DETAILED ACTION
Claims 1-4, 6, 8-16, 18, 20-24 and 26-28 are currently pending.
Response to Arguments
Applicant's arguments filed 9/1/26 have been fully considered but they are not persuasive.
The Applicant argues on page 11 of the response in essence that: As submitted before, generating fractional KL grades from multiple feature values requires evaluating combinatorial number of feature splits across multi- dimensional feature spaces, which cannot be performed via pen and paper or in a person's mind.
A doctor has traditionally performed analysis of an x-ray image to classify the severity of osteoarthritis using a Kellgren-Lawernce grade. Such analysis considers osteophytes and joint space narrowing to determine the grade. The Applicant’s Background Section demonstrates this in the discussions of definitions of KL grades on page 1 and 2. Thus, the generating of feature values for osteophytes and joint space narrowing does not preclude the steps from being performed in the human mind.
The Applicant argues on pages 11 and 12 of the response in essence that: Besides, the present application is not simply applying generic machine learning to new data environments. In contrast, the claimed method shows prediction performance superior to previously available approaches. As will be discussed in more detail in NONOBVIOUSNESS section, a regression tree provides inherent interpretability by decomposing each individual prediction into feature contributions, and the feature values selected in the present application indeed provide better prediction ability compared to prior art references. Moreover, selecting regression tree-based model for fractional KL grade predictions further provides the model with an intrinsic ability to establish a feature importance estimation module and perform feature importance analysis (e.g., SHAP) for each single prediction. Those advantageous effects are not taught or suggested by the cited prior art references.
The recitation of a regression tree algorithm does not provide additional elements that are sufficient to amount to significantly more than the judicial exception because the recitation involves no more than a generic computer performing generic computer functions that are well understood, routine and conventional activities previously known in the industry. Regression tree algorithms are commonly used in artificial intelligence and have been known since 1984. See Breiman, L., Friedman, J., Olshen, R.A., & Stone, C.J. (1984). Classification and Regression Trees (1st ed.). Chapman and Hall/CRC. https://doi.org/10.1201/9781315139470. Furthermore, “patents that do no more than claim the application of generic machine learning to new data environments, without disclosing improvements to the machine learning models to be applied, are patent ineligible under § 101.” Recentive Analytics, Inc. v. Fox Corp., No. 2023-2437 (Fed. Cir. Apr. 18, 2025).
The Applicant argues on pages 12 and 13 of the response in essence that: First of all, what discloses in Shamir on page 1 is the standard for a doctor to perform KL grading, not Shamir's approach to implementing KL prediction. Shamir's model does not apply characteristic values characterizing joint status (e.g., joint space narrowing and osteophyte) to generate predicted KL grades. Specifically, Shamir discloses in Section 2 (Methods) that "It]he image analysis method used in this study is wndchrm, which has been found effective for the analysis and detection of osteoarthritis using knee X-rays." "The image features used by wndchrm can be classified into high contrast features, textures (e.g., Haralick, Tamura), statistical distribution of the pixel values (e.g., multiscale histogram, first four moments), and polynomial decomposition of the image." In other words, the features used in Shamir are image features of an image, instead of OA-related features.
Shamir discloses computing features that include osteophytes and narrowing of part or all of the tibial-femoral joint space (page 1). While Applicant argues that these are image features and not characteristic values characterizing joint status, the image features must include a characterization of the joint status in order to derive a KL grade.
The Applicant argues on page 14 of the response in essence that: The Examiner indicates on page 3 of the Office Action that "[w]hile Shamir discusses rejecting some of the feature values, the feature values the remain in will necessarily include joint space narrowing and osteophyte feature values." Applicant respectfully disagrees. Besides the above arguments suggesting that Shamir does not apply OA-related features for KL prediction, Shamir also describes on page 3 (e.g., FIGS. 2-3) that only the center part of a knee joint is included in the image to be analyzed. This might exclude existing osteophytes from consideration. Shamir's disclosure on page 4 that "KL grade 1 is determined by the presence of mild osteophytes which are not present in the joint itself' confirms this.
