure a post-operative kinematic model to predict the patient outcome as a result of a future knee surgery.”

18. I do not see the kinematic simulation operating on a computer in the claim as anything more than an idea of a program that is necessarily operating on a non-descript computer.  This is precisely the manner in which abstract ideas may be implemented on generic computer technology merely at the level of an idea for a program with no consideration given to how the computer is to function.  This resonates with the decision in Encompass Corporation Pty Ltd v InfoTrack Pty Ltd[33] where an expanded Full Court found that an invention was: 

    “…really an idea for a computer program, it being left (as we have said) to the user to carry out that idea in an electronic processing device.”

    The same can be said of the other features discussed in the submission as together constituting the algorithm.  While the features might arguably not be algorithmic activities that are commonly done in the general sense, they are nothing more than elements of an idea for a program operating merely as functions on a non-descript computer.  They must logically be conducted using well-known computer functions as there are simply no specific computer operations present in the claimed invention.  

    [33] [2019] FCAFC 161 at [101]

19. To the allegation of a lack of routineness of the relevant features the same logic applies.  The present consideration is not one of inventive step but a characterisation of the claimed invention as a matter of substance to determine whether there is any technical ingenuity present.  I have already determined that I do not see the necessary patentable technicality in the algorithm or significantly, in its end result.  For these same reasons, it is clear to me that there is no improvement to a computer or to computer technology.  The applicant adds the additional argument that “the processor can derive a kinematic model, which is a simplified representation that disregards 3D details that are not needed when considering the movement of the knee joint” and this “improves the processing capabilities of the computer and also reduces memory requirements of the computer”.  Plainly the mere reduction of data to be processed does not in any way change how a computer operates.  It merely means a computer must do less work and use less storage to do that work.  This cannot be an improvement to computer technology. 

    Addressing a technical problem outside of the computer

20. The applicant submits that the claimed invention in substance lies in an improvement in a technical field outside of a computer.  They pointed to a matter decided by a delegate of Commissioner who considered that an invention in substance directed to a testing procedure that reduced the number of reference samples required to be used resulted in a material effect outside a computer and thereby was a manner of manufacture[34].  They argue that when the presently claimed invention is put into effect during a knee surgery, the result of improved patient outcome ought to qualify as an improvement in a technical field outside of a computer.

    [34] Bio-rad Laboratories, Inc. [2018] APO 24

21. As already assessed earlier, I consider that the nature of the effect in the present matter as provided by the claimed invention to be too broad in encompassing an abstract result.  I do not see difficultly in contrasting this with the delegate’s findings in the above-mentioned decision where he found a claimed invention patentable subject matter on the basis of it inevitably producing a specific technical effect.  The effect of the present invention is not limited to some kind of physical improvement or measurement of a physical variable.  As I noted earlier, it may well be the case that within the scope of the claimed invention and the concept of a “patient outcome” there may be non-abstract subject matter present.  For example, the claim may be used to predict a particular patient outcome that could be characterised as a prediction of a particular measurable physical state or result of a surgery and in this context, may provide for patentable subject matter.  However, such limitation is not currently claimed and thus I do not see the claim as addressing a technical problem or necessarily being in a particular technical field. 

    Is the invention technical in nature?

1. It can easily be seen that many considerations regarding patentability of computer implemented inventions are heavily overlapping and addressing each of them in turn can become somewhat repetitious.  In arguing the invention is technical in nature the applicant points to the claimed invention as increasing “the amount of medical information in a CT scan to provide more accurate surgical parameters”.  They also point to the features of building a 3D model with tomography data, virtually performing surgery, and performing kinematic simulation as not being abstract features, and argue that assisting a surgeon performing knee surgery and improving patient quality of care means the invention is of technical nature.

