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 .
Priority
Acknowledgment is made of applicant’s claim for foreign priority under 35 U.S.C. 119 (a)-(d).
Information Disclosure Statement
The information disclosure statements (IDS) submitted on 3-1-2022, 11-5-2025 (1), and 11-5-2025 (2) are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statements are being considered by the examiner.
Drawings
The drawings filed on 12-27-2021 are objected to because FIGS. 3, 4, and 5 contains text that is illegible [see 37 CFR 1.84(p)(1)]. Additionally, FIGS. 4, 5, and 6 contain text that are not oriented in the same direction as the view [see 37 CFR 1.84(p)(1)].
The drawings filed on 3-11-2022 are objected to because FIGS. 4A, 4B, 4C, 4D, 4E, 4F, 4G, 4H, 4I, 5A, 5B, 5C, 5D, 5E, 5F, 5G, 5H, and 5I contain text that are not oriented in the same direction as the view [see 37 CFR 1.84(p)(1)].
Corrected drawing sheets in compliance with 37 CFR 1.121(d) are required in reply to the Office action to avoid abandonment of the application. Any amended replacement drawing sheet should include all of the figures appearing on the immediate prior version of the sheet, even if only one figure is being amended. The figure or figure number of an amended drawing should not be labeled as “amended.” If a drawing figure is to be canceled, the appropriate figure must be removed from the replacement sheet, and where necessary, the remaining figures must be renumbered and appropriate changes made to the brief description of the several views of the drawings for consistency. Additional replacement sheets may be necessary to show the renumbering of the remaining figures. Each drawing sheet submitted after the filing date of an application must be labeled in the top margin as either “Replacement Sheet” or “New Sheet” pursuant to 37 CFR 1.121(d). If the changes are not accepted by the examiner, the applicant will be notified and informed of any required corrective action in the next Office action. The objection to the drawings will not be held in abeyance.
Specification
The disclosure is objected to because it contains an embedded hyperlink and/or other form of browser-executable code. Applicant is required to delete the embedded hyperlink and/or other form of browser-executable code; references to websites should be limited to the top-level domain name without any prefix such as http:// or other browser-executable code. See MPEP § 608.01. The following links are noted on: pg. 5, line 12, “http://urn”, and pg. 42, lines 4-5, “https://www.cbioportal.org/study/clinicalData?id=brca_tcga_pan_can_atlas_2018”.
Claim Objections
Claims 16-19 are objected to because of the following informalities:
In claim 16, “cause the processor to be configured to” should read “cause the processor to be configured to:”, “instructions for of a function” should read “instructions for a function”, and “causes the processor to be configured to” should read “causes the processor to be configured to:”.
Claims 17-19 are objected to for inheriting the deficiencies of claim 16.
Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 11-19 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Regarding Claims 11-15:
Claim 11 recites numerous limitations in the body of the claim identical to limitations found in the preamble of the claim. This makes it unclear if the limitations recited in the body are referring to a separate subject from the limitations found in the preamble or if both limitations are referring to the same subject. The following italicized limitations appear to be repeated:
(1a) “applying a function machine-learnt by machine-learning the function that is configured… to output a distribution of transition-specific probabilities for each interval of a set of intervals”
(1b) “applying the function to the input covariates, thereby outputting…a distribution of transition-specific probabilities for each interval of a set of time intervals”
Regarding (1a) and (1b), it is unclear if there are two separate distribution of transition-specific probabilities and two sets of intervals, or if the distribution of transition-probabilities and set of time intervals from (1b) is referring to the distribution of transition-probabilities and the set of intervals from (1a).
(2a) “based on input covariates representing medical characteristics of a patient”
(2b) “obtaining input covariates representing medical characteristics of a patient”
Regarding (2a) and (2b), it is unclear if there are two separate input covariates for a patient, or if the input covariates of a patient from (1b) is referring to the input covariates of a patient from (1a).
(3a) “with respect to a multi-state model of an illness having states and transitions between the states”
(3b) “with respect to a multi-state model of an illness having states and transitions between states”
Regarding (3a) and (3b), it is unclear if there are two separate multi-state models, or if the multi-state model from (1b) is referring to the multi-state model from (1a).
