DETAILED ACTION
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 .
Status of Claims
This action is a final rejection
Claims 1-11, 13-20 are pending
Claims 16-20 are withdrawn
Claim 12 was cancelled
Claims 1, 9-10, 13-15 were amended
Claims 1-11, 13-15 are rejected under 35 USC § 101
Claims 1-11, 13-15 are rejected under 35 USC § 103
Priority
Acknowledgement is made of Applicant’s claim for a foreign priority date of 3-18-2022
Information Disclosure Statement
The information disclosure statements (IDS) submitted on 10-23-2024, is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
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 1-11, 13-15 are not patent eligible because the claimed invention is directed to an abstract idea without significantly more.
Analysis
First, claims are directed to one or more of the following statutory categories: a process, a machine, a manufacture, and a composition of matter. Regarding claims 1-11, 13-15 the claims recite an abstract idea of “multi-modal patient representation”.
Independent Claims 1 and 15 are rejected under 35 U.S.C 101 based on the following analysis.
-Step 1 (Does the claim fall within a statutory category? YES): Claims 1 and 15 recite a method, and a system respectively of “multi-modal patient representation“.
-Step 2A Prong One (Does the claim fall within at least one of the groupings of abstract ideas?: YES): The claimed invention:
accessing a set of medical data associated with a patient, wherein the set of medical data includes a plurality of modalities of medical data, wherein each of the modalities, consists of one data type and is associated with one data source;
inputting one or more of the modalities each having a data type of laboratory testing data into a first machine-learning model trained to generate a first vector representation;
inputting another one of the modalities of medical data, into a second machine- learning model trained to generate a second vector representation of the second modality of medical data, wherein the second modality of medical data consists of a second data type;
generating by a third machine-learning model, a combined vector representation based on the first vector representation and the second vector representation;
storing the combined vector representation to a database
retrieving, based on the patient, the combined vector representation from the database;
identifying, by an ensemble model, a cohort of patients similar to the patient based on a patient similarity inference using the retrieved combined vector representation; and
generating, by the ensemble model, a clinical prediction for the patient based on the retrieved combined vector representation.
belonging to the grouping of certain methods of organizing human activity under managing personal behavior or relationships or interactions between people as it recites “multi-modal patient representation”. (refer to MPP 2106.04(a)(2)). Accordingly this claim recites an abstract idea.
-Step 2A Prong Two (Are there additional elements in the claim that imposes a meaningful limit on the abstract idea? NO).
Claims 1, 15 recite:
first machine-learning model trained to generate a first vector representation;
second machine- learning model trained to generate a second vector representation;
generating by a third machine-learning model, a combined vector representation
storing the combined vector representation to a database;
retrieving, the combined vector representation from the database;
identifying, by an ensemble model, a cohort of patients;
generating, by the ensemble model, a clinical prediction.
Claim 15 recites:
one or more non-transitory computer-readable storage media including instructions;
one or more processors coupled to the one or more storage media, the one or more processors configured to execute the instructions and cause the system to access a set of medical data associated with a patient.
Amounting to mere instructions to implement an abstract idea on a computer, or merely use a computer as a tool to implement the abstract idea. (refer to MPEP 2106.05(f)). Accordingly, these additional elements, when considered separately and as an ordered combination do not integrate the judicial exception/abstract idea into a “practical application” of the judicial exception because they do not impose any meaningful limit on practicing the judicial exception.
-Step 2B (Does the additional elements of the claim provide an inventive concept?: NO. As discussed previously with respect to Step 2A Prong Two:
first machine-learning model trained to generate a first vector representation;
second machine- learning model trained to generate a second vector representation;
generating by a third machine-learning model, a combined vector representation
storing the combined vector representation to a database;
retrieving, the combined vector representation from the database;
identifying, by an ensemble model, a cohort of patients;
generating, by the ensemble model, a clinical prediction.
Claim 15 recites:
one or more non-transitory computer-readable storage media including instructions;
one or more processors coupled to the one or more storage media, the one or more processors configured to execute the instructions and cause the system to access a set of medical data associated with a patient.
Amounting to mere instructions to implement an abstract idea on a computer, or merely use a computer as a tool to implement the abstract idea. (refer to MPEP 2106.05(f)) Accordingly, even when viewed as a whole the claim does not provide an inventive concept (significantly more than the abstract idea) and hence the claim is ineligible.