2. Fundamentally I think my analysis above already addresses much of these points.  There does not appear to be any ingenuity in the steps of building a 3D model, virtually performing surgery or performing kinematic simulation as such.  The true nature of the invention lies in algorithmic steps and the data that is being processed to arrive at an improved estimation and representation of a “patient outcome” which itself does not appear to me necessarily technical in nature.  I do not see anything in the idea of assisting a surgeon to perform knee surgery and improving patient care to itself be necessarily technical at the level of generality these are put.  If on the other hand the method of the claimed invention predicted a particular measurable physical state of a patient’s knee after surgery, then an invention of a technical nature may be present.  However, as noted, the claimed invention at present is not so narrowly defined.  

   Conclusion regarding manner of manufacture

3. Considering my discussion above I find that the claimed invention as presently drafted is not to a manner of manufacture.  I consider that the claimed invention is characterised as a mere scheme/algorithm involving particular modelling steps that does not produce a technical effect and is therefore as a matter of substance, not technical in nature. 

4. While there may by a practical and useful nature to the invention, it simply does not appear to me to represent technical innovation, and there is no adaptation or improvement to computer technology present in the claims that accommodates the implementation of the mathematical algorithm.  The most key aspect of my finding in this regard is that the computer implemented algorithm of the claimed invention merely produces a result with “subjective”, “non-technical” or “abstract” character. 

5. It may be the case that the claimed invention, in relation to the resultant data of a “patient outcome”, could be amended so as to limit the claimed invention to adequately technical subject matter.  For example, the claim may be used to predict a particular patient outcome that could be characterised as a prediction of a particular measurable physical state or result of a surgery and in this context, may provide for patentable subject matter.  However, at this stage the claimed invention is not so limited. 

   NOVELTY

   Legal Principles

6. For the purposes of subsection 7(1) of the Patents Act (1990), an invention is to be taken to be novel when compared with the prior art base unless it is not novel in the light of any one of the pieces of prior art information.  It is well established that the general test for anticipation is the reverse infringement test. The classic formulation of this test is that given by Aickin J in Meyers Taylor Pty Ltd v Vicarr Industries Ltd[35]:

   “The basic test for anticipation or want of novelty is the same as that for infringement and generally one can properly ask oneself whether the alleged anticipation would, if the patent were valid, constitute an infringement.”

   [35] [1977] HCA 19 at [20]

7. This test is satisfied if the alleged anticipation discloses all of the essential features of the invention as claimed with “clear and unmistakable directions to do what the patentee claims to have invented”[36]:

   “A signpost, however clear, upon the road to the patentee’s invention will not suffice. The prior inventor must be clearly shown to have planted his flag at the precise destination before the patentee”.
   
   

   [36] The General Tire & Rubber Company v The Firestone Tyre and Rubber Company Limited [1972] RPC 457 at [486]

8. The examiner has maintained a novelty objection on the basis of document D1 (US 2010/0332194 A1) along the following lines with my emphasis in underline:

   “A method for assisting a surgeon with a knee surgery, the method comprising: receiving computer tomography data of a knee of a patient: see para [0050].

   building a 3D model of the knee of the patient using the computer tomography data: see [0030] ‘The data cloud 110 may be populated for a specific patient, using patient data such as a digital X-ray, CT scan or a MRI scan.’ Also see figure 3 and [0050] ‘Additional patient data, for example X-ray, CT scan or an MRI scan, may be extracted from the data cloud 110. The patient data (presented on patient data tab 310), regardless of source (e.g., a simulated subject or a particular patient, etc.) is the starting point for the rest of the simulation.’

   receiving user input from the surgeon, the user input comprising an identifier of a knee implant, the knee implant comprising a tibial component: see paras [0010] and [0034] where the implant is selected by the surgeon. Also see [0056] ‘tibial implant or component 730’.

   retrieving multiple machine learning model parameters indicative of machine learning performed on historical patient records, the historical patient records comprising multiple historical kinematic parameters of each of multiple historical patients as inputs and a reported patient outcome for each historical patient as output, the machine learning model parameters being indicative of a relationship between the multiple historical kinematic parameters and the reported patient outcome: see para [0027] ‘Turning to FIG. 1, data flow for a system 100 in accordance with an embodiment is shown. System 100 includes a database or data cloud 110 of surgical outcomes for various surgical implantation and component alignment techniques.’

