NON-FINAL REJECTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This action is in response to the amendments filed on October 17th, 2022.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on November 20th, 2025 has been entered.
Response to Amendment
The Examiner thanks the applicant for the remarks, edits and arguments.
Regarding Claim Rejections – 35 U.S.C. 101
Applicant Remarks: Step 2A, Prong One
The applicant cites the MPEP 2106.04(a)(2)(III)(A) which states that a limitation cannot be an abstract idea if the limitation cannot practically be performed by a human. The applicant argues that the training a machine learning audit model is a complex task requiring processing of data and images. Further the applicant argues that that the training is occurring on special purpose computing devices. Therefore, the limitations recited in the independent claims cannot practically performed in a human mind and would not qualify as an abstract idea.
The applicant argues that that the claims do not recite mathematical relationships, formulas, and method of organizing human activity therefore the claims do not recite an abstract idea.
Step 2A, Prong Two
The applicant argues, even if limitations of the claims recite an abstract idea, the claims still integrate the practical application of an improvement to computing or field. The remarks state that, “The robustness of the machine learning based medical analysis network is improved by using a machine learning based audit network.”. According to the applicant this technical improvement is recited in the claims and specification.
Step 2B
Finally, the applicant states that that the claims are not Well understood, routine or conventional and point to the first claim as an example. According to the applicant, taking in these considerations, amended claims and the specification, the current application does not recite any abstract idea. Further, even if the claims recite an abstract idea, there is sufficient evidence of an improvement to technical field or technology and that the claim limitations are not well understood, routine or conventional. For these reasons, the applicant believes the 101 rejection should be withdrawn.
Examiner Response:
The applicant argues that the tasks recited are too complex and require special purpose computing devise to perform. According to the MPEP 2106.04(a)(2)(III) it states: “Accordingly, the "mental processes" abstract idea grouping is defined as concepts performed in the human mind, and examples of mental processes include observations, evaluations, judgments, and opinions.”. The examiner believes there are elements of abstract ideas in the independent claims. Such as using an algorithm/ML model to produce a result and to make a judgment after an evaluation. However, the amendment presented provides enough support for the process of using a trained machine learning model to perform designated actions. Because of this, the examiner no longer believes that the independent claims recite any abstract ideas. Therefore, the 101 rejection has been withdrawn.
Regarding Claim Rejections – 35 U.S.C. 103
Applicant Remarks:
The applicant argues that Fernandez fails to teach the limitations of the independent claims. According to the applicant Fernandez does not teach the use of a machine learning model as disclosed in the claims and specification. Further Fernandez fails to teach the training of a fault model as disclosed in the claims. Because of this the applicant believes Fernandez and Zhang fail to explicitly teach the amended claims. Finally, since Fernandez is unable to teach the independent claims, then the other art proposed fails to overcome the deficiencies of Fernandez. Because of this the applicant believes the current amended claims overcome the 103 rejection and ask for the rejection to be withdrawn.
Examiner Response:
The applicant has pointed out that the art Fernandez does not use a machine learning model as disclosed in the claims to perform robustness testing. The examiner has reviewed Fernandez and found elements of machine learning and automation but it does explicitly teach the use a machine learning model to perform an evaluation of another ML model. After the amendments made to the claims, Fernandez does fail to teach the limitations of the independent claims. After each amendment a full and complete search is performed. In doing this old art is reviewed and any new art is considered. After further search was conducted, new art was found by the examiner and is proposed. The art Fernandez no longer meets the obviousness requirements and does not explicitly teach the claimed limitations and has been discarded by the examiner. The new art found discloses the use of machine learning models to determine robustness and trustworthiness of other machine learning models. With this new art combined with the previously proposed art the examiner believes the claimed invention would have been obvious to an ordinary person of the art at the time of filing. Therefore, the rejection under 35 U.S.C. 103 is upheld, see 103 rejection below.
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, 2, 7, 9-11, 15, 18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang et al. (Zhang et al. "Deep Reinforcement Learning for Vessel Centerline Tracing in Multi-Modality 3D Volumes", 2018, hereinafter "Zhang") in view of Bie et al., (Bie et al., “To Trust or Not to Trust a Regressor: Estimating and Explaining Trustworthiness of Regression Predictions” Jul. 28th, 2021, hereinafter “Bie”).
Regarding claim 1, Zhang discloses, “A computer-implemented method comprising:” (Introduction, pp. 756; "In this paper, we address the vessel centerline tracing problem with an end-to-end trainable deep reinforcement learning (DRL) network. An artificial agent is learned to interact with surrounding environment and collect rewards from the interaction." This article teaches a method to determine the centerline of vessels using an artificial agent.)