Shamir discloses computing features that include osteophytes and narrowing of part or all of the tibial-femoral joint space (page 1). It is irrelevant that in certain circumstances Shamir may or may not consider the presence of osteophytes.
The Applicant argues on page 14 of the response in essence that: On the other hand, a regression tree model provides inherent interpretability by decomposing each individual prediction into feature contributions along its decision path, thus enabling construction of a feature importance estimation module for the established first machine learning model. Other model architectures (such as CNN as exemplified in Kim) do not show this characteristic. It is respectfully submitted that the advantageous characteristic of being able to decompose each individual prediction into feature contributions is not taught or suggested by any of the cited references.
Before the effective filing date of the claimed invention, it would have obvious to a person of ordinary skill in the art to use a regression tree algorithm. The motivation for doing so would have been to utilize a widely used manner of improving the results of a machine learning model. While Applicant argues that deep learning models do not traditionally use regression tree algorithms, regression tree algorithms are commonly used in artificial intelligence and have been known since 1984. See Breiman, L., Friedman, J., Olshen, R.A., & Stone, C.J. (1984). Classification and Regression Trees (1st ed.). Chapman and Hall/CRC. https://doi.org/10.1201/9781315139470.
The Applicant argues on pages 15 and 16 of the response in essence that: Since the plurality of feature importance indicators indicates contributions of the plurality of features to the quantitative KL grade, and since the quantitative KL grade is generated by the first machine learning model (which is an established model), the claim language of claim 1 and claim 13 automatically requires the feature importance indicators to be generated only after the machine learning model has been established.
A skilled artisan would have been motivated to apply the feature importance estimation of Shamir into the first machine learning model of Kim. The motivation for doing so would have been to improve the diagnosis of osteoarthritis by analyzing the features that are most important in diagnosing osteoarthritis. In response to Applicant's arguments against the references individually, one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-4, 6, 8-16, 18, 20-24 and 26-28 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more. The claims recite the abstract idea of grading an image to diagnose osteoarthritis.
The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the receiving of a plurality of features from a skeletal image is simply appending well-understood, routine and conventional activities previously known in the industry. Doctors’ analysis of an x-ray image to grade osteoarthritis is commonplace in the art.
Furthermore, the steps of “generating a quantitative Kellgren-Lawrence (KL) grade based on the plurality of feature values”, “the quantitative KL grade has an integer part and a fractional part”, “the quantitative KL grade is used to diagnose osteoarthritis” and “the plurality of feature values comprises at least one joint space narrowing (JSN) feature value and at least one osteophyte (OST) feature value” do not preclude the steps from being performed in the human mind. The dependent claims likewise do not preclude the steps from being performed in the human mind, and are directed to the abstract idea of grading an image to diagnose osteoarthritis.
The addition of a computer running a grading module, analysis logic or a machine learning trained by a regression tree algorithm does not provide additional elements that are sufficient to amount to significantly more than the judicial exception because the recitations to hardware involve no more than a generic computer performing generic computer functions that are well understood, routine and conventional activities previously known in the industry. That is, other than reciting “by a processor,” nothing in the claim precludes the steps from practically being performed in the human mind. Furthermore, “patents that do no more than claim the application of generic machine learning to new data environments, without disclosing improvements to the machine learning models to be applied, are patent ineligible under § 101.” Recentive Analytics, Inc. v. Fox Corp., No. 2023-2437 (Fed. Cir. Apr. 18, 2025).
The claims do not provide an inventive concept as they do not provide an improvement to any type of particular machine. Thus, even when viewed as a whole, nothing in the claim adds significantly more (i.e., an inventive concept) to the abstract idea.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-4, 6, 8, 11-14, 16, 18, 20, 22, 24 and 26 are rejected under 35 U.S.C. 103 as being unpatentable over Kim US Publication 2021/0193326 (hereafter “Kim”), Shamir et al. “Progression Analysis and Stage Discovery in Continuous Physiological Processes Using Image Computing” (hereafter “Shamir”) and Couture et al. US Publication 2019/0272917 (hereafter “Couture”).