   for each of multiple values of rotation of the tibial component and slope of the tibial component: see figures 4 and 7, and paras [0051] and [0056].

   configuring a post-operative kinematic model of the knee of the patient using the 3D model, the user input and the value of the rotation and the value of the slope: see para [0031].

   wherein configuring the post-operative kinematic model comprises: virtually performing a surgery on the 3D model by introducing cut surfaces to change a shape of bones in the 3D model, adding a shape of the knee implant, and simplifying the 3D model to the post-operative kinematic model: see figures 4, 5, 6 and 7, and [0051]-[0056].

   performing a kinematic simulation on the post-operative kinematic model to determine multiple simulated kinematic parameters: see [0058]-[0060].

   and estimating a current patient outcome by applying the multiple machine learning model parameters to the determined multiple simulated kinematic parameters: see [0061]-[0064].

   and generating a shaded surface on portions of a user interface, the portions spanning the multiple values of rotation of the tibial component and slope of the tibial component on the user interface to graphically represent the estimated current patient outcome for each of the multiple values of rotation of the tibial component and the slope of the tibial component: see figure 8 and [0061] ‘Each indicator shows the current value as a vertical line along a scale with green and red ranges. Values in the green range indicate a positive result, and values in the red range suggest that a change to the surgical parameters may be advisable.’ See figure 8 where the surgical parameters include ‘tibial rotation’ and ‘posterior slope’. As discussed in [0051], ‘posterior slope’ relates to the tibia.”

9. In response the applicant suggests that document D1 does not disclose:

   ·     retrieving multiple machine learning model parameters indicative of machine learning performed on historical patient records, the historical patient records comprising multiple historical kinematic parameters of each of multiple historical patients as inputs and a reported patient outcome for each historical patient as output, the machine learning model parameters being indicative of a relationship between the multiple historical kinematic parameters and the reported patient outcome;

   ·     building a 3D model of the patient's knee to represent the computer tomography data;

   ·     further performing for each of multiple values of rotation of the tibial component and slope of the tibial component the steps of: configuring a post-operative kinematic model of the current patient's knee based on the computer tomography data, the user input and that value of the rotation and the slope, wherein configuring the post-operative kinematic model comprises: virtually performing the surgery by introducing cut surfaces to change a shape of bones in the 3D model and adding a shape of the knee implant;

   ·     performing a kinematic simulation based on the post-operative kinematic model to determine multiple simulated kinematic parameters; and

   ·     estimating a current patient outcome by applying the multiple machine learning model parameters to the multiple simulated kinematic parameters of the current patient.

10. However, their arguments only focus on the absence of machine learning model parameters as follows:

    “For example, nowhere does D1 disclose “retrieving multiple machine learning model parameters”, “applying the multiple machine learning model parameters to the multiple simulated kinematic parameters of the current patient” or any aspect of machine learning. The examiner points to paragraph [0027] as disclosure of “retrieving multiple machine learning model parameters”. However, paragraph [0027] recites “a database or data cloud 110 of surgical outcomes for various surgical implantation and component alignment techniques.” This is not related or indicative of machine learning.

    Further, the examiner points to paragraphs [0061]-[0064] as disclosure of “estimating a current patient outcome by applying the multiple machine learning model parameters to the multiple simulated kinematic parameters of the current patient”. However, paragraph [0061] recites “The outcome prediction values 840 are calculated (or retrieved from a database) based on the patient parameters and the surgical parameters entered in the application.” The patient parameters and the surgical parameters are not multiple machine learning model parameters indicative of machine learning performed on historical patient records. Moreover, paragraphs [0062]-[0064] do not recite anything to do with machine learning.”