(4a) “the set of intervals forming a subdivision of a follow-up period”
(4b) “the set of time intervals forming a subdivision of a follow-up period”
Regarding (4a) and (4b), it is unclear if there are two separate sets of intervals and two separate follow-up periods, or if the set of time intervals and the follow-up period from (1b) is referring to the set of intervals and the follow-up period from (1a).
(5a) “obtaining an input dataset of covariates and time-to-event data of a set of patients”
(5b) “obtaining input covariates representing medical characteristics of a patient”
Regarding (5a) and (5b), it is unclear if there are two separate sets of covariate data, where (5a) obtains covariates for a set of patients and (5b) obtains a single covariate set for a single patient that is not a part of the set of patients, or if the single patient from (5b) is a subset of the set of patients. Examiner is interpreting (5b) as a subset of (5a).
Examiner is interpreting each b claim limitation to be referring to their respective a claim limitation.
Claims 12-15 are rejected to for inheriting the deficiencies of claim 11.
Regarding Claims 16-19:
Claim 16 is rejected to for at least the same reasons as discussed above with respect to claim 11. Additionally, claim 16 recites:
(1a) “instructions for machine-learning a function”
(1b) “instructions for a function machine-learnt according to the machine-learning that when executed by the processor causes the processor to be configured to”
(1c) “a function machine-learnt according to the machine-learning”
Regarding (1a) and (1b), it is unclear if there are two separate instructions, or if the instructions from (1b) are referring to the instructions from (1a). In view of (1b) reciting “the machine-learning” and “the processor”, the Examiner is interpreting the limitation from (1b) to be referring to (1a).
Further regarding (1a), (1b), and (1c), it is unclear if there are three separate functions of it the functions from (1b) and (1c) are referring to the function from (1a). In view of (1b) and (1c) reciting “the machine-learning”, the Examiner is interpreting the limitations from (1b) and (1c) to be referring to (1a).
Claims 17-19 are rejected to for inheriting the deficiencies of claim 16.
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1, 8-12, 16, and 20 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Groha et al. ("A General Framework for Survival Analysis and Multi-State Modelling"), hereinafter Groha.
Regarding Claim 1:
Groha discloses:
A computer-implemented method for machine-learning a function that is configured…to output a distribution of transition-specific probabilities for each interval of a set of intervals
Groha, pg. 2, col. 2, “The aim of multi-state models is a more granular analysis of time-to-event phenomena…Such models and their state-space can be described by a directed graph as shown in Figure 1(b).”
Pg. 4, col. 2, “With this model, we also have direct access to the hazard rate (the instantaneous risk for a given transition) over time.”
On pg. 2, Groha discloses a multi-state model [machine-learning a function that is configured to…output a distribution of transition-specific probabilities…]. Pg. 4 further specifies that the model works with data over time [for each interval of a set of intervals]. Lastly, as Groha’s models are implemented on a computer, it is construed as disclosing a computer-implemented method.
…based on input covariates representing medical characteristics of a patient with respect to a multi-state model of an illness having states and transitions between the states…
PNG
media_image1.png
182
333
media_image1.png
Greyscale
PNG
media_image2.png
184
327
media_image2.png
Greyscale
Groha, pg. 3, col. 1, “Figure 1. Example graphs corresponding to the 2-state competing risks model (a) and the illness-death model (b), a popular multi-state model. Competing risks models are a special case of multi-state models that only have one non-absorbing state, whereas in general multi-state models can have arbitrary, even cyclical connections.”
Pg. 4, col. 2, “The initial conditions are encoded by the covariates of the patient
m
(
0
)
=
f
(
x
)
, where
f
is given by a neural net.”
Pg. 13, C. Algorithm 1
PNG
media_image3.png
503
812
media_image3.png
Greyscale
On pg. 3 and in view of FIG. 1, Groha discloses 1(b), a general multi-state model [a multi-state model of an illness having states and transitions between the states]. On pg. 4, Groha discloses patient covariates [covariates representing medical characteristics of a patient], and further on pg. 13, Algorithm 1 discloses the covariates are used as input [input covariates…].
the set of intervals forming a subdivision of a follow-up period
Groha, pg. 3, col. 2, “For each individual we will observe the process
Y
(
t
)
in the form of discrete jumps over the relevant time interval
[
0
,
T
]
.”