Dependent Claims:
Step 2A Prong One: The following dependent claims recites additional limitations that further define the abstract idea of “managing exposure of confidential user information in a user data record to a third party”. The claim limitations include:
Claim 2: wherein a data type consists of whole slide images, radiological images, medical graph images, other medical images, genomics data, proteomics data, transcriptomics data, metabolomics data, radiomics data, toxigenomics data, multi-omics data, medication data, medical diagnostics data, medical procedures data, medical symptoms data, demographics data, patient lifestyle data, physical activity data, body mass index (BMI) data, family history data, socioeconomics data, geographic environment data, or another type of digital medical data relating to the patient.
Claim 3: wherein a data source consists of a randomized controlled trial for medical treatment, a provider of real-world medical data, or a provider of patient knowledge graphs.
Claim 4: wherein at least one of the plurality of modalities of medical data, comprises a longitudinal dataset of medical data.
Claim 5: wherein the first vector representation comprises a first dimensionless value representative of a first plurality of datasets of the first data type.
Claim 6: wherein the first .. model was trained by:
inputting a first plurality of datasets to the first .. model, the first plurality of datasets corresponding to a first modality of medical data and utilizing the first .. model to encode the first plurality of datasets into the first vector representation; and
wherein a dimension of the first vector representation is reduced with respect to a dimension of the first plurality of datasets
Claim 7: wherein the second vector representation comprises a second dimensionless value representative of a second plurality of datasets of the second data type.
Claim 8:
inputting a second plurality of datasets to the second model, the second plurality of datasets corresponding to the second modality of medical data and utilizing the second model to encode the second plurality of datasets into the second vector representation and
wherein a dimension of the second vector representation is reduced with respect to a dimension of the second plurality of datasets
Claim 9:
prior to generating the combined vector representation, inputting a third modality of medical data of the plurality of modalities of medical data, (104), into a fourth model trained to generate a third vector representation of the third modality of medical data, wherein the third modality of medical data consists of a third data type; and
generating the combined vector representation based on the first vector representation, the second vector representation, and the third vector representation
Claim 10: wherein generating the combined vector representation comprises generating a comprehensive data representation of a biomedical state of the patient.
Claim 11: wherein generating the combined vector representation comprises generating a reduced-dimension dataset as compared to the set of medical data.
Claim 13: wherein retrieving, based on the patient, the combined vector representation in response to receiving one or more requests for medical data associated with the patient..;
Claim 14: wherein generating the clinical prediction comprises generating a predicted survivability for the patient, generating a predicted future disease development for the patient, generating a predicted treatment response for the patient, generating a predicted diagnosis for the patient,.
Step 2A Prong Two (Are there additional elements in the claim that imposes a meaningful limit on the abstract idea? NO). The following dependent claims recite mere instructions to implement an abstract idea on a computer, or merely use a computer as a tool to implement the abstract idea. (refer to MPEP 2106.05(f)). Accordingly, the claims as a whole do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. The claims include:
Claim 6: the first machine-learning model.
Claim 8: second machine-learning model.
Claim 9: fourth machine-learning model
Claim 13: database;
Claim 14: generating, by the ensemble model, the clinical prediction.
Step 2B (Does the additional elements of the claim provide an inventive concept?: NO). As discussed previously with respect to Step 2A Prong Two, the following dependent claims recite mere instructions to implement an abstract idea on a computer, or merely use a computer as a tool to implement the abstract idea. (refer to MPEP 2106.05(f)). Accordingly, the claim does not provide an inventive concept (significantly more than the abstract idea) and hence the claim is ineligible. The claims include:
Claim 6: the first machine-learning model.
Claim 8: second machine-learning model.
Claim 9: fourth machine-learning model
Claim 13: database;
Claim 14: generating, by the ensemble model, the clinical prediction.
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 set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied 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
non-obviousness.
Claims 1-4, 9-10, 12-15 are rejected under 35 U.S.C. 103 as being un-patentable by Liu et.al. (EP 4199002 A1) hereinafter “Liu”; in view of Yamamoto et.al (US 20220245523 A1) hereinafter “Yamamoto”; in view of Ribbing et.al (WO 2011070461 A2) hereinafter “Ribbing”; in view of Will et.al (US 20200243167 A1) hereinafter “Will”.