11. I consider it clear that the focus of the applicant’s rebuttal as a distinguishment from document D1 turns to the fact that the claimed invention is directed towards the use of a machine learning model to estimate the relationship between kinematic parameters (such as tibial rotation and slope) and patient outcome.  The applicant notes that document D1 does not discuss any use of machine learning.  I will consider document D1 in detail.

12. Document D1 is directed to a biomechanical modelling system that simulates physical parameters of a surgery using a preloaded 3D representation of a patient’s knee.  Surgical parameters, such as properties of a tibial implant, are received as input and correlations are identified in order to estimate outcome information for a potential surgery[37].  The examiner is plainly correct in that the document discloses a method for assisting a surgeon with a knee surgery, the method comprising: receiving computer tomography data of a knee of a patient and building a 3D model of the knee of the patient using the computer tomography data[38], and receiving user input from the surgeon, the user input comprising an identifier of a knee implant, the knee implant comprising a tibial component[39].

    [37] Abstract, Figures, and paragraph [0056].

    [38] Paragraphs [0030] and [0050]

    [39] Paragraphs [0010], [0034] and [0056]

13. The key aspect of the applicant’s argument focusses on the feature of retrieving multiple machine the machine learning model parameters generated from historical patent records, these parameters being indicative of a relationship between the multiple historical kinematic parameters and the reported patient outcome.  In terms of this feature the examiner points to paragraph [0027] of the document which states:

    Turning to FIG. 1, data flow for a system 100 in accordance with an embodiment is shown. System 100 includes a database or data cloud 110 of surgical outcomes for various surgical implantation and component alignment techniques. For example the data can be generated using software based biomechanical modeling (e.g, using analysis engine 120) that has been used extensively in product development department for many major orthopedic companies for several years. For example, analysis engine 120 may include a computer simulation tool, such as LifeMOD/KneeSIM, previously run for a range of test cases expected to include a case of interest. Generally, data cloud 110 uses the input parameters to incorporate the results from analysis engine 120 for a case with similar inputs, for example, interpolating between similar cases.

14. Clearly there is no identification of using machine learning to develop a model of the relationship between surgical implant parameters and surgical/patient outcomes.  At most reference is made to an “analysis engine” that may include a computer simulation tool such as LifeMOD/KneeSIM.  This broad reference to an analysis engine remains throughout the citation and I agree that there is no clear and unmistakeable direction to machine learning as such.  I thus agree with the applicant that there is no reference in the citation to the use of machine learning to model the relationships in the data of historical patient records. 

15. Subsequent features of claim 1 warranting consideration relate to the iterative simulated surgery involving multiple values of tibial rotation and slope.  The claimed invention currently requires that for multiple values of rotation and slope of a tibial component, a model of the patient’s knee is generated via simulated cut surfaces of bones in the model whereby the shape of the knee implant is added into the model.  Through this process the simulated parameters of the surgery are identified.  Paragraph [0051] of document D1 discusses that across the various tibial settings, cuts can be made to simulate surgery and that relevant parameters are updated and used at the end of the procedure to calculate results.  Various outcome prediction values for the surgery are calculated (or retrieved from the database) on the basis of patient parameters and surgical parameters entered into the software application that is simulating the surgery[40].   The applicant does not appear to take issue with the examiner’s suggestion of disclosure of these features of the claim.  What remains clear is that there is no disclosure that patient outcomes are estimated by applying machine learning model parameters to multiple simulated kinematic parameters of the current patient.  Instead, as mentioned above, the useful relationships are generated using an analysis engine and held in a database or data cloud. 

    [40] Paragraph [0061]

16. The final feature of claim 1 that was before the examiner for consideration related to the generation of a shade surface on the user interface with portions spanning multiple values of rotation and slope of a tibial component to graphically represent patient outcomes for these tibial parameters.  I agree with the examiner that FIG 8 of document D1 reproduced above shows selectable values for these tibial parameters (presented as posterior slope and tibial rotation) along with corresponding patient outcomes (840).  While coloured shading is not depicted in the image paragraph [0061] notes that ‘Each indicator shows the current value as a vertical line along a scale with green and red ranges. Values in the green range indicate a positive result, and values in the red range suggest that a change to the surgical parameters may be advisable.’  Again, no issue is taken by the applicant with this interpretation by the examiner. 