Referring to claim 1, Kim discloses a method for osteoarthritis diagnosis, comprising:
receiving, by a grading module implemented in a computer system, a plurality of feature values generated based on a set of analysis logic from at least one input skeletal image of a joint (paragraph 44, Generating the joint model at step S220 may be configured to generate a 3D joint model for the joint area by performing a 3D-FEA-based simulation for the input data, to generate a joint cross-section image by performing 2D orthogonal projection on the 3D joint model, and to generate a joint model including the joint cross-section image); and
generating, by a first machine learning model implemented in the grading module (paragraph 51, Particularly, FIG. 3 illustrates a method for providing osteoarthritis prediction information by performing two types of analysis including FEA-based simulation and bone tissue pattern analysis and performing machine learning based on the analysis result), a quantitative Kellgren-Lawrence (KL) grade having an integer part based on the plurality of feature values (paragraph 49, Predicting the disease in the joint area at step S240 may further include determining a Kellgren-Lawrence (KL) grade based on the joint model and the bone tissue pattern and predicting a KL grade after the preset time period); and
wherein:
the first machine learning model is trained to generate the quantitative KL grade from the plurality of feature values; and
the quantitative KL grade represents a current osteoarthritis status of a patient (paragraph 49, Predicting the disease in the joint area at step S240 may further include determining a Kellgren-Lawrence (KL) grade based on the joint model and the bone tissue pattern and predicting a KL grade after the preset time period);
While Kim discloses that the quantitative KL grade has an integer part, Kim does not disclose expressly that the quantitative KL grade has a fractional part or that the features include at least one joint space narrowing (JSN) feature value and at least one osteophyte (OST) feature value.
Shamir discloses the quantitative KL grade has an integer part and a fractional part (page 3, For instance, an X-ray of an osteoarthritic knee that is between KL grade 2 and KL grade 3 can be assigned with the score of 2.6, rather than classified into KL grade 2 or KL grade 3);
wherein:
the plurality of feature values comprises at least one joint space narrowing (JSN) feature value and at least one osteophyte (OST) feature value, wherein the at least one joint space narrowing (JSN) feature value and at least one osteophyte (OST) feature value are characteristic values characterizing statuses of joint space narrowing and osteophyte, respectively (page 1, The KL classification scheme is a validated method for classifying individual joints into one of five grades, with 0 representing healthy joints, 1 representing doubtful OA, 2 representing mild OA, 3 moderate OA, and 4 being the most severe radiographic disease. This classification is based on features ofosteophytes (bony growths adjacent to the joint space), narrowing of part or all of the tibial-femoral joint space, and sclerosis of the subchondral bone, which reflect the progression of the disease);
wherein the first machine learning model is trained to generate the quantitative KL grade from the plurality of feature values (paragraph 48, Here, the machine-learning model may be a model trained using a supervised learning method using joint models and bone tissue pattern information acquired from multiple users); and
the quantitative KL grade represents a current osteoarthritis status of a patient (page 3, This assumption can be used to provide an automated classifier, which can automatically determine the KL grade for a given knee X-ray, and also estimate the KL grade for each test X-ray image with resolution that is higher than the individual KL grades).
Before the effective filing date of the claimed invention, it would have obvious to a person of ordinary skill in the art to provide an average score that has an integer and fractional part and to use joint space narrowing and osteophyte to classify osteoarthritis. The motivation for doing so would have been to provide a more accurate representation of the knee diagnosis and to accurately classify osteoarthritis in images by using widely known indicators of osteoarthritis.
While Kim discloses that the first machine learning model is trained, Kim does not disclose expressly training by a regression tree algorithm.
Couture discloses wherein the first machine learning model is trained by a regression tree algorithm (paragraph 38, The machine learning algorithm 412 may be selected from among many different potential supervised or unsupervised machine learning algorithms. Examples of supervised learning algorithms include Classification and Regression Tree (CART)).
Before the effective filing date of the claimed invention, it would have obvious to a person of ordinary skill in the art to use a regression tree algorithm. The motivation for doing so would have been to utilize a widely used manner of improving the results of a machine learning model. Therefore, it would have been obvious to combine Shamir and Couture with Kim to obtain the invention as specified in claim 1.