Regarding claims 1, 15 Liu teaches:
one or more non-transitory computer-readable storage media including instructions and one or more processors coupled to the one or more storage media, the one or more processors configured to execute the instructions and cause the system to access a set of medical data associated with a patient (See at least [Page 5, lines 26-31] via: “..The processor may be associated with one or more non-transitory storage media such as volatile and non-volatile computer memory such as RAM, PROM, EPROM, and EEPROM. The non-transitory storage media may be encoded with one or more programs that, when executed on one or more processors and/or controllers, perform the required functions. Various storage media may be fixed within a processor or controller or may be transportable, such that the one or more programs stored thereon can be loaded into the signal processing unit, property determination unit and/or processor...”)
accessing a set of medical data associated with a patient, wherein the set of medical data includes a plurality of modalities of medical data, wherein each of the modalities, consists of one data type and is associated with one data source; (See at least [Page 2, lines 28-29] via: “...receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type...”; in addition see at least [Page 5, lines 6-11] via: “... plurality of patient data records comprises: (i) a first patient data record comprising at least one medical image (i.e. medical imaging data); (ii) a second patient data record comprising a non-image medical data comprising a physiological signal; and (iii) a third patient data record comprising text-based medical data. This combination of data is traditionally particularly difficult to combine but the disclosed methods provide improved pre-processing that enables the more accurate determination of health parameters using these three data types...”; in addition see at least [Page 6, lines 30-36] via: “... the data received is heterogeneous. In this embodiment, the medical imaging data and non-medical imaging data are related in that they have relevance to a health condition under investigation. The input data may be extracted or received from a healthcare database or plural databases. For example, this data may be extracted or received from EMRs, Radiology Information System, Cardiology Information System and/or PACS. Alternatively, this may be extracted or received from the devices or instruments used to obtain the data, such as a medical device or sensor. Interfacing with clinical, operational, and demographic data sources can be done with any available state-of-art IT communication protocol such as HL7, DICOM and FHIR...”; in addition see at least [Page 4, lines 15-23] via: “...the plurality of patient data records comprise patient data records obtained at a plurality of points in time; and wherein analyzing the vectors using a machine learning model comprising applying a weighting to each patient data record based on the point in time that the record was created. The weighting may apply more weight to the more recently obtained data, in some embodiments. An example of this may be weighting a scan that has been taken recently more heavily than a scan that was taken at an earlier point in time. This enables the accuracy of predictions pertaining to the diagnosis, onset and/or progression of a disease to be improved. Although some risk scores have developed (e.g. Framingham risk score) for disease prediction, these are general and only depend on one study or investigation. Embodiments taking into account longitudinal data for disease prediction provide improved accuracy over these by taking into account data obtained over time and monitoring the changes...”..)
inputting one or more of the modalities each having a data type of laboratory testing data into a first machine-learning model trained to generate a first vector representation; (See at least [Page 5, lines 38-41] via: “...generating a vector for each of the plural patient data records by processing each patient data record using a corresponding encoding algorithm, wherein the encoding algorithm used to generate the vector is selected based on the type of patient data record and wherein the vectors are for use by a machine learning model...”; in addition see at least [Page 5, lines 6-7] via: “... plurality of patient data records comprises: (i) a first patient data record comprising at least one medical image (i.e. medical imaging data)..”; in addition see at least [Page 6, lines 37-41] via: “...generating a first vector 120 for the medical imaging data received in the first step 110. Generating the first vector 120 comprises processing each patient data record (i.e. the medical imaging data in this case) using an encoding algorithm which is optimized and specific to medical imaging data. The encoding algorithm may be an AI or machine learning algorithm or model. ..”; )..”; in addition see at least [Page 7, lines 3-4] via: “... for each type of patient data, a model that is suitable for the data type can be built. These model wills convert the data into vectors from different data sources...”)
inputting another one of the modalities of medical data, into a second machine- learning model trained to generate a second vector representation of the second modality of medical data, wherein the second modality of medical data consists of a second data type; (See at least [Page 2 lines 28-32] via: “...receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type... generating a vector for each of the plural patient data records by processing each patient data record using a corresponding encoding algorithm, wherein the encoding algorithm used to generate the vector is selected based on the type of patient data record and wherein the vectors are for use by a machine learning model...”; in addition see at least [Page 5, lines 6-8] via: “... plurality of patient data records comprises: ... (ii) a second patient data record comprising a non-image medical data comprising a physiological signal..”; in addition see at least [Page 6, lines 43-46 and Page 7, lines 1-2] via: “...generating a second vector 120' for the non-medical imaging data received in the first step 110'. In this case, generating the vector 120' comprises processing each patient data record (i.e. the non-medical imaging data in this case) using an encoding algorithm which is optimized and specific to the specific data type. This is a different encoding algorithm to that used to encode the first vector such that the data extraction is optimized for each data type thereby providing an output that is especially accurate and that key aspects and information in each particular data type is retained and taken into account in subsequent processing...”)