17. As discussed earlier in the decision, there is a new feature added by amendment this being reducing one or more 3D details of the 3D model to simplify the 3D model to the postoperative kinematic model.  I can find no reference to this reduction in 3D details in document D1. 

18. In summary, on balance I consider all features of document D1 are disclosed except the use of machine learning to model the claimed data relationships, and the reduction of one or more details of the initial patient 3D model to simply it to the post operative kinematic model.  The claimed invention is novel in view of document D1 and the examiner’s objection under this ground is not sustainable.  I will consider these features further under the ground of inventive step. 

    INVENTIVE STEP

    Legal Principles

19. The test for obviousness was provided by Justice Aicken in Wellcome Foundation Ltd v VR Laboratories (Aust) Pty Ltd[41] as follows:

    “The test is whether the hypothetical addressee faced with the same problem would have taken as a matter of routine whatever steps might have led from the prior art to the invention, whether they be the steps of the inventor or not.”

    [41] [1981] HCA 12 at [45]

20. The High Court in Aktiebolaget Hässle v Alphapharm Pty Ltd[42] approved this approach, in addition to that taken in Olin Mathieson Chemical Corporation v Biorex Laboratories Ltd[43]in which Graham J had posed the question:

    “Would the notional research group at the relevant date in all the circumstances directly be led as a matter of course to try [the claimed invention] in the expectation that it might well produce a useful [desired result]?”

    [42] [2002] HCA 59 at [51]-[53]

    [43] [1970] RPC 157 at [187]

21. The usual approach to determining inventive step is the problem-solution approach.  Once the problem has been formulated and the common general knowledge and the prior art base has been determined, the question of whether the claimed solution is obvious must be addressed.

    Examiner’s Objections

22. In objecting to a lack of novelty of the claimed invention against document D1 in the first report, the examiner also argued that those reasons gave rise to a lack of inventive step.  I have already found that the novelty objection was not sustainable and identified features that I consider are not disclosed in document D1.  The examiner also raises an inventive step objection on the basis of document D3 as follows:

    “Regarding independent claims 1 and 18, D3 discloses or renders obvious:

    A method for assisting a surgeon with a knee surgery, the method comprising: receiving computer tomography data of a knee of a patient: see [0042].

    building a 3D model of the knee of the patient using the computer tomography data: see [0074].

    receiving user input from the surgeon, the user input comprising an identifier of a knee implant, the knee implant comprising a tibial component: see [0074].

    retrieving multiple machine learning model parameters indicative of machine learning performed on historical patient records, the historical patient records comprising multiple historical kinematic parameters of each of multiple historical patients as inputs and a reported patient outcome for each historical patient as output, the machine learning model parameters being indicative of a relationship between the multiple historical kinematic parameters and the reported patient outcome: see paras [0092], [0094] and [0098].

    for each of multiple values of rotation of the tibial component and slope of the tibial component; configuring a post-operative kinematic model of the knee of the patient using the 3D model, the user input and the value of the rotation and the value of the slope: see paras [0072]-[0074]. D3 fails to explicitly teach simulating multiple values of a slope of the tibial component. However it would have been obvious to a person skilled in the art that the system of D3 may be utilised to additionally simulate the slope of the tibial component as claimed. Consequently this feature cannot contribute an inventive step over D3.

    wherein configuring the post-operative kinematic model comprises: virtually performing a surgery on the 3D model by introducing cut surfaces to change a shape of bones in the 3D model, adding a shape of the knee implant, and simplifying the 3D model to the post-operative kinematic model: see [0074].

    performing a kinematic simulation on the post-operative kinematic model to determine multiple simulated kinematic parameters: see [0074].