On pg. 3, Groha discloses a relevant time interval
[
0
,
T
]
from the time intervals [the set of intervals forming a subdivision of a follow-up period].
the method comprising: obtaining an input dataset of covariates and time-to-event data of a set of patients
Groha, pg. 15, “we choose three covariates, where one of the covariates has time varying coefficients to model a proportional hazards violation. We choose all coefficients to be of O(1), with a saw-tooth time dependence for the time dependent covariate. We sample 2048 patients for the training set and 1024 patients for the validation and test set respectively with event times between 0 and 100.”
Groha discloses choosing three covariates with time varying coefficients [obtaining an input dataset of covariates] with event times between 0 and 100 [time-to-event data] for a set of 2048 patients for training and 1024 patients for validation [a set of patients].
training the function based on the input dataset
As cited above on pg. 15, Groha discloses that the sampled patient data is for the training set [training the function based on the input dataset].
Regarding Claim 8:
As discussed above, Groha teaches [the] computer-implemented method of claim 1, and Groha further discloses:
wherein the multi-state model is an illness-death model
PNG
media_image1.png
182
333
media_image1.png
Greyscale
PNG
media_image2.png
184
327
media_image2.png
Greyscale
Groha, pg. 3, col. 1, “Figure 1. Example graphs corresponding to the 2-state competing risks model (a) and the illness-death model (b), a popular multi-state model. Competing risks models are a special case of multi-state models that only have one non-absorbing state, whereas in general multi-state models can have arbitrary, even cyclical connections.”
Regarding Claim 9:
As discussed above, Groha teaches [the] computer-implemented method of claim 1, and Groha further discloses:
wherein the illness is a cancer disease including a breast cancer or a disease having an intermediate state and a final state
Groha, pg. 5, col. 2, “To benchmark our proposed model against various survival frameworks, we examine the performance of SURVNODE on the METABRIC breast cancer data set…”
Regarding Claim 10:
As discussed above, Groha teaches [the] computer-implemented method of claim 1, and Groha further discloses:
wherein the medical characteristics comprise characteristics representing a general state of the patient and/or characteristics representing a condition of the patient with respect to the illness
PNG
media_image4.png
434
1002
media_image4.png
Greyscale
Groha, pg. 7, col. 2, Figure 5 depicts medical characteristics of a patient representing a condition of the patient with respect to the illness.
Regarding Claim 11:
Groha discloses:
A method for applying a function machine-learnt by machine-learning the function that is configured… to output a distribution of transition-specific probabilities for each interval of a set of intervals
Groha, pg. 2, col. 2, “The aim of multi-state models is a more granular analysis of time-to-event phenomena…Such models and their state-space can be described by a directed graph as shown in Figure 1(b).”
Pg. 4, col. 2, “With this model, we also have direct access to the hazard rate (the instantaneous risk for a given transition) over time.”
On pg. 2, Groha discloses a multi-state model [machine-learning a function that is configured to…output a distribution of transition-specific probabilities…]. Pg. 4 further specifies that the model works with data over time [for each interval of a set of intervals].
…based on input covariates representing medical characteristics of a patient with respect to a multi-state model of an illness having states and transitions between the states…
PNG
media_image1.png
182
333
media_image1.png
Greyscale
PNG
media_image2.png
184
327
media_image2.png
Greyscale
Groha, pg. 3, col. 1, “Figure 1. Example graphs corresponding to the 2-state competing risks model (a) and the illness-death model (b), a popular multi-state model. Competing risks models are a special case of multi-state models that only have one non-absorbing state, whereas in general multi-state models can have arbitrary, even cyclical connections.”
Pg. 4, col. 2, “The initial conditions are encoded by the covariates of the patient
m
(
0
)
=
f
(
x
)
, where
f
is given by a neural net.”
Pg. 13, C. Algorithm 1
PNG
media_image3.png
503
812
media_image3.png
Greyscale
On pg. 3 and in view of FIG. 1, Groha discloses 1(b), a general multi-state model [a multi-state model of an illness having states and transitions between the states]. On pg. 4, Groha discloses patient covariates [covariates representing medical characteristics of a patient], and further on pg. 13, Algorithm 1 discloses the covariates are used as input [input covariates…].
the set of intervals forming a subdivision of a follow-up period
Groha, pg. 3, col. 2, “For each individual we will observe the process
Y
(
t
)
in the form of discrete jumps over the relevant time interval
[
0
,
T
]
.”