However, Liu is silent the following limitation that is taught by Yamamoto
generating by a third machine-learning model a combined vector representation based on the first vector representation and the second vector representation; (See at least [0043] via: “..The machine learning device 100 combines the extracted first feature vector 111 and the extracted second feature vector 112 together to generate a third feature vector 113. For example, the machine learning device 100 couples the second feature vector 112 with first feature vector 111 to generate the third feature vector 113..”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify Liu with Yamamoto. Liu teaches collating patient data for analysis comprising receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type; and generating a vector for each of the plural patient data records by processing each patient data record using a corresponding encoding algorithm, wherein the encoding algorithm used to generate the vector is selected based on the type of patient data record and wherein the vectors are for use by a machine learning model. However, Liu fails to disclose combining a first feature vector with a second feature vector to obtain a third feature vector as taught by Yamamoto. Combining Liu and Yamamoto is helpful to medical practitioners that may desire to combine into one vector multiple types and modes of medical data of a particular patient into one summary vector that contains the combined patient data.
However, Liu and Yamamoto are silent the following limitation that is taught by Ribbing:
storing the combined vector representation to a database database 60. In this way, recomputation of the features vectors can be avoided...”; in addition see at least [Page 6, lines 31-32 and Page 7, lines 1-8] via: “... The stored medical data includes both image medical data acquired by the medical imagers 10, 12, 14 (or by different, additional, or other medical imagers) and non-image medical data acquired by the non-image diagnostic apparatuses or facilities 20, 22 (or by different, additional, or other non-image diagnostic apparatuses or facilities). Stored data are suitably tagged with metadata identifying data type, data source, acquisition date, laboratory identification, and other relevant information. Medical data are stored in the database 30 as the data are acquired, and accordingly at any given point in time the stored medical data pertaining to a particular patient may be "incomplete" in the sense that certain tests ordered by medical personnel may not yet have been performed, or certain tests may not yet have been ordered..”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify Liu and Yamamoto with Ribbing. Liu and Yamamoto teach collating patient data for analysis comprising receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type; and generating a summary vector that combines vectors for each of the plural patient data records. However, Liu and Yamamoto fails to disclose saving the summary vector into a database and subsequently retrieving the summary vector from the database when needed as taught by Ribbing. Combining Liu, Yamamoto and Ribbing is helpful to medical practitioners that may desire to store the summary medical vector of a patient into a database and subsequently be able to recall it at a later time when needed.
retrieving, based on the patient, the combined vector representation from the database; (See at least [Page 10, lines 7-10] via: “...If the patient of interest undergoes an additional test, then the features vector 56 can be retrieved from the features vectors database 60, any vector elements affected by the results of the additional test are updated by the components 52, 54, and the updated features vector stored back into the features vectors database 60..”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify Liu and Yamamoto with Ribbing. Liu and Yamamoto teach collating patient data for analysis comprising receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type; and generating a summary vector that combines vectors for each of the plural patient data records. However, Liu and Yamamoto fails to disclose saving the summary vector into a database and subsequently retrieving the summary vector from the database when needed as taught by Ribbing. Combining Liu, Yamamoto and Ribbing is helpful to medical practitioners that may desire to store the summary medical vector of a patient into a database and subsequently be able to recall it at a later time when needed.
However, Liu, Yamamoto and Ribbing are silent the following limitation that is taught by Will:
identifying, by an ensemble model, a cohort of patients similar to the patient based on a patient similarity inference using the retrieved combined vector representation; (See at least [0049] via: “.. FIG. 4 illustrates example operations 400 for predicting clinical trial eligibility for patients...”; in addition see at least [0051] via: “..At step 420, a cohort of patients that are clinically similar to the first patient is identified based on the plurality of attributes. For example, trial recommendation engine 144 of FIG. 1 may provide the plurality of current values for the plurality of attributes along with identifying information of a first attribute as inputs to a predictive model, and the predictive model may determine the cohort of patients that are clinically similar to the first patient based on the cohort of patients having similar attribute values to the first patient..”; in addition see at least [0024] via: “..In certain embodiments a single prediction model is used, while in other embodiments a plurality (or ensemble) of prediction models may be used ..” )
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify Liu , Yamamoto and Ribbing with Will. Liu, Yamamoto, and Ribbing teach collating patient data for analysis comprising receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type; and generating a summary vector that combines vectors for each of the plural patient data records, whereby the summary vector may be saved and later retrieved from a database. However, Liu, Yamamoto and Ribbing fails to disclose identifying a cohort of patients similar to the patient as described by the summary vector and predicting with a probability or ensemble model medical issues pertaining to the patient based data related to the cohort of patients.. Combining Liu, Yamamoto, Ribbing and Will is helpful to medical practitioners that may desire to predict medical issues of a particular patient based on known and established medical issues from a cohort of patients that have been identified to have similar attributes to the patient