“receiving input medical data;” (Method, pp. 757; "Given a 3D volumetric image I and the list of ground truth vessel centerline points
G
=
[
g
0
,
g
1
,
…
,
g
n
]
we aim to learn a navigation model for an agent to trace the centerline through an optimal trajectory
P
=
[
p
0
,
p
1
,
…
,
p
n
]
" This model used takes in 3D volumetric image data to perform a medical task.)
“receiving results of a medical analysis task performed based on the input medical data using a machine learning based medical analysis network;” (Method, pp. 759; "From the neural network we generate an action
a
0
which moves the current point to
p
1
. Then, the current state is updated as
s
1
=
I
p
1
and fed into the neural network to generate action again. We repeat this process until the path converges on oscillatory-like cycles. To further stabilize the tracing process, we also apply momentum on action-values from network output:
r
t
←
α
r
t
-
1
+
(
1
-
α
)
r
t
, where a is the momentum factor. The centerline tracing process stops if the agent moves out of the volume or if a cycle is formed, i.e., moving to a position already visited previously." This model will take in input data and trace a vascular centerline and return a result.)
Zhang fails to explicitly disclose, “determining a robustness of the machine learning based medical analysis network for performing the medical analysis task based on the input medical data and the results of the medical analysis task using a machine learning based audit network, the machine learning based audit network receiving as input the input medical data and the results of the medical analysis task and generating as output the robustness of the machine learning based medical analysis network, wherein the machine learning based audit network is trained using training medical images with labels of ground truth results of the medical analysis task; and” and “outputting the determination of the robustness of the machine learning based medical analysis network.”.
However, Bie discloses, “determining a robustness of the machine learning based medical analysis network for performing the medical analysis task based on the input medical data and the results of the medical analysis task using a machine learning based audit network, the machine learning based audit network receiving as input the input medical data and the results of the medical analysis task and generating as output the robustness of the machine learning based medical analysis network, wherein the machine learning based audit network is trained using training medical images with labels of ground truth results of the medical analysis task; and” (Method, pp. 3; “First, we calculate the RETRO score. RETRO requires (i) a trained regression model, (ii) the data that the model was trained on
(
X
t
r
a
i
n
,
Y
t
r
a
i
n
)
, and (iii) a new instance
x
p
and its predicted target value
y
^
p
. All input and output variables must be numeric. We treat the model as a black box and do not require access to the internals or parameters. The method consists of three phases, which are discussed below.” This model will evaluate the trustworthiness of a prediction machine learning model. The model which analyzes the prediction model uses k-NN methods to determine how trustworthy a prediction output and model is. As stated above, the model will use the training data used to train the prediction model, the new instance of the data representing the input to the prediction model, and the predicted target value representing the output of the prediction model. These are used to evaluate a RETRO score.)
“outputting the determination of the robustness of the machine learning based medical analysis network.” (VIZ: Visually Explaining Trustworthiness, pp. 4; “Besides identifying predictions that are potentially erroneous, we want to provide users with an actionable tool that helps them understand why a particular prediction is (un)trustworthy. We use Parallel Coordinate Plots, as proposed by Inselberg (1985) for displaying multi-dimensional data, to visualize the new instance and its K neighbors as retrieved through the RETRO-method. Example VIZ-plots are provided in Figures 1, 2 and 3.” This model will be able to show the user using an interface the trustworthiness of a prediction result. This method is able to chart the different RETRO scores and will be able to show the user different values and inferences.) It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Zhang and Bie. Zhang teaches a model that is able to recognize centerlines of vessels using a machine learning model. Bie teaches a machine learning model that is able to assess the trustworthiness and accuracy of another machine learning model. One of ordinary skill would have motivation to combine a machine learning model that is able to find centerlines of vessels and a machine learning model which is able to assess another machine learning model for trustworthiness and accuracy, “We propose RETRO-VIZ, a method for assessing the trustworthiness of regression predictions. In all 117 experimental settings, we found a negative correlation between the RETRO-score for a prediction and its correctness. This shows that RETRO can help to gauge the trustworthiness of a prediction in the absence of the ground-truth error. In addition, in our user study, we find that VIZ-plots help users distinguish whether and why algorithmic predictions are trustworthy.” (Bie, Discussion and Conclusion, pp. 7).