    and estimating a current patient outcome by applying the multiple machine learning model parameters to the determined multiple simulated kinematic parameters: see [0102] ‘the method 100 comprises the step 110 of comparing the plurality of modified 3D knee joint data to a database of 3D knee joint data to determine which treatment from the plurality of treatments will produce optimal results as a result of the treatment and/or surgery.’

    and generating a shaded surface on portions of a user interface, the portions spanning the multiple values of rotation of the tibial component and slope of the tibial component on the user interface to graphically represent the estimated current patient outcome for each of the multiple values of rotation of the tibial component and the slope of the tibial component: see [0018]-[0019] where the result from the potential treatment plans are evaluated. D3 fails to teach displaying these results as a ‘shaded surface’ as claimed, however this does not represent an inventive step. The use of ‘shaded surfaces’ to represent the relationships between values is common general knowledge in the art. Therefore, the use of this well-known data visualisation technique would be obvious to the person skilled in the art when seeking to visually represent to the user the relationship between the surgical plans and the simulated outcomes. Therefore this feature fails to contribute an inventive step over the disclosure of D3.”

1. I will deal with both of these documents under this ground.      

   Applicant’s Submissions

2. Regarding document D1 the applicant submits:

   “Given that D1 does not disclose many of the features of claim 1, we submit that claim 1 is inventive over D1. In particular, the system of D1 cannot achieve the same advantages provided by the claimed invention discussed earlier. For example, in D1, the surgeon only sees a large number of simulation outputs and but has no way of knowing how to adjust the surgery parameters. However, in the claimed invention, the surgeon can easily visualise the association between the surgical parameters and the patient outcomes, enabling the surgeon to easily understand and modify the surgical parameters to achieve an improved patient outcome.”

3. Regarding document D3 the applicant focusses on features identified by the examiner as not being disclosed by the citation and argues that the examiner has failed to demonstrate that these features are obvious.  I will discuss these submissions in more detail below.

   Consideration

   Document D1

4. I have considered document D1 above under the ground of novelty.  I found that the following features are not disclosed:

   ·Machine learning to model the claimed data relationships and create machine learning model parameters that are used in the process

   ·The reduction of one or more details of the initial patient 3D model to simplify it to the post operative kinematic model

5. Firstly, I cannot see how the applicant’s submissions as to an inventive step regarding document D1 address these differences between the prior art and the claimed invention.  Inventive step is assessed by understanding whether the features not disclosed by a relevant citation are obvious to the person skilled in the art, who in the present circumstances would appear to include a person with experience in computer modelling the human body and surgical parameters.  The submission of the applicant does not take this approach.  With this in mind I turn to each of the relevant features.

6. To date there has been no discussion as to the potential obviousness of the features relating to machine learning parameters in view of document D1.  The applicant has not taken the opportunity in making submissions for this hearing to clearly state why the addition of such a feature would have been inventive.  Clearly the claims broadly encompass the concept whereby data relationships between implant parameters and surgical outcomes are generated using some kind of machine learning.  In contrast document D1 generates data relationships between implant parameters (such as tibial slope and rotation) and surgical outcomes using an “analysis engine” exemplified by biomechanical software modelling wherein interpolation is used to generate data between similar cases with similar inputs[44].   

   [44] Document D1 at [0027]

7. The specification discusses problems in the art relate to the need for improved data regarding patient outcomes to optimise surgical decision making[45].  The specification[46] notes that in order to increase the success of patient outcomes, surgeons might make small changes to surgery parameters, but they rarely have the tools available to allow them to investigate how parameter changes might have a positive impact for a particular patient.  The applicant identifies the problem in the art in their submissions as follows:

   “The problem addressed by the claimed invention is that a surgeon cannot accurately determine the appropriate surgical parameters (such as the rotation and slope of the tibial component) from a CT scan, due to the limited amount of information available in the static information.”