generating, by the ensemble model, a clinical prediction for the patient based on the retrieved combined vector representation. (See at least [0052] via: “..At step 430, data associated with the cohort of patients is analyzed to determine an attribute trend for at least a first attribute of the plurality of attributes. For example, the attribute trend for the first attribute (e.g., identified at step 420) may be identified by the predictive model according to its training based on attribute values over time for the cohort of patients. In one example, the trend comprises an increase in a prescribed dosage of a particular medication that commonly occurs when other attributes are present...”; in addition see at least [0053] via: “..At step 440, a predicted value for the first attribute is generated based on the current value of the first attribute and the attribute trend for the first attribute. For example, the predictive model may output the predicted value for the first attribute by applying the attribute trend to the current value of the first attribute. In one example, the attribute trend comprises an increase in a prescribed dosage of a particular medication, and the predicted value is determined (e.g., using the predictive model) by applying the increase to the current prescribed dosage for the particular medication for the patient. For example, the attribute trend may comprise a direction and magnitude, and the predicted value may be generated by modifying the current value of the first attribute based on the direction and magnitude of the attribute trend...”; in addition see at least [0054] via: “..At step 450, a plurality of clinical trials is identified based on the first attribute. For example, trial recommendation engine 144 of FIG. 1 may retrieve clinical trial data from clinical trial data store 150 of FIG. 1, and may identify the plurality of clinical trials by determining which clinical trials in the clinical trial data include qualifying or disqualifying criteria related to the first attribute...”; in addition see at least [0055] via: “..At step 460, a probability that the first patient will be eligible for each of the plurality of clinical trials at a future time is generated based on the predicted value for the first attribute. For example, the predicted value for the first attribute may be provided (e.g., along with the plurality of attributes and identifying information of each of the plurality of clinical trials) as an input to a predictive model, which may output the probability..”; in addition see at least [0056] via: “..The probability determined at step 460 may be used to determine whether the first patient is eligible to participate in each of the plurality of clinical trials, such as based on whether the probability for a given clinical trial is above a threshold. Once eligibility is determined, additional processing may be performed to determine the most relevant and/or most beneficial clinical trials for a given patient, which may be provided in a ranked list to a user via a user interface...”; in addition see at least [0024] via: “..In certain embodiments a single prediction model is used, while in other embodiments a plurality (or ensemble) of prediction models may be used ..” )
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify Liu , Yamamoto and Ribbing with Will. Liu, Yamamoto, and Ribbing teach collating patient data for analysis comprising receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type; and generating a summary vector that combines vectors for each of the plural patient data records, whereby the summary vector may be saved and later retrieved from a database. However, Liu, Yamamoto and Ribbing fails to disclose identifying a cohort of patients similar to the patient as described by the summary vector and predicting with a probability or ensemble model medical issues pertaining to the patient based data related to the cohort of patients. Combining Liu, Yamamoto, Ribbing and Will is helpful to medical practitioners that may desire to predict medical issues of a particular patient based on known and established medical issues from a cohort of patients that have been identified to have similar attributes to the patient.
Regarding claim 2: Liu, Yamamoto, Ribbing and Will teach the invention as claimed and detailed above with respect to claim 1. Liu also teaches:
wherein a data type consists of whole slide images, radiological images, medical graph images, other medical images, genomics data, proteomics data, transcriptomics data, metabolomics data, radiomics data, toxigenomics data, multi-omics data, medication data, medical diagnostics data, medical procedures data, medical symptoms data, demographics data, patient lifestyle data, physical activity data, body mass index (BMI) data, family history data, socioeconomics data, geographic environment data, or another type of digital medical data relating to the patient. (See at least [Page 5, lines 6-9] via: “...the plurality of patient data records comprises: (i) a first patient data record comprising at least one medical image (i.e. medical imaging data); (ii) a second patient data record comprising a non-image medical data comprising a physiological signal; and (iii) a third patient data record comprising text-based medical data...”)
Regarding claim 3: Liu, Yamamoto, Ribbing and Will teach the invention as claimed and detailed above with respect to claim 1. Liu also teaches:
wherein a data source consists of a randomized controlled trial for medical treatment, a provider of real-world medical data, or a provider of patient knowledge graphs. (See at least [Page 6, lines 30-35] via: “...the medical imaging data and non-medical imaging data are related in that they have relevance to a health condition under investigation. The input data may be extracted or received from a healthcare database or plural databases. For example, this data may be extracted or received from EMRs, Radiology Information System, Cardiology Information System and/or PACS. Alternatively, this may be extracted or received from the devices or instruments used to obtain the data, such as a medical device or sensor...”)