Regarding claim 2, Zhang fails to explicitly disclose this claim. However, Bie discloses, “in response to determining that the machine learning based medical analysis network is not robust, determining that the machine learning based medical analysis network is not robust due to the input medical data being out-of-distribution with respect to training data on which the machine learning based medical analysis network was trained or due to an artifact in at least one of the input medical data or the results of the medical analysis task.” (VIZ: Visually Explaining Trustworthiness, pp. 4; “VIZ-plots make it straightforward to identify why a prediction has received a high or low RETRO-score. A low RETRO-score can be obtained for a new instance for two main reasons. Firstly, it could be that the new instance has no instances in the reference set that lie relatively close to it, because the new instance deviates substantially in one or multiple features. An example of this is provided in Figure 1. Secondly, it could be that the predicted target variable lies far away from the ground-truth target variables of the K neighbors (see Figure 2 for an example). In contrast, when a new instance and its prediction are aligned with the reference data, they will receive a high RETRO-score, and the lines on the VIZ-plot will lie close to each other, like in Figure 3.” This model is able to show a user the trustworthiness of a prediction model based on its output. This model will be able to take in data and evaluate the prediction model and output to produce a score. This score is then presented to show the level of trustworthiness. This will show the end result of the prediction model and how accurate it is.)
Regarding claim 7, Zhang fails to explicitly disclose this claim. However, Bie discloses, “in response to determining that the machine learning based medical analysis network is not robust, generating an alert to a user notifying the user that the machine learning based medical analysis network is not robust or requesting input from the user.” (VIZ: Visually Explaining Trustworthiness, pp. 4; “Besides identifying predictions that are potentially erroneous, we want to provide users with an actionable tool that helps them understand why a particular prediction is (un)trustworthy. We use Parallel Coordinate Plots, as proposed by Inselberg (1985) for displaying multi-dimensional data, to visualize the new instance and its K neighbors as retrieved through the RETRO-method. Example VIZ-plots are provided in Figures 1, 2 and 3.” This model will display the prediction results to the user. As seen in figures 1, 2 and 3 the RETRO score is displayed at the top and by a highlighted line on the graph. The interface will also have the prediction score on a red and green sliding scale. If the prediction score is low, the line will be represented in red and if the score is high, it will be represented in green. An ordinary person in the art would be able to determine that red line would alert the user of a low score and the presence of a green line would signify the opposite.)
Regarding claim 9, Bie fails to explicitly disclose this claim. However, Zhang discloses, “wherein the medical analysis task comprises at least one of segmentation, determining centerlines of vessels, or computing a fractional flow reserve (FFR).” (Method, pp. 757; "In this section we propose a deep reinforcement learning based method for vessel centerline tracing in 3D volumes. Given a 3D volumetric image I and the list of ground truth vessel centerline points
G
=
[
g
0
,
g
1
,
…
,
g
n
]
, we aim to learn a navigation model for an agent to trace the centerline through an optimal trajectory
P
=
[
p
0
,
p
1
,
…
,
p
n
]
." This article teaches a ML model that is designed to determine the centerline of vessels.)
Regarding claim 10, Zhang discloses, “An apparatus comprising:” (Network Architecture and Implementation, pp. 760; "The experiments was conducted on a server with one Nvidia Titan X GPU." This model was executed on a computer system or server.)
“means for receiving input medical data;” (Method, pp. 757; "Given a 3D volumetric image I and the list of ground truth vessel centerline points
G
=
[
g
0
,
g
1
,
…
,
g
n
]
we aim to learn a navigation model for an agent to trace the centerline through an optimal trajectory
P
=
[
p
0
,
p
1
,
…
,
p
n
]
" This model used takes in 3D volumetric image data to perform a medical task.)
“means for receiving results of a medical analysis task performed based on the input medical data using a machine learning based medical analysis network;” (Method, pp. 759; "From the neural network we generate an action
a
0
which moves the current point to
p
1
. Then, the current state is updated as
s
1
=
I
p
1
and fed into the neural network to generate action again. We repeat this process until the path converges on oscillatory-like cycles. To further stabilize the tracing process, we also apply momentum on action-values from network output:
r
t
←
α
r
t
-
1
+
(
1
-
α
)
r
t
, where a is the momentum factor. The centerline tracing process stops if the agent moves out of the volume or if a cycle is formed, i.e., moving to a position already visited previously." This model will take in input data and trace a vascular centerline and return a result.)
Zhang fails to explicitly disclose, “means for determining a robustness of the machine learning based medical analysis network for performing the medical analysis task based on the input medical data and the results of the medical analysis task using a machine learning based audit network, the machine learning based audit network receiving as input the input medical data and the results of the medical analysis task and generating as output the robustness of the machine learning based medical analysis network, wherein the machine learning based audit network is trained using training medical images with labels of ground truth results of the medical analysis task; and” and “means for outputting the determination of the robustness of the machine learning based medical analysis network.”.