On pg. 3, Groha discloses a relevant time interval
[
0
,
T
]
from the time intervals [the set of intervals forming a subdivision of a follow-up period].
the method comprising: obtaining an input dataset of covariates and time-to-event data of a set of patients
Groha, pg. 15, “we choose three covariates, where one of the covariates has time varying coefficients to model a proportional hazards violation. We choose all coefficients to be of O(1), with a saw-tooth time dependence for the time dependent covariate. We sample 2048 patients for the training set and 1024 patients for the validation and test set respectively with event times between 0 and 100.”
Groha discloses choosing three covariates with time varying coefficients [obtaining an input dataset of covariates] with event times between 0 and 100 [time-to-event data] for a set of 2048 patients for training and 1024 patients for validation [a set of patients].
training the function based on the input dataset
As cited above on pg. 15, Groha discloses that the sampled patient data is for the training set [training the function based on the input dataset].
obtaining input covariates representing medical characteristics of a patient
Groha, pg. 4, col. 2, “The initial conditions are encoded by the covariates of the patient
m
(
0
)
=
f
(
x
)
, where
f
is given by a neural net.”
PNG
media_image3.png
503
812
media_image3.png
Greyscale
On pg. 4, Groha discloses patient covariates [covariates representing medical characteristics of a patient], and further on pg. 13, Algorithm 1 discloses the covariates are used as input [input covariates…]
applying the function to the input covariates, thereby outputting…a distribution of transition-specific probabilities for each interval of a set of time intervals
Groha, pg. 2, col. 2, “The aim of multi-state models is a more granular analysis of time-to-event phenomena…Such models and their state-space can be described by a directed graph as shown in Figure 1(b).”
Pg. 4, col. 2, “With this model, we also have direct access to the hazard rate (the instantaneous risk for a given transition) over time.”
On pg. 2 and in view of Algorithm 1, Groha discloses inputting the covariates and the multi-state model [applying the function to the input covariates, thereby outputting…a distribution of transition-specific probabilities…]. Pg. 4 further specifies that the model works with data over time [for each interval of a set of intervals].
…with respect to a multi-state model of an illness having states and transitions between states…
PNG
media_image1.png
182
333
media_image1.png
Greyscale
PNG
media_image2.png
184
327
media_image2.png
Greyscale
Groha, pg. 3, col. 1, “Figure 1. Example graphs corresponding to the 2-state competing risks model (a) and the illness-death model (b), a popular multi-state model. Competing risks models are a special case of multi-state models that only have one non-absorbing state, whereas in general multi-state models can have arbitrary, even cyclical connections.”
On pg. 3 and in view of FIG. 1, Groha discloses 1(b), a general multi-state model [a multi-state model of an illness having states and transitions between the states].
the set of time intervals forming a sub-division of a follow-up period
Groha, pg. 3, col. 2, “For each individual we will observe the process
Y
(
t
)
in the form of discrete jumps over the relevant time interval
[
0
,
T
]
.”
On pg. 3, Groha discloses a relevant time interval
[
0
,
T
]
from the time intervals [the set of intervals forming a subdivision of a follow-up period].
Regarding Claim 12:
As discussed above, Groha teaches [the] method of claim 11, and further discloses:
computing one or more transition-specific cumulative incidence functions (CIF)
Groha, pg. 15, “Figure 8. Cumulative incidence functions for the two competing outcomes for all benchmarked models on the top…Risk 1 is shown in blue for the estimates and purple for the ground truth, whereas risk 2 is color coded in green for estimates and teal for ground truth.”