Regarding claim 4: Liu, Yamamoto, Ribbing and Will teach the invention as claimed and detailed above with respect to claim 1. Liu also teaches:
wherein at least one of the plurality of modalities of medical data, comprises a longitudinal dataset of medical data. (See at least [Page 4, lines 10-11] via: “... data used in the method has been obtained at different time points (i.e. longitudinal data) ...”; in addition see at least [Page 4, lines 22-23] via: “... Embodiments taking into account longitudinal data for disease prediction provide improved accuracy over these by taking into account data obtained over time and monitoring the changes...”; in addition see at least [Page 4, lines 24-25 ] via: “...there is almost always a vast amount of longitudinal information in a patient's medical records, and the healthcare system more generally, which can be used..”; in addition see at least [Page 4, lines 26-29] via: “ ...the patient data comprises data obtained at plural points in time (e.g. different days or studies), each patient data comprises time stamps associated therewith and generation of the vector comprises time-resolved feature vectors based on the time stamps. In such embodiments, subsequent processing determines the disease onset/progress prediction based on a plurality of time-resolved feature vectors..”; in addition see at least [Page 5, lines 1-3] via: “... the plurality of patient data records comprise patient data records obtained at a plurality of points in time; and generating a vector for each of the plural patient data record comprises generating time-resolved feature vectors...”)
Regarding claim 9: Liu, Yamamoto, Ribbing and Will teach the invention as claimed and detailed above with respect to claim 1. Liu also teaches:
prior to generating the combined vector representation, inputting a third modality of medical data of the plurality of modalities of medical data, into a fourth machine-learning model trained to generate a third vector representation of the third modality of medical data, wherein the third modality of medical data consists of a third data type; (See at least [Page 8, lines 45-47] via: “...generating a third vector for the text-based medical data 323 using a text-based data encoding model (i.e. a model adapted or configured for converting text-based data into a vector)..”) and
generating the combined vector representation based on the first vector representation, the second vector representation, and the third vector representation. (See at least [Page 8, lines 41-49] via: “...generating vectors for the first data set 320, which itself comprises generating a first vector for the medical imaging data 321 using a medical imaging encoding model (i.e. a model adapted or configured for converting medical imaging data into a vector), generating a second vector for the medical data relating to a physiological signal 322 using a physiological signal encoding model (i.e. a model adapted or configured for converting physiological signal data into a vector) and generating a third vector for the text-based medical data 323 using a text-based data encoding model (i.e. a model adapted or configured for converting text-based data into a vector). These are subsequently combined 325 to provide a multi-dimensional or stacked vector encoding temporal information. The use of data-type-specific models ensures that all of the key information present in each of the data is retained and encoded into the vectors...”)
Regarding claim 10: Liu, Yamamoto, Ribbing and Will teach the invention as claimed and detailed above with respect to claim 1. Liu also teaches:
wherein generating the combined vector representation comprises generating a comprehensive data representation of a biomedical state of the patient. (See at least [Page 9, lines 10-14] via: “...The resultant vectors from this process are then combined and analyzed in a subsequent processing step 330. In particular, the vectors are used to train a machine learning model, which in this embodiment comprises a recurrent neural network which analyses and takes into account the temporal aspect of the data. The method in this embodiment includes applying a weighting to each patient data record based on the point in time that the record was created...”; in addition see at least [Page 9, lines 15-16] via: “...The result from the analysis is an output 340 in the form of a prediction relating to the onset of a condition for a patient, which takes account of the three studies and the data within each of the studies...”; in addition see at least [Page 3, lines 10-12] via: “... the use of a machine learning model or algorithm to provide an output relating to the health of the patient (e.g. a prediction relating to the onset or progression of a condition), such an output being more accurate and more sensitive to changes due to the data-type-specific pre-processing steps.)
Regarding claim 13: Liu, Yamamoto, Ribbing and Will teach the invention as claimed and detailed above with respect to claim 1. However Liu, Yamamoto and Will are silent the following claim that is taught by Ribbing:
wherein retrieving, based on the patient, the combined vector representation from the database is in response to receiving one or more requests for medical data associated with the patient, generated for each patient is accumulated in a features vectors database 60. In this way, recomputation of the features vectors can be avoided. If the patient of interest undergoes an additional test, then the features vector 56 can be retrieved from the features vectors database 60, any vector elements affected by the results of the additional test are updated by the components 52, 54, and the updated features vector stored back into the features vectors database 60...”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify Liu and Yamamoto with Ribbing. Liu and Yamamoto teach collating patient data for analysis comprising receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type; and generating a summary vector that combines vectors for each of the plural patient data records. However, Liu and Yamamoto fails to disclose saving the summary vector into a database and subsequently retrieving the summary vector from the database when needed as taught by Ribbing. Combining Liu, Yamamoto and Ribbing is helpful to medical practitioners that may desire to store the summary medical vector of a patient into a database and subsequently be able to recall it at a later time when needed..