However, Bie discloses, “means for determining a robustness of the machine learning based medical analysis network for performing the medical analysis task based on the input medical data and the results of the medical analysis task using a machine learning based audit network, the machine learning based audit network receiving as input the input medical data and the results of the medical analysis task and generating as output the robustness of the machine learning based medical analysis network, wherein the machine learning based audit network is trained using training medical images with labels of ground truth results of the medical analysis task; and” (Method, pp. 3; “First, we calculate the RETRO score. RETRO requires (i) a trained regression model, (ii) the data that the model was trained on
(
X
t
r
a
i
n
,
Y
t
r
a
i
n
)
, and (iii) a new instance
x
p
and its predicted target value
y
^
p
. All input and output variables must be numeric. We treat the model as a black box and do not require access to the internals or parameters. The method consists of three phases, which are discussed below.” This model will evaluate the trustworthiness of a prediction machine learning model. The model which analyzes the prediction model uses k-NN methods to determine how trustworthy a prediction output and model is. As stated above, the model will use the training data used to train the prediction model, the new instance of the data representing the input to the prediction model, and the predicted target value representing the output of the prediction model. These are used to evaluate a RETRO score.)
“means for outputting the determination of the robustness of the machine learning based medical analysis network.” (VIZ: Visually Explaining Trustworthiness, pp. 4; “Besides identifying predictions that are potentially erroneous, we want to provide users with an actionable tool that helps them understand why a particular prediction is (un)trustworthy. We use Parallel Coordinate Plots, as proposed by Inselberg (1985) for displaying multi-dimensional data, to visualize the new instance and its K neighbors as retrieved through the RETRO-method. Example VIZ-plots are provided in Figures 1, 2 and 3.” This model will be able to show the user using an interface the trustworthiness of a prediction result. This method is able to chart the different RETRO scores and will be able to show the user different values and inferences.)
Regarding claim 11, Zhang fails to explicitly disclose this claim. However, Bie discloses, “means for determining that the machine learning based medical analysis network is not robust due to the input medical data being out-of-distribution with respect to training data on which the machine learning based medical analysis network was trained or due to an artifact in at least one of the input medical data or the results of the medical analysis task in response to determining that the machine learning based medical analysis network is not robust.” (VIZ: Visually Explaining Trustworthiness, pp. 4; “VIZ-plots make it straightforward to identify why a prediction has received a high or low RETRO-score. A low RETRO-score can be obtained for a new instance for two main reasons. Firstly, it could be that the new instance has no instances in the reference set that lie relatively close to it, because the new instance deviates substantially in one or multiple features. An example of this is provided in Figure 1. Secondly, it could be that the predicted target variable lies far away from the ground-truth target variables of the K neighbors (see Figure 2 for an example). In contrast, when a new instance and its prediction are aligned with the reference data, they will receive a high RETRO-score, and the lines on the VIZ-plot will lie close to each other, like in Figure 3.” This model is able to show a user the trustworthiness of a prediction model based on its output. This model will be able to take in data and evaluate the prediction model and output to produce a score. This score is then presented to show the level of trustworthiness. This will show the end result of the prediction model and how accurate it is.)
Regarding claim 15, Zhang discloses, “A non-transitory computer readable medium storing computer program instructions, the computer program instructions when executed by a processor cause the processor to perform operations comprising:” (Network Architecture and Implementation, pp. 760; "The experiments was conducted on a server with one Nvidia Titan X GPU." This model was executed on a computer system or server which contain memory with stored computer instructions to execute the functions.)
“receiving input medical data;” (Method, pp. 757; "Given a 3D volumetric image I and the list of ground truth vessel centerline points
G
=
[
g
0
,
g
1
,
…
,
g
n
]
we aim to learn a navigation model for an agent to trace the centerline through an optimal trajectory
P
=
[
p
0
,
p
1
,
…
,
p
n
]
" This model used takes in 3D volumetric image data to perform a medical task.)
“receiving results of a medical analysis task performed based on the input medical data using a machine learning based medical analysis network;” (Method, pp. 759; "From the neural network we generate an action
a
0
which moves the current point to
p
1
. Then, the current state is updated as
s
1
=
I
p
1
and fed into the neural network to generate action again. We repeat this process until the path converges on oscillatory-like cycles. To further stabilize the tracing process, we also apply momentum on action-values from network output:
r
t
←
α
r
t
-
1
+
(
1
-
α
)
r
t
, where a is the momentum factor. The centerline tracing process stops if the agent moves out of the volume or if a cycle is formed, i.e., moving to a position already visited previously." This model will take in input data and trace a vascular centerline and return a result.)
Zhang fails to explicitly disclose, “determining a robustness of the machine learning based medical analysis network for performing the medical analysis task based on the input medical data and the results of the medical analysis task using a machine learning based audit network, the machine learning based audit network receiving as input the input medical data and the results of the medical analysis task and generating as output the robustness of the machine learning based medical analysis network, wherein the machine learning based audit network is trained using training medical images with labels of ground truth results of the medical analysis task; and” and “outputting the determination of the robustness of the machine learning based medical analysis network.”.