   [45] Specification at [0003]

   [46] Specification at [0004]

8. This appears a fair summary of the problem which on my reckoning clearly focusses upon an improved data analysis and presentation for a surgeon to determine which parameters to choose to optimise patient outcomes.

9. With this problem in mind, I can see no reason why it would be inventive to simply substitute a piece of modelling software for generating relevant parameters as disclosed in document D1, with some kind of generic machine learning model, as per the claimed invention.  I understand machine learning to be a modern and useful tool for data analysis that could easily substitute for a software coded mathematical model.  Machine learning has been a tool for data analysis for many decades[47] and as such, I see no reason why its prevalence and usefulness would not have been common general knowledge in the art at the priority date in 2016.  The specification discusses no barriers to the mere use of machine learning. 

   [47] >

   In seeking to improve data analysis such that a surgeon might be equipped with a better understanding of connection between historical parameters of tibial rotation and slope and patient outcome I consider that it would have been obvious to the person skilled in the art to seek to use machine learning techniques.  I consider that machine learning is merely a type of “analysis engine”.

10. The remaining feature that I consider is not disclosed by document D1 is the reduction of one or more details of the initial patient 3D model to simplify it to the post operative kinematic model.  This is a new feature, and it is not discussed in any detail regarding inventive step by the applicant in their submissions.  Considering the claim as a whole the feature is rather peripheral to its function and appears to me to be broadly claiming the idea that the data complexity in the patient 3D model is simply reduced in some non-descript way in generating a post operative kinematic model.  I see the feature as nothing more than removing data that may not be necessary or relevant to further analysis.  I see no reason why a person skilled in the art of computer modelling would not see that as a mere matter of routine optimisation of data processing.

11. I consider that an inventive step objection is sustainable against the claimed invention, in particular independent claims 1 and 18, in light of document D1 and the common general knowledge.   I will not consider the dependent claims further noting that the applicant does not make submissions refuting the examiners objection to the obviousness of these claims and that an amendment would be required to address my findings.   

    Document D3

12. I will briefly deal with the examiner’s objection against document D3.  The examiner’s objection in relation to document D3 (US 2013/0185310 A1) is as follows:

    “Claims 1-13 and 16-19 do not involve an inventive step over the disclosure of citation D3 when viewed in light of the common general knowledge in the art. Regarding independent claims 1 and 18, D3 discloses or renders obvious:

    A method for assisting a surgeon with a knee surgery, the method comprising: receiving computer tomography data of a knee of a patient: see [0042].

    building a 3D model of the knee of the patient using the computer tomography data: see [0074].

    receiving user input from the surgeon, the user input comprising an identifier of a knee implant, the knee implant comprising a tibial component: see [0074].

    retrieving multiple machine learning model parameters indicative of machine learning performed on historical patient records, the historical patient records comprising multiple historical kinematic parameters of each of multiple historical patients as inputs and a reported patient outcome for each historical patient as output, the machine learning model parameters being indicative of a relationship between the multiple historical kinematic parameters and the reported patient outcome: see paras [0092], [0094] and [0098].

    for each of multiple values of rotation of the tibial component and slope of the tibial component; configuring a post-operative kinematic model of the knee of the patient using the 3D model, the user input and the value of the rotation and the value of the slope: see paras [0072]-[0074]. D3 fails to explicitly teach simulating multiple values of a slope of the tibial component. However it would have been obvious to a person skilled in the art that the system of D3 may be utilised to additionally simulate the slope of the tibial component as claimed. Consequently this feature cannot contribute an inventive step over D3.

    wherein configuring the post-operative kinematic model comprises: virtually performing a surgery on the 3D model by introducing cut surfaces to change a shape of bones in the 3D model, adding a shape of the knee implant, and simplifying the 3D model to the post-operative kinematic model: see [0074].

    performing a kinematic simulation on the post-operative kinematic model to determine multiple simulated kinematic parameters: see [0074].