Regarding claim 14: Liu, Yamamoto, Ribbing and Will teach the invention as claimed and detailed above with respect to claim 1 & 13. Liu also teaches:
wherein generating, by the ensemble model, the clinical prediction comprises generating a predicted survivability for the patient, generating a predicted future disease development for the patient, generating a predicted treatment response for the patient, generating a predicted diagnosis for the patient
Claims 5, 7, 11 are rejected under 35 U.S.C. 103 as being un-patentable by Liu in view of Yamamoto, in view of Ribbing, in view of Will, in further view of Chen et.al (CN 114550946 A) hereinafter “Chen”
Regarding claim 5: Liu, Yamamoto, Ribbing and Will teach the invention as claimed and detailed above with respect to claim 1. Liu, Yamamoto, Ribbing and Will are silent the following claim that is taught by Chen:
wherein the first vector representation comprises a first dimensionless value representative of a first plurality of datasets of the first data type. (See at least [Page 11, lines 40-43] via: “...perform softmax normalization processing to the vector, so as to obtain the numerical value vector. by normalization processing, so that each component in the numerical value vector is defined in a certain interval, for example, [0, 1], the numerical value vector is the normalized vector of the numerical value data...”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify Liu, Yamamoto, Ribbing and Will with Chen. Liu teaches collating patient data for analysis comprising receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type; and generating a vector for each of the plural patient data records by processing each patient data record using a corresponding encoding algorithm, wherein the encoding algorithm used to generate the vector is selected based on the type of patient data record and wherein the vectors are for use by a machine learning model. However, Liu fails to disclose normalizing each component of the numerical value vector (equivalent to a dimensionless representation of a vector) as taught by Chen. Combining Liu and Chen is helpful in combining multiple vectors by ensuring that each vector contributing to the sum has equal weight.
Regarding claim 7: Liu, Yamamoto, Ribbing and Will teach the invention as claimed and detailed above with respect to claim 1. Liu, Yamamoto, Ribbing and Will are silent the following claim that is taught by Chen:
wherein the second vector representation comprises a second dimensionless value representative of a second plurality of datasets of the second data type. (See at least [Page 11, lines 40-43] via: “...perform softmax normalization processing to the vector, so as to obtain the numerical value vector. by normalization processing, so that each component in the numerical value vector is defined in a certain interval, for example, [0, 1], the numerical value vector is the normalized vector of the numerical value data...”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify Liu, Yamamoto, Ribbing and Will with Chen. Liu teaches collating patient data for analysis comprising receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type; and generating a vector for each of the plural patient data records by processing each patient data record using a corresponding encoding algorithm, wherein the encoding algorithm used to generate the vector is selected based on the type of patient data record and wherein the vectors are for use by a machine learning model. However, Liu fails to disclose normalizing each component of the numerical value vector (equivalent to a dimensionless representation of a vector) as taught by Chen. Combining Liu and Chen is helpful in combining multiple vectors by ensuring that each vector contributing to the sum has equal weight.
Regarding claim 11: Liu, Yamamoto, Ribbing and Will teach the invention as claimed and detailed above with respect to claim 1. Liu, Yamamoto, Ribbing and Will are silent the following claim that is taught by Chen:
wherein generating the combined vector representation comprises generating a reduced-dimension dataset as compared to the set of medical data. (See at least [Page 3, lines 18-19] via: “...fusing the first weight vector and the second weight vector to obtain the disease analysis vector corresponding to the medical record data...”; in addition see at least [Page 11, lines 40-43] via: “...perform softmax normalization processing to the vector, so as to obtain the numerical value vector. by normalization processing, so that each component in the numerical value vector is defined in a certain interval, for example, [0, 1], the numerical value vector is the normalized vector of the numerical value data...” )
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify Liu, Yamamoto, Ribbing and Will with Chen. Liu teaches collating patient data for analysis comprising receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type; and generating a vector for each of the plural patient data records by processing each patient data record using a corresponding encoding algorithm, wherein the encoding algorithm used to generate the vector is selected based on the type of patient data record and wherein the vectors are for use by a machine learning model. However, Liu fails to disclose normalizing each component of the numerical value vector (equivalent to a dimensionless representation of a vector) and subsesquently fusing or combining multiple vectors as taught by Chen. Combining Liu and Chen is helpful in combining multiple vectors each representing a set of medical data into one by ensuring that each vector contributing to the sum has equal weight resulting in a reduced dimension medical representation of a patient.