However, Bie discloses, “determining a robustness of the machine learning based medical analysis network for performing the medical analysis task based on the input medical data and the results of the medical analysis task using a machine learning based audit network, the machine learning based audit network receiving as input the input medical data and the results of the medical analysis task and generating as output the robustness of the machine learning based medical analysis network, wherein the machine learning based audit network is trained using training medical images with labels of ground truth results of the medical analysis task; and” (Method, pp. 3; “First, we calculate the RETRO score. RETRO requires (i) a trained regression model, (ii) the data that the model was trained on
(
X
t
r
a
i
n
,
Y
t
r
a
i
n
)
, and (iii) a new instance
x
p
and its predicted target value
y
^
p
. All input and output variables must be numeric. We treat the model as a black box and do not require access to the internals or parameters. The method consists of three phases, which are discussed below.” This model will evaluate the trustworthiness of a prediction machine learning model. The model which analyzes the prediction model uses k-NN methods to determine how trustworthy a prediction output and model is. As stated above, the model will use the training data used to train the prediction model, the new instance of the data representing the input to the prediction model, and the predicted target value representing the output of the prediction model. These are used to evaluate a RETRO score.)
“outputting the determination of the robustness of the machine learning based medical analysis network.” (VIZ: Visually Explaining Trustworthiness, pp. 4; “Besides identifying predictions that are potentially erroneous, we want to provide users with an actionable tool that helps them understand why a particular prediction is (un)trustworthy. We use Parallel Coordinate Plots, as proposed by Inselberg (1985) for displaying multi-dimensional data, to visualize the new instance and its K neighbors as retrieved through the RETRO-method. Example VIZ-plots are provided in Figures 1, 2 and 3.” This model will be able to show the user using an interface the trustworthiness of a prediction result. This method is able to chart the different RETRO scores and will be able to show the user different values and inferences.)
Regarding claim 18, Zhang fails to explicitly disclose this claim. However, Bie discloses, “in response to determining that the machine learning based medical analysis network is not robust, generating an alert to a user notifying the user that the machine learning based medical analysis network is not robust or requesting input from the user.” (VIZ: Visually Explaining Trustworthiness, pp. 4; “Besides identifying predictions that are potentially erroneous, we want to provide users with an actionable tool that helps them understand why a particular prediction is (un)trustworthy. We use Parallel Coordinate Plots, as proposed by Inselberg (1985) for displaying multi-dimensional data, to visualize the new instance and its K neighbors as retrieved through the RETRO-method. Example VIZ-plots are provided in Figures 1, 2 and 3.” This model will display the prediction results to the user. As seen in figures 1, 2 and 3 the RETRO score is displayed at the top and by a highlighted line on the graph. The interface will also have the prediction score on a red and green sliding scale. If the prediction score is low, the line will be represented in red and if the score is high, it will be represented in green. An ordinary person in the art would be able to determine that red line would alert the user of a low score and the presence of a green line would signify the opposite.)
Regarding claim 20, Bie fails to explicitly disclose this claim. However, Zhang discloses, “wherein the medical analysis task comprises at least one of segmentation, determining centerlines of vessels, or computing a fractional flow reserve (FFR).” (Method, pp. 757; "In this section we propose a deep reinforcement learning based method for vessel centerline tracing in 3D volumes. Given a 3D volumetric image I and the list of ground truth vessel centerline points
G
=
[
g
0
,
g
1
,
…
,
g
n
]
, we aim to learn a navigation model for an agent to trace the centerline through an optimal trajectory
P
=
[
p
0
,
p
1
,
…
,
p
n
]
." This article teaches a ML model that is designed to determine the centerline of vessels.)
Claims 3 and 12 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang and Bie in view of Daughton et al, (Daughton et al., “PREDICTION OF PROBABILITY DISTRIBUTION FUNCTION OF CLASSIFIERS”, US 11,901,076 B1, Filed Jun. 11th, 2021, hereinafter “Daughton”).