    and estimating a current patient outcome by applying the multiple machine learning model parameters to the determined multiple simulated kinematic parameters: see [0102] ‘the method 100 comprises the step 110 of comparing the plurality of modified 3D knee joint data to a database of 3D knee joint data to determine which treatment from the plurality of treatments will produce optimal results as a result of the treatment and/or surgery.’

    and generating a shaded surface on portions of a user interface, the portions spanning the multiple values of rotation of the tibial component and slope of the tibial component on the user interface to graphically represent the estimated current patient outcome for each of the multiple values of rotation of the tibial component and the slope of the tibial component: see [0018]-[0019] where the result from the potential treatment plans are evaluated. D3 fails to teach displaying these results as a ‘shaded surface’ as claimed, however this does not represent an inventive step. The use of ‘shaded surfaces’ to represent the relationships between values is common general knowledge in the art. Therefore, the use of this well-known data visualisation technique would be obvious to the person skilled in the art when seeking to visually represent to the user the relationship between the surgical plans and the simulated outcomes. Therefore this feature fails to contribute an inventive step over the disclosure of D3.

13. Regarding this last feature relating to the shaded surface being common general knowledge the applicant submits:

    “…the examiner has provided no evidence that this feature is common general knowledge. It is noted that, according to Australian Examiner’s Manual section 2.5.2.1.3, common general knowledge must be formulated on the basis of written information. However, it is not sufficient to provide written information as a source of common general knowledge if there is no evidence that the written information is generally accepted by those skilled in the art, as discussed in British Acoustic Films Ld v Nettlefold Productions (1936) 53 RPC 221. Therefore, it is not sufficient to simply state that the above feature is common general knowledge for the skilled person in the context of the subject matter of pending claim 1, without evidence of written information and evidence that the written information is generally accepted by those skilled in the art.”

100. I appreciate the point made here by the applicant.  There is no information before me to suggest that the use of a shaded surface to display surgical parameters against outcomes is common general knowledge to the person skilled in the art being one experienced in computer modelling the human body and surgical parameters.  In my opinion, the addition of this feature to the claimed invention is not of the same nature of the feature of claim 1 relating to machine learning that I discussed in relation to document D1.  Without supporting documentary information, I cannot see how I can consider the use of a shaded surface to depict surgical outcome as being common general knowledge.  It may be the case that such documentation exists or so as to render the feature obvious, however such information is not before me as demonstrating such common general knowledge.  On the contrary, on my reckoning machine learning is merely a form of “analysis engine” that a person skilled in the art would substitute for the type analysis engine specifically described in document D1. 

101. Turning now to the feature whereby for each of multiple values of rotation of the tibial component and slope of the tibial component there is configured a post-operative kinematic model of the knee of the patient using the 3D model, the user input and the value of the rotation and the value of the slope.  The examiner acknowledges that D3 fails to disclose simulating values of a slope of the tibial component but suggests it would have been obvious to a person skilled in the art that the system of D3 may be utilised to additionally simulate the slope of the tibial component as claimed.  The applicant responds by pointing out that the examiner has not provided any evidence that this would be obvious to the person skilled in the art.  They argue that simply stating that a feature would be obvious to the person skilled in the art without justification is not sufficient to assert that a claim lacks inventive step.

102. I agree with the applicant.  It is not discussed in document D3 that use of the tibial slope is an important parameter for simulation and there is nothing to suggest that this knowledge is common general knowledge. 

103. Hence at least for the reasons above, the objection under the ground of inventive step formulated against document D3 in the light of the common general knowledge as presented by the examiner is not sustainable. 

CONCLUSION

104. I consider some objections of the examiner are sustainable.  I consider the claimed invention lacks an inventive step in view of document D1 and is not for a manner of manufacture. 

105. In accordance with paragraph 13.4(1)(g) and sub-regulation 13.4(3) of the Patent Regulations 1991 I provide six (6) months from the date of this decision for the applicant to gain acceptance.

Dr N. R. Madsen

Deputy Commissioner of Patents