Referring to claim 2, Kim discloses before receiving the plurality of feature values by the grading module further comprising:
receiving, by the computer system, the at least one input skeletal image; and
generating, by the computer system, the plurality of feature values from the at least one input skeletal image (paragraph 40, input data including medical image data corresponding to an image of a joint area of a user is acquired at step S210).
Referring to claim 3, Kim discloses before generating the plurality of feature values further comprising:
extracting, by the computer system, at least one recognition area from the at least one input skeletal image; and
determining, by the computer system, a recognition result by the at least one recognition area;
wherein the recognition result determines the set of analysis logic (paragraph 57, Analysis of a bone tissue pattern in an image may be the process of finding a bone area in an entire medical image provided as input and performing image analysis on the tissue pattern of the bone area, thereby detecting the extent to which the bone area is changed. Finding a bone area in the image may be manually performed in such a way that a user specifies the bone area, or the bone area may be automatically found using an image-processing method (edge detection, contour detection, or the like) or machine learning).
Referring to claim 4, Kim discloses wherein the method is for knee osteoarthritis (KOA) diagnosis (paragraph 30, For example, based on the most commonly used Kellgren-Lawrence (KL) grade, the possibility that a person whose knee joints are currently determined to be KL 0 (normal) will be diagnosed as having knee joints graded as KL 2 in four years may be predicted as a percentage (%) value).
Referring to claims 6, 18 and 26, Sharmir discloses wherein the plurality of feature values further comprises at least one of:
one or more joint space width (JSW) feature values;
one or more joint space area (JSA) feature values;
one or more sclerosis (SCL) feature values;
one or more alignment feature values;
one or more attrition feature values; and
one or more cyst feature values (page 1, The KL classification scheme is a validated method for classifying individual joints into one of five grades, with 0 representing healthy joints, 1 representing doubtful OA, 2 representing mild OA, 3 moderate OA, and 4 being the most severe radiographic disease. This classification is based on features of osteophytes (bony growths adjacent to the joint space), narrowing of part or all of the tibial-femoral joint space, and sclerosis of the subchondral bone, which reflect the progression of the disease).
Referring to claims 8 and 20, Kim discloses wherein the first machine learning model is trained with multiple training data (paragraph 48, Here, the machine-learning model may be a model trained using a supervised learning method using joint models and bone tissue pattern information acquired from multiple users), each of which comprises:
a predetermined KL grade predetermined based on a training skeletal image as training ground truth (paragraph 58, When input data and correct output data corresponding thereto are regarded as Ground Truth (GT) and provided to input/output nodes, learning based on a learning algorithm is performed); and
a set of training feature values derived from the training skeletal image as training input (paragraph 58, Then, when new data is given as input, the inference result of the new data may be derived based on the learned content. Here, a hidden layer including at least one layer may be present between the input/output nodes).
Referring to claim 11, Kim discloses wherein the set of training feature values of each of the multiple training data is generated based on the training skeletal image (paragraph 58, When input data and correct output data corresponding thereto are regarded as Ground Truth (GT) and provided to input/output nodes, learning based on a learning algorithm is performed).
Referring to claim 12, Kim discloses wherein the first machine learning model is trained to output a predictive grade with an integer part (paragraph 30, For example, based on the most commonly used Kellgren-Lawrence (KL) grade, the possibility that a person whose knee joints are currently determined to be KL 0 (normal) will be diagnosed as having knee joints graded as KL 2 in four years may be predicted as a percentage (%) value).
Shamir discloses wherein the first machine learning model is trained to output a predictive grade with an integer part and a fractional part (page 3, For instance, an X-ray of an osteoarthritic knee that is between KL grade 2 and KL grade 3 can be assigned with the score of 2.6, rather than classified into KL grade 2 or KL grade 3).
Referring to claims 13 and 22, Kim discloses receiving a plurality of feature values, but does not disclose expressly generating a plurality of feature importance indicators.
Shamir discloses generating, by a feature importance estimation module implemented in the computer system, a plurality of feature importance indicators indicating contributions of a plurality of features to the quantitative KL grade;
wherein the plurality of features corresponds to the plurality of feature values (page 2, After all image features are computed, Fisher scores are computed individually for each image feature, and 93% of the features with the lowest Fisher scores are rejected. The rejection of the 93% of the features was determined experimentally, but as thoroughly discussed in [22, 23], changing this value has marginal effect on the performance. The classification can then be made by using the remaining 7% of the features with a Weighted Nearest Neighbor rule, such that the Fisher scores are used as weights).