Regarding claim 3, Zhang and Bie fail to explicitly disclose this claim. However, Daughton discloses, “in response to determining that the machine learning based medical analysis network is not robust, retraining the machine learning based medical analysis network and the machine learning based audit network based on the input medical data.” (Detailed Description, pp. 22, Col. 11, Ln. 55-61; “In step 230, program 115 determines if the uncertainty level is above a threshold associated with an acceptable level of uncertainty. Accordingly, if the threshold is exceeded, then program 115 proceeds to retrain the models in step 230. Stated another way, there may be models generated that do not have a high enough confidence level for program 115 to make accurate evaluations of medical information 130.” The method in this article is able to evaluate many different kinds of models which can perform different medical related predictions and tasks. This model also is able to determine when models need to be retrained. The citation above states that it will evaluate models and after a threshold is met the models will be retrained. This is similar to the limitation because it solves the same problem, after a particular threshold is met, the models, which includes different kinds of models, will be retrained.)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Zhang, Bie and Daughton. Zhang teaches a model that is able to recognize centerlines of vessels using a machine learning model. Bie teaches a machine learning model that is able to assess the trustworthiness and accuracy of another machine learning model. Daughton teaches a method that is able to determine the accuracy of a machine learning model and determine when and how retraining would occur. One of ordinary skill would have motivation to combine a machine learning model that is able to find centerlines of vessels and a machine learning model which is able to assess another machine learning model for trustworthiness and accuracy with a model that is able to determine when and how to retain prediction models after certain criteria is met, “During operation, the computer system may receive information (step 1010) corresponding to medical imaging and clinical data for a plurality of individuals. Then, the computer system may apply a pretrained predictive model (step 1020) to the information for at least a subset of the plurality of individuals. Moreover, the computer system may determine levels of uncertainty (step 1020) in results of the pretrained predictive model for at least the subset of the plurality of individuals. Next, the computer system may dynamically adapt (step 1030) a lower acceptable limit and an upper acceptable limit that define at least one threshold range based at least in part on the determined levels of uncertainty and a predefined target performance of the pretrained predictive model for the plurality of individuals.” (Example 4, Daughton, Col. 17, pp. 25)
Regarding claim 12, Zhang and Bie fail to explicitly disclose this claim. However, Daughton discloses, “means for retraining the machine learning based medical analysis network and the machine learning based audit network based on the input medical data in response to determining that the machine learning based medical analysis network is not robust.” (Detailed Description, pp. 22, Col. 11, Ln. 55-61; “In step 230, program 115 determines if the uncertainty level is above a threshold associated with an acceptable level of uncertainty. Accordingly, if the threshold is exceeded, then program 115 proceeds to retrain the models in step 230. Stated another way, there may be models generated that do not have a high enough confidence level for program 115 to make accurate evaluations of medical information 130.” The method in this article is able to evaluate many different kinds of models which can perform different medical related predictions and tasks. This model also is able to determine when models need to be retrained. The citation above states that it will evaluate models and after a threshold is met the models will be retrained. This is similar to the limitation because it solves the same problem, after a particular threshold is met, the models, which includes different kinds of models, will be retrained.)
Claims 4 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang and Bie in view of Kaditz et al., (Kaditz et al., “ITERATNE MEDICAL TESTING OF BIOLOGICAL SAMPLES”, US 11,650,195 B2, Filed, Sep. 26th, 2018, hereinafter “Kaditz”).
Regarding claim 4, Zhang and Bie fail to explicitly disclose this claim. However, Kaditz discloses, “in response to determining that the machine learning based medical analysis network is not robust, presenting one or more alternate results of the medical analysis task from other machine learning based medical analysis networks.” (Summary, pp. 13, Col. 1- 2, Ln. 65- 67, 1- 9; "Next, the system determines one or more additional medical tests to perform that can improve an accuracy of the first test result. Furthermore, the system accesses the biological sample to obtain at least a second portion of the biological sample, and performs the one or more additional medical tests on at least the second portion of the biological sample to obtain one or more additional test results. Additionally, the system computes a second test result based at least in part on the first test result and the one or more additional test results, where the second test result has an improved accuracy relative to the first test result." This system will determine whether another test is necessary and perform it. It will the present this second, more accurate result to the user. This second test is more accurate than the first test.)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Zhang, Bie, and Kaditz. Zhang teaches a model that is able to recognize centerlines of vessels using a machine learning model. Bie teaches a machine learning model that is able to assess the trustworthiness and accuracy of another machine learning model. Kaditz teaches improved testing of medical analysis by automating medical testing of samples. One of ordinary skill would have motivation to combine a machine learning model that is able to find centerlines of vessels and a machine learning model which is able to assess another machine learning model for trustworthiness and accuracy and improved medical testing practices, "By iteratively performing the medical testing of a previously acquired and stored biological sample, this testing technique may allow a diagnosis and/or a treatment for the individual to be rapidly and accurately determined." (Kaditz, Detailed Description, pp. 14, Col. 4, In. 48- 51).
Regarding claim 13, Zhang and Bie fail to explicitly disclose this claim. However, Kaditz discloses, “means for presenting one or more alternate results of the medical analysis task from other machine learning based medical analysis networks in response to determining that the machine learning based medical analysis network is not robust.” Kaditz (Summary, pp. 13, Col. 1- 2, Ln. 65- 67, 1- 9; "Next, the system determines one or more additional medical tests to perform that can improve an accuracy of the first test result. Furthermore, the system accesses the biological sample to obtain at least a second portion of the biological sample, and performs the one or more additional medical tests on at least the second portion of the biological sample to obtain one or more additional test results. Additionally, the system computes a second test result based at least in part on the first test result and the one or more additional test results, where the second test result has an improved accuracy relative to the first test result." This system will determine whether another test is necessary and perform it. It will the present this second, more accurate result to the user. This second test is more accurate than the first test.)