Groha discloses cumulative incidence functions for two competing outcomes.
optionally displaying said one or more transition-specific cumulative incidence functions
Groha, pg. 15, Figure 8,
PNG
media_image5.png
731
959
media_image5.png
Greyscale
Groha depicts in Figure 8, the cumulative incidence functions (CIF) [optionally displaying said one or more transition-specific cumulative incidence functions]
identify relapse risk and/or death risk associate to the patient
Groha, pg. 7, Figure 5,
PNG
media_image4.png
434
1002
media_image4.png
Greyscale
As cited above on pg. 15, Groha discloses estimated risks and Figure 5 depicts identifying states such as relapse and death.
and/or
determining a treatment, a treatment adaptation, a follow-up visit, and/or a surveillance with diagnostic tests, including based on the identified relapse risk and/or death risk
Groha, pg. 1, col. 2, “For example, in the case of acute myeloid leukemia, individualized genetic prediction based on a sophisticated multi-stage model was used to tailor personalized treatment within first complete remission…”
Groha discloses determining a personized treatment for a patient [determining a treatment] using the multi-stage model [based on the identified relapse risk and/or death risk].
Regarding Claim 16:
Groha discloses:
A device comprising: a processor; and a non-transitory data storage medium having recorded thereon a data structure having a computer program including instructions for machine-learning a function configured…to output a distribution of transition-specific probabilities for each interval of a set of intervals that when executed by the processor, cause the processor to be configured to
Groha, pg. 2, col. 2, “The aim of multi-state models is a more granular analysis of time-to-event phenomena…Such models and their state-space can be described by a directed graph as shown in Figure 1(b).”
Pg. 4, col. 2, “With this model, we also have direct access to the hazard rate (the instantaneous risk for a given transition) over time.”
On pg. 2, Groha discloses a multi-state model [machine-learning a function that is configured to…output a distribution of transition-specific probabilities…]. Pg. 4 further specifies that the model works with data over time [for each interval of a set of intervals]. Lastly, as Groha’s models are implemented on a computer, it is construed as disclosing a device with a processor, non-transitory data storage medium, computer program, and machine instructions.
…based on input covariates representing medical characteristics of a patient with respect to a multi-state model of an illness having states and transitions between the states…
PNG
media_image1.png
182
333
media_image1.png
Greyscale
PNG
media_image2.png
184
327
media_image2.png
Greyscale
Groha, pg. 3, col. 1, “Figure 1. Example graphs corresponding to the 2-state competing risks model (a) and the illness-death model (b), a popular multi-state model. Competing risks models are a special case of multi-state models that only have one non-absorbing state, whereas in general multi-state models can have arbitrary, even cyclical connections.”
Pg. 4, col. 2, “The initial conditions are encoded by the covariates of the patient
m
(
0
)
=
f
(
x
)
, where
f
is given by a neural net.”
Pg. 13, C. Algorithm 1
PNG
media_image3.png
503
812
media_image3.png
Greyscale
On pg. 3 and in view of FIG. 1, Groha discloses 1(b), a general multi-state model [a multi-state model of an illness having states and transitions between the states]. On pg. 4, Groha discloses patient covariates [covariates representing medical characteristics of a patient], and further on pg. 13, Algorithm 1 discloses the covariates are used as input [input covariates…].
the set of intervals forming a subdivision of a follow-up period
Groha, pg. 3, col. 2, “For each individual we will observe the process
Y
(
t
)
in the form of discrete jumps over the relevant time interval
[
0
,
T
]
.”
On pg. 3, Groha discloses a relevant time interval
[
0
,
T
]
from the time intervals [the set of intervals forming a subdivision of a follow-up period].
obtain an input data set of covariates and time-to-event data of a set of patients
Groha, pg. 15, “we choose three covariates, where one of the covariates has time varying coefficients to model a proportional hazards violation. We choose all coefficients to be of O(1), with a saw-tooth time dependence for the time dependent covariate. We sample 2048 patients for the training set and 1024 patients for the validation and test set respectively with event times between 0 and 100.”
Groha discloses choosing three covariates with time varying coefficients [obtaining an input dataset of covariates] with event times between 0 and 100 [time-to-event data] for a set of 2048 patients for training and 1024 patients for validation [a set of patients].
train the function based on the input data set
As cited above on pg. 15, Groha discloses that the sampled patient data is for the training set [training the function based on the input dataset].
and/or instructions for of a function machine-learnt according to the machine-learning that when executed by the processor causes the processor to be configured to obtain input covariates representing medical characteristics of a patient apply the function to the input covariates, thereby outputting, with respect to a multi-state model of an illness having states and transitions between the states, a distribution of transition-specific probabilities for each interval of a set of time intervals, the set of time intervals forming a subdivision of a follow-up period
and/or a function machine-learnt according to the machine-learning
Regarding Claim 20:
As discussed above, Groha teaches [the] method according to claim 1, and Groha further discloses:
A non-transitory computer readable medium having stored thereon a program that when executed by a computer causes the computer to implement the method according to claim 1.