Before the effective filing date of the claimed invention, it would have obvious to a person of ordinary skill in the art to generate a plurality of feature importance indicators. The motivation for doing so would have been to improve the diagnosis of osteoarthritis by analyzing the features that are most important in diagnosing osteoarthritis. Therefore, it would have been obvious to combine Shamir with Kim to obtain the invention as specified in claim 13.
Referring to claim 14, Shamir discloses wherein the feature importance estimation module is built based on the sets of training feature values of the multiple training data and the grading module (page 2, After all image features are computed, Fisher scores are computed individually for each image feature, and 93% of the features with the lowest Fisher scores are rejected. The rejection of the 93% of the features was determined experimentally, but as thoroughly discussed in [22, 23], changing this value has marginal effect on the performance. The classification can then be made by using the remaining 7% of the features with a Weighted Nearest Neighbor rule, such that the Fisher scores are used as weights).
Referring to claim 16, Kim discloses a non-transitory computer-readable medium having stored thereon a set of instructions that are executable by a processor of a computer system to carry out a method of generating a quantitative KL grade comprising:
receiving, by the computer system, at least one input skeletal image (paragraph 40, input data including medical image data corresponding to an image of a joint area of a user is acquired at step S210);
generating, by the computer system, a plurality of feature values based on the at least one input skeletal image (paragraph 44, Generating the joint model at step S220 may be configured to generate a 3D joint model for the joint area by performing a 3D-FEA-based simulation for the input data, to generate a joint cross-section image by performing 2D orthogonal projection on the 3D joint model, and to generate a joint model including the joint cross-section image); and
generating, by the computer system, the quantitative KL grade having an integer part from the plurality of feature values for osteoarthritis diagnosis (paragraph 47, Predicting the disease in the joint area at step S240 includes predicting a new joint model and a new bone tissue pattern after a preset time period based on the joint model and the bone tissue pattern information, and this may be a process of predicting the new joint model and the new bone tissue pattern after a preset time period based on a value output from a machine-learning model when the joint model and the bone tissue pattern information are input to the machine-learning model as input values thereof);
wherein the quantitative KL grade represents a current osteoarthritis status of a patient (paragraph 49, Predicting the disease in the joint area at step S240 may further include determining a Kellgren-Lawrence (KL) grade based on the joint model and the bone tissue pattern and predicting a KL grade after the preset time period), and is generated by a first machine learning model implemented in the non-transitory computer-readable medium (paragraph 51, Particularly, FIG. 3 illustrates a method for providing osteoarthritis prediction information by performing two types of analysis including FEA-based simulation and bone tissue pattern analysis and performing machine learning based on the analysis result);
wherein the first machine learning model is trained to generate the quantitative KL grade from the plurality of feature values (paragraph 48, Here, the machine-learning model may be a model trained using a supervised learning method using joint models and bone tissue pattern information acquired from multiple users).
While Kim discloses that the quantitative KL grade has an integer part, Kim does not disclose expressly that the quantitative KL grade has a fractional part or that the features include at least one joint space narrowing (JSN) feature value and at least one osteophyte (OST) feature value.