Claims 5, 8, 14, 17, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang and Bie in view of Mitchell et al., (Mitchell et al., “Experience With a Learning Personal Assistant”, 1994, hereinafter “Mitchell”).
Regarding claim 5, Bie discloses, “wherein determining a robustness of the machine learning based medical analysis network for performing the medical analysis task based on the input medical data and the results of the medical analysis task using a machine learning based audit network comprises: determining the robustness of the machine learning based medical analysis network based on the final results of the medical analysis tasks.” (Method, pp. 3; “First, we calculate the RETRO score. RETRO requires (i) a trained regression model, (ii) the data that the model was trained on
(
X
t
r
a
i
n
,
Y
t
r
a
i
n
)
, and (iii) a new instance
x
p
and its predicted target value
y
^
p
. All input and output variables must be numeric. We treat the model as a black box and do not require access to the internals or parameters. The method consists of three phases, which are discussed below.” This model will determine the trustworthiness of a prediction model. This will take in the stated inputs and evaluate the model and output from the model. This will generate a score and will return the result to the user in a graphical interface. This model will evaluate the prediction model after it produces a, including and not limited to, a final output.)
Zhang and Bie fail to explicitly disclose, “receiving user input editing the results of the medical analysis task to generate final results of the medical analysis task and”.
However, Mitchell discloses, “receiving user input editing the results of the medical analysis task to generate final results of the medical analysis task and” (System Organization, pp. 3; "This suggestion is derived from a previously learned rule that matches the known features of this new meeting (i.e., those features for which the user has already been prompted, plus any features inferred from these). The user may accept this advice or override it by entering the desired value. In this figure, the user is overriding the advice, and instructing the system to allocate 30 minutes for this meeting. Whenever the user accepts or overrides CAP's advice, a training example is captured that is used for subsequent learning." The user can override the results of the output and enter their own final result. This override is taken in by the system and is used for subsequent training.)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Zhang, Bie and Mitchell. Zhang teaches a model that is able to recognize centerlines of vessels using a machine learning model. Bie teaches a machine learning model that is able to assess the trustworthiness and accuracy of another machine learning model. Mitchell teaches a machine learning model whose results can be altered by the user to better fit their needs. One of ordinary skill would have motivation to combine a machine learning model that is able to find centerlines of vessels and a machine learning model which is able to assess another machine learning model for trustworthiness and accuracy and a machine learning model that is able to respond to a user and tailor their needs to the model, "While rules learned by CAP are useful for providing interactive advice to be approved or overridden by the user, they are not sufficiently accurate to support autonomous negotiation of all meetings by the agent on the user's behalf." (Mitchell, Conclusion and Prospects, pp. 11).
Regarding claim 8, Zhang and Bie fail to explicitly disclose this claim. However, Mitchell discloses, “receiving the input from the user overriding the determination that the machine learning based medical analysis network is not robust or editing the results of the medical analysis task.” (System Organization, pp. 3; "This suggestion is derived from a previously learned rule that matches the known features of this new meeting (i.e., those features for which the user has already been prompted, plus any features inferred from these). The user may accept this advice or override it by entering the desired value. In this figure, the user is overriding the advice, and instructing the system to allocate 30 minutes for this meeting. Whenever the user accepts or overrides CA P's advice, a training example is captured that is used for subsequent learning." This system will allow the user to override a recommendation. This recommendation is a result of a ML model designed to be a personal assistant. The user will be allowed to edit the results.)
Regarding claim 14, Bie discloses, “wherein the means for determining a robustness of the machine learning based medical analysis network for performing the medical analysis task based on the input medical data and the results of the medical analysis task using a machine learning based audit network comprises: means for determining the robustness of the machine learning based medical analysis network based on the final results of the medical analysis tasks.” (Method, pp. 3; “First, we calculate the RETRO score. RETRO requires (i) a trained regression model, (ii) the data that the model was trained on
(
X
t
r
a
i
n
,
Y
t
r
a
i
n
)
, and (iii) a new instance
x
p
and its predicted target value
y
^
p
. All input and output variables must be numeric. We treat the model as a black box and do not require access to the internals or parameters. The method consists of three phases, which are discussed below.” This model will determine the trustworthiness of a prediction model. This will take in the stated inputs and evaluate the model and output from the model. This will generate a score and will return the result to the user in a graphical interface. This model will evaluate the prediction model after it produces a, including and not limited to, a final output.)