Groha, pg. 2, col. 2, “The aim of multi-state models is a more granular analysis of time-to-event phenomena…Such models and their state-space can be described by a directed graph as shown in Figure 1(b).”
Since Groha’s models are implemented on a computer, it is construed as disclosing a device with a non-transitory computer readable medium, a program, and a computer.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 2-6, 13-15, and 17-19 are rejected under 35 U.S.C. 103 as being unpatentable over Groha in view of Lee et al. ("DeepHit: A Deep Learning Approach to Survival Analysis with Competing Risks"), hereinafter Lee.
Regarding Claim 2:
As discussed above, Groha teaches [the] computer-implemented method of claim 1, but does not explicitly disclose:
wherein the function further comprises a covariate-shared network and/or a transition-specific subnetwork per transition
However, in the same field, analogous art Lee teaches:
wherein the function further comprises a covariate-shared network and/or a transition-specific subnetwork per transition
Lee, pg. 3, Figure 2, and pg. 3-4, “The shared sub-network takes as inputs the clinical covariates x…Each cause-specific sub-network takes as inputs the pairs
z
=
(
f
s
(
x
)
,
x
)
and produces as output a vector
f
c
k
(
z
)
, which corresponds to the probability of the first hitting time of a specific cause k.”
PNG
media_image6.png
596
712
media_image6.png
Greyscale
Lee discloses a shared sub-network [a covariate-shared subnetwork] and cause-specific sub-networks [transition-specific subnetwork per transition].
Groha, Lee, and the instant application are analogous art because they are all directed to survival analysis.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Groha with Lee to use a shared sub-network and cause-specific sub-networks in order to maintain the input for each cause-specific sub-network. “Second, we maintain a residual connection (He et al. 2016) from the input covariates into the input of each cause-specific sub-network” (Lee, pg. 3, col. 2).
Regarding Claim 3:
As discussed above, Groha in view of Lee teaches [the] computer-implemented method of claim 2, and Lee further teaches:
wherein the covariate-shared subnetwork comprises a respective fully connected neural network and/or at least one transition-specific subnetwork comprises a fully connected neural network
Lee, pg. 3, Figure 2,
PNG
media_image6.png
596
712
media_image6.png
Greyscale
Figure 2 depicts the shared sub-network [covariate-shared subnetwork] and the cause-specific sub-networks [transition-specific subnetwork] as fully-connected.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Groha with Lee for at least the same reasons as given with respect to claim 2.
Regarding Claim 4:
As discussed above, Groha and Lee teaches [the] computer-implemented method of claim 2, and Lee further teaches:
wherein the covariate-shared subnetwork comprises a respective non-linear activation function and/or at least one transition-specific subnetwork comprises a respective non-linear activation function
Lee, pg. 3, col. 2, “First, we utilize a single softmax layer as the output layer of DeepHit in order to ensure that the network learns the joint distribution of K competing events not the marginal distributions of each event.”
Lee discloses using a single softmax layer as the output layer [at least one transition-specific subnetwork comprises a respective non-linear activation function].
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Groha with Lee in order to learn non-linear relationships. “The output of the softmax layer is a probability distribution…given a patient with covariates x, an output element
y
k
,
s
is the (estimated) probability
P
^
(
s
,
k
|
x
)
that the patient will experience the event
k
at time
s
. This architecture drives the network to learn potentially non-linear, even non-proportional, relation ships between covariates and risks” (Lee, pg. 4, col. 1).
Regarding Claim 5:
As discussed above, Groha in view of Lee teaches [the] computer-implemented method of claim 2, and Lee further teaches:
wherein each transition-specific subnetwork is followed by a softmax layer
Lee, pg. 3, Figure 2,
PNG
media_image6.png
596
712
media_image6.png
Greyscale
Lee, pg. 3, col. 2, “First, we utilize a single softmax layer as the output layer of DeepHit in order to ensure that the network learns the joint distribution of K competing events not the marginal distributions of each event.”