Shamir discloses the quantitative KL grade has an integer part and a fractional part (page 3, For
instance, an X-ray of an osteoarthritic knee that is between KL grade 2 and KL grade 3 can be assigned with the score of 2.6, rather than classified into KL grade 2 or KL grade 3);
wherein the quantitative KL grade represents a current osteoarthritis status of a patient (page 3, This assumption can be used to provide an automated classifier, which can automatically determine the KL grade for a given knee X-ray, and also estimate the KL grade for each test X-ray image with resolution that is higher than the individual KL grades);
wherein the plurality of feature values comprises at least one joint space narrowing (JSN) feature value and at least one osteophyte (OST) feature value, and wherein the at least one joint space narrowing (JSN) feature value and at least one osteophyte (OST) feature value are characteristic values characterizing statuses of joint space narrowing and osteophyte, respectively (page 1, The KL classification scheme is a validated method for classifying individual joints into one of five grades, with 0 representing healthy joints, 1 representing doubtful OA, 2 representing mild OA, 3 moderate OA, and 4 being the most severe radiographic disease. This classification is based on features ofosteophytes (bony growths adjacent to the joint space), narrowing of part or all of the tibial-femoral joint space, and sclerosis of the subchondral bone, which reflect the progression of the disease);
Before the effective filing date of the claimed invention, it would have obvious to a person of ordinary skill in the art to provide an average score that has an integer and fractional part and to use joint space narrowing and osteophyte to classify osteoarthritis. The motivation for doing so would have been to provide a more accurate representation of the knee diagnosis and to accurately classify osteoarthritis in images by using widely known indicators of osteoarthritis.
While Kim discloses that the first machine learning model is trained, Kim does not disclose expressly training by a regression tree algorithm.
Couture discloses wherein the first machine learning model is trained by a regression tree algorithm (paragraph 38, The machine learning algorithm 412 may be selected from among many different potential supervised or unsupervised machine learning algorithms. Examples of supervised learning algorithms include Classification and Regression Tree (CART)).
Before the effective filing date of the claimed invention, it would have obvious to a person of ordinary skill in the art to use a regression tree algorithm. The motivation for doing so would have been to utilize a widely used manner of improving the results of a machine learning model. Therefore, it would have been obvious to combine Shamir and Couture with Kim to obtain the invention as specified in claim 16.
Referring to claim 24, Kim discloses a method for training a first machine learning model to generate a quantitative Kellgren-Lawrence (KL) grade based on a plurality of feature values, comprising:
obtaining multiple training data, each of which comprises a set of training feature values derived from a training skeletal image as training input, and a predetermined KL grade predetermined based on the training skeletal image as training ground truth (paragraph 58, When input data and correct output data corresponding thereto are regarded as Ground Truth (GT) and provided to input/output nodes, learning based on a learning algorithm is performed); and
training the first machine learning model with the multiple training data (paragraph 48, Here, the machine-learning model may be a model trained using a supervised learning method using joint models and bone tissue pattern information acquired from multiple users);
wherein:
the predetermined KL grade represents a current osteoarthritis status of a patient when taking the training skeletal image (paragraph 49, Predicting the disease in the joint area at step S240 may further include determining a Kellgren-Lawrence (KL) grade based on the joint model and the bone tissue pattern and predicting a KL grade after the preset time period);
the predetermined KL grade comprises only an integer part (paragraph 58, When input data and correct output data corresponding thereto are regarded as Ground Truth (GT) and provided to input/output nodes, learning based on a learning algorithm is performed); and
the first machine learning model is trained to generate a quantitative KL grade comprising an integer part (paragraph 49, Predicting the disease in the joint area at step S240 may further include determining a Kellgren-Lawrence (KL) grade based on the joint model and the bone tissue pattern and predicting a KL grade after the preset time period).
While Kim discloses that the quantitative KL grade has an integer part, Kim does not disclose expressly that the quantitative KL grade has a fractional part or that the features include at least one joint space narrowing (JSN) feature value and at least one osteophyte (OST) feature value.
Shamir discloses wherein:
the predetermined KL grade represents a current osteoarthritis status of a patient when taking the training skeletal image (page 3, This assumption can be used to provide an automated classifier, which can automatically determine the KL grade for a given knee X-ray, and also estimate the KL grade for each test X-ray image with resolution that is higher than the individual KL grades).
the first machine learning model is trained to generate a quantitative KL grade comprising an integer part and a fractional part (page 3, For instance, an X-ray of an osteoarthritic knee that is between KL grade 2 and KL grade 3 can be assigned with the score of 2.6, rather than classified into KL grade 2 or KL grade 3); and
the plurality of feature values comprises at least one joint space narrowing (JSN) feature value and at least one osteophyte (OST) feature value, wherein the at least one joint space narrowing (JSN) feature value and at least one osteophyte (OST) feature value are characteristic values characterizing statuses of joint space narrowing and osteophyte, respectively (page 1, The KL classification scheme is a validated method for classifying individual joints into one of five grades, with 0 representing healthy joints, 1 representing doubtful OA, 2 representing mild OA, 3 moderate OA, and 4 being the most severe radiographic disease. This classification is based on features ofosteophytes (bony growths adjacent to the joint space), narrowing of part or all of the tibial-femoral joint space, and sclerosis of the subchondral bone, which reflect the progression of the disease).