Zhang and Bie fail to explicitly disclose, “means for receiving user input editing the results of the medical analysis task to generate final results of the medical analysis task and”.
However, Mitchell discloses, “means for receiving user input editing the results of the medical analysis task to generate final results of the medical analysis task and” (System Organization, pp. 3; "This suggestion is derived from a previously learned rule that matches the known features of this new meeting (i.e., those features for which the user has already been prompted, plus any features inferred from these). The user may accept this advice or override it by entering the desired value. In this figure, the user is overriding the advice, and instructing the system to allocate 30 minutes for this meeting. Whenever the user accepts or overrides CAP's advice, a training example is captured that is used for subsequent learning." The user can override the results of the output and enter their own final result. This override is taken in by the system and is used for subsequent training.)
Regarding claim 17, Zhang and Bie fail to explicitly disclose this claim. However, Mitchell discloses, “the operations further comprising receiving user input editing the results of the medical analysis task to generate final results of the medical analysis task and” (System Organization, pp. 3; "This suggestion is derived from a previously learned rule that matches the known features of this new meeting (i.e., those features for which the user has already been prompted, plus any features inferred from these). The user may accept this advice or override it by entering the desired value. In this figure, the user is overriding the advice, and instructing the system to allocate 30 minutes for this meeting. Whenever the user accepts or overrides CAP's advice, a training example is captured that is used for subsequent learning." The user can override the results of the output and enter their own final result. This override is taken in by the system and is used for subsequent training.)
Regarding claim 19, Zhang and Bie fail to explicitly disclose this claim. However, Mitchell discloses, “receiving the input from the user overriding the determination that the machine learning based medical analysis network is not robust or editing the results of the medical analysis task.” (System Organization, pp. 3; "This suggestion is derived from a previously learned rule that matches the known features of this new meeting (i.e., those features for which the user has already been prompted, plus any features inferred from these). The user may accept this advice or override it by entering the desired value. In this figure, the user is overriding the advice, and instructing the system to allocate 30 minutes for this meeting. Whenever the user accepts or overrides CA P's advice, a training example is captured that is used for subsequent learning." This system will allow the user to override a recommendation. This recommendation is a result of a ML model designed to be a personal assistant. The user will be allowed to edit the results.)
Claims 6 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang and Bie in view of Valiuddin et al. (Valiuddin et al., "Out-Of-Distribution Detection of Melanoma using Normalizing Flows", 2021, hereinafter "Valiuddin").
Regarding claim 6, Zhang and Bie fail to explicitly disclose this claim. However, Valiuddin discloses, “wherein the machine learning based audit network is implemented using a normalizing flows model.” (Abstract, pp. 1; "While the generative abilities of NFs are typically explored, we focus on exploring the data distribution modelling for Out-of-Distribution (OOD) detection. Using one of the state-of-the-art NF models, GLOW, we attempt to detect OOD examples in the ISIC dataset." The model used in this article utilizes normalizing flow architectures to detect OOD in datasets.)
It would have obvious to one of ordinary skill in the art before the effective filing date of the present application to combine Zhang, Bie and Valiuddin. Zhang teaches a model that is able to recognize centerlines of vessels using a machine learning model. Bie teaches a machine learning model that is able to assess the trustworthiness and accuracy of another machine learning model. Valiuddin teaches a machine learning model that uses a normalizing flow model to ensure correctness of data used in machine learning models. One of ordinary skill would have motivation to combine a machine learning model that is able to find centerlines of vessels and a machine learning model which is able to assess another machine learning model for trustworthiness and accuracy with another machine learning model which uses normalizing flows to ensure input data correctness, "We further implemented Wavelet Flow which was originally introduced due to the computational efficiency and competitiveness with GLOW in bits per dimension. We have observed a big improvement in BPD evaluation and find similar OOD detection performance." (Valiuddin, Conclusion, pp. 9).
Regarding claim 16, Zhang and Bie fail to explicitly disclose this claim. However, Valiuddin discloses, “wherein the machine learning based audit network is implemented using a normalizing flows model.” (Abstract, pp. 1; "While the generative abilities of NFs are typically explored, we focus on exploring the data distribution modelling for Out-of-Distribution (OOD) detection. Using one of the state-of-the-art NF modes, GLOW, we attempt to detect OOD examples in the ISIC dataset." The model used in this article utilizes normalizing flow architectures to detect OOD in datasets.)
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to PAUL MICHAEL GALVIN-SIEBENALER whose telephone number is (571)272-1257. The examiner can normally be reached Monday - Friday 8AM to 5PM.
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, Viker Lamardo can be reached at (571) 270-5871. 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.
/PAUL M GALVIN-SIEBENALER/Examiner, Art Unit 2147
/ERIC NILSSON/Primary Examiner, Art Unit 2151