Lee discloses using a softmax layer as the output layer after the cause-specific sub-networks [each transition-specific subnetwork is followed by a softmax layer].
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Groha with Lee in order to learn non-linear relationships. “The output of the softmax layer is a probability distribution…given a patient with covariates x, an output element
y
k
,
s
is the (estimated) probability
P
^
(
s
,
k
|
x
)
that the patient will experience the event
k
at time
s
. This architecture drives the network to learn potentially non-linear, even non-proportional, relation ships between covariates and risks” (Lee, pg. 4, col. 1).
Regarding Claim 6:
As discussed above, Groha in view of Lee teaches [the] computer-implemented method of claim 2, and Lee further teaches:
wherein the multi-state model further comprises competing transitions, and transition-specific subnetworks of the competing transitions share a common softmax layer
Lee, pg. 3, Figure 2,
PNG
media_image6.png
596
712
media_image6.png
Greyscale
Lee, pg. 3, col. 2, “First, we utilize a single softmax layer as the output layer of DeepHit in order to ensure that the network learns the joint distribution of K competing events not the marginal distributions of each event.”
Lee discloses using a softmax layer as the output layer after the competing cause-specific sub-networks [comprises competing transitions, and transition-specific subnetworks of the competing transitions share a common softmax layer].
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Groha with Lee in order to learn joint distributions. “First, we utilize a single softmax layer as the output layer of DeepHit in order to ensure that the network learns the joint distribution of K competing events not the marginal distributions of each event.” (Lee, pg. 3, col. 2).
Regarding Claims 13-15:
Claims 13-15 are method claims corresponding to the computer-implemented method claims 2-4 and is rejected for at least the same reasons as given in the rejection of claim 2-4. In particular, 13:2, 14:3, 15:4.
Regarding Claims 17-19:
Claims 17-19 are device claims corresponding to the computer-implemented method claims 2-4 and is rejected for at least the same reasons as given in the rejection of claim 2-4. In particular, 17:2, 18:3, 19:4.
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Groha in view of Raghavan et al. ("Real-Time Update of Multi-State System Reliability using Prognostic Data-Driven Techniques"), hereinafter Raghavan.
Regarding Claim 7:
As discussed above, Groha teaches [the] computer-implemented method of claim 1, and further discloses:
wherein the training further comprises minimizing a loss function which includes a likelihood term, and/or a regularization term that penalizes
Groha, pg. 13, “For training the model minimizing the negative log-likelihood”
Groha discloses training the model [training further comprises] minimizing the negative log-likelihood [minimizing a loss function which includes a likelihood term that penalizes…].
Groha does not explicitly disclose:
in weight matrices, first order differences of weights associated with two adjacent time intervals, and/or in bias vectors, first order differences of biases associated with two adjacent time intervals
However, in the same field, analogous art Raghavan teaches:
…first order differences…associated with two adjacent time intervals…
Raghavan, pg. 1, col. 2, “The reliability of [multi-state systems] MSS is commonly assessed using the Markov chain theory [7] where the time-dependent state probabilities can be expressed in terms of the transition rates (λ) between adjacent states through a system of first order differential equations. The most important piece of data needed for quantifying the reliability of MSS is the value of for the different inter-state transitions.”
Raghavan teaches that it is common in multi-state systems to use first order differential equations [first order differences…associated with two adjacent time intervals] which expresses time-dependent state probabilities.
Groha, Raghavan, and the instant application are analogous art because they are all directed to multi-state models.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Groha with Raghavan to use first order differential equations in order to increase the reliability and accuracy of the system. “The reliability of MSS is commonly assessed using the Markov chain theory [7] where the time-dependent state probabilities can be expressed in terms of the transition rates (λ) between adjacent states through a system of first order differential equations. The most important piece of data needed for quantifying the reliability of MSS is the value of λ for the different inter-state transitions” (Raghavan, pg. 1).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to STEVEN PHUNG whose telephone number is (703) 756-1499. The examiner can normally be reached Monday-Thursday: 9:00AM-4:00PM ET.
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, KAMRAN AFSHAR can be reached at (571) 272-7796. 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.
/STEVEN PHUNG/Examiner, Art Unit 2125
/KAMRAN AFSHAR/Supervisory Patent Examiner, Art Unit 2125