Before the effective filing date of the claimed invention, it would have obvious to a person of ordinary skill in the art to provide an average score that has an integer and fractional part and to use joint space narrowing and osteophyte to classify osteoarthritis. The motivation for doing so would have been to provide a more accurate representation of the knee diagnosis and to accurately classify osteoarthritis in images by using widely known indicators of osteoarthritis.
While Kim discloses that the first machine learning model is trained, Kim does not disclose expressly training by a regression tree algorithm.
Couture discloses wherein:
the first machine learning model is a regression tree algorithm (paragraph 38, The machine learning algorithm 412 may be selected from among many different potential supervised or unsupervised machine learning algorithms. Examples of supervised learning algorithms include Classification and Regression Tree (CART)).
Before the effective filing date of the claimed invention, it would have obvious to a person of ordinary skill in the art to use a regression tree algorithm. The motivation for doing so would have been to utilize a widely used manner of improving the results of a machine learning model. Therefore, it would have been obvious to combine Shamir and Couture with Kim to obtain the invention as specified in claim 25.
Claims 9, 10, 15, 21, 22, 27 and 28 are rejected under 35 U.S.C. 103 as being unpatentable over Kim US Publication 2021/0193326, Shamir et al. “Progression Analysis and Stage Discovery in Continuous Physiological Processes Using Image Computing” and Couture et al. US Publication 2019/0272917 as applied to claims 7, 9, 13 and 19 above, and further in view of well known prior art.
Referring to claims 9, 21 and 27, Kim discloses the first machine learning model, but does not disclose expressly wherein the first machine learning model is trained by a boosted regression tree algorithm.
Official Notice is taken that it is well known and obvious in the art to train a model using a boosted regression tree algorithm (See MPEP 2144.03). The motivation for doing so would have been to utilize a widely used manner of improving the results of a machine learning model. Therefore, it would have been obvious to combine well known prior art with Kim to obtain the invention as specified in claims 9 and 21.
Referring to claim 10, Kim discloses the first machine learning model, but does not disclose expressly wherein the boosted regression tree algorithm is XGBRegressor algorithm.
Official Notice is taken that it is well known and obvious in the art to train a model using a boosted regression tree algorithm (See MPEP 2144.03). The motivation for doing so would have been to utilize a widely used manner of improving the results of a machine learning model. Therefore, it would have been obvious to combine well known prior art with Kim to obtain the invention as specified in claim 10.
Referring to claims 15 and 22, Shamir discloses the feature importance estimation module, but does not disclose expressly wherein the feature importance estimation module is a SHAP estimation model, and the plurality of feature importance indicators is a plurality of SHAP values.
Official Notice is taken that it is well known and obvious in the art to use a SHAP estimation model to determine a plurality of SHAP values (See MPEP 2144.03). The motivation for doing so would have been to utilize a widely used manner of improving the results of a machine learning model. Therefore, it would have been obvious to combine well known prior art with Kim and Shamir to obtain the invention as specified in claims 15 and 22.
Referring to claim 28, Kim discloses training the first machine learning model, but does not disclose expressly wherein the first machine learning model is trained by trying to minimize a mean square error (MSE).
Official Notice is taken that it is well known and obvious in the art to train a model by trying to minimize a mean square error (See MPEP 2144.03). The motivation for doing so would have been to improve the performance of the machine training model. Therefore, it would have been obvious to combine well known prior art with Kim and Shamir to obtain the invention as specified in claim 28.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to PETER K HUNTSINGER whose telephone number is (571)272-7435. The examiner can normally be reached Monday - Friday 8:30 - 5:00.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Benny Q Tieu can be reached at 571-272-7490. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/PETER K HUNTSINGER/Primary Examiner, Art Unit 2682
Prosecution Insights