Claims 6, 8 are rejected under 35 U.S.C. 103 as being un-patentable by Liu in view of Yamamoto, in view of Ribbing, in view of Will, in further view of Molero et.al (WO 2021236702 A1) hereinafter “Molero”
Regarding claim 6: Liu, Yamamoto, Ribbing and Will teach the invention as claimed and detailed above with respect to claim 1. Liu also teaches:
inputting a first plurality of datasets to the first machine-learning model, the first plurality of datasets corresponding to a first modality of medical data and utilizing the first machine-learning model to encode the first plurality of datasets into the first vector representation; (See at least [Page 5, lines 38-41] via: “...generating a vector for each of the plural patient data records by processing each patient data record using a corresponding encoding algorithm, wherein the encoding algorithm used to generate the vector is selected based on the type of patient data record and wherein the vectors are for use by a machine learning model...”; in addition see at least [Page 5, lines 6-7] via: “... plurality of patient data records comprises: (i) a first patient data record comprising at least one medical image (i.e. medical imaging data)..”; in addition see at least [Page 6, lines 38-41] via: “...generating a first vector 120 for the medical imaging data received in the first step 110. Generating the first vector 120 comprises processing each patient data record (i.e. the medical imaging data in this case) using an encoding algorithm which is optimized and specific to medical imaging data. The encoding algorithm may be an AI or machine learning algorithm or model. ..”; in addition see at least [Page 7, lines 3-4] via: “... for each type of patient data, a model that is suitable for the data type can be built. These model wills convert the data into vectors from different data sources...”)
Nevertheless Liu, Yamamoto, Ribbing and Will are silent the following limitation that is taught by Molero:
wherein a dimension of the first vector representation is reduced with respect to a dimension of the first plurality of datasets. (See at least [0030] via: “... data elements that include images (e.g., MRIs) or image frames of a video (e.g., a video of an ultrasound), each image or image frame may be transformed into a numerical representation (e.g., vector) using a trained auto-encoder neural network, which is trained to generate a latent-space representation of an input image. The condensed representation of the input image (e.g., the latent-space representation) may serve as the vector that numerically represents the input image...”; in addition see at least [0028] via: “...instead of generating a vector to numerically represent each data element of a subject record, techniques may be executed to reduce the dimensionality of the subject record by identifying and selecting a subset of data elements from the set of data elements. The subset of data elements may represent the “important” data elements, where “importance” of a data element is determined based on a prediction using feature extraction techniques, such as Singular Value Decomposition (SVD). For example, transforming a subject record into a transformed representation that is consumable by machine-learning and artificial-intelligence models may include performing one or more feature extraction techniques on the non-numerical values included in the data elements of a subject record to generate a feature vector that numerically represents a decomposed version of the non-numerical values. In some implementations, feature extraction techniques may include, for example, reducing the dimensionality of a set of data elements of a subject record (e.g., each data element representing a feature or dimension of a subject) into an optimal subset of features that can be used to, for example, predict an outcome or event. Reducing the dimensionality of the set of data elements may include reducing N data elements into a subset of M elements, where M is smaller than N. In these implementations, each element of the subset of M elements may be transformed into a numerical value. In some implementations, a feature vector may be generated to represent the N data elements of a subject record. The feature vector may include a vector for each data element of the set of data elements...”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify Liu, Yamamoto, Ribbing and Will with Molero. Liu teaches collating patient data for analysis comprising receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type; and generating a vector for each of the plural patient data records by processing each patient data record using a corresponding encoding algorithm, wherein the encoding algorithm used to generate the vector is selected based on the type of patient data record and wherein the vectors are for use by a machine learning model. However, Liu fails to disclose techniques to reduce the dimensionality of the subject record by identifying and selecting a subset of data elements from the set of data elements, whereby reducing the dimensionality of the set of data elements may include reducing N data elements into a subset of M elements, where M is smaller than N as taught by Molero. Combining Liu and Molero is helpful reducing vector dimensionality which enhances processing by improving computational efficiency, reducing storage costs, and mitigating the "curse of dimensionality". It speeds up machine learning training and inference, reduces noise, and prevents overfitting by focusing on key, informative features, ultimately enhancing model accuracy and performance.
Regarding claim 8: Liu, Yamamoto, Ribbing and Will teach the invention as claimed and detailed above with respect to claim 1. Liu also teaches:
wherein the second machine-learning model was trained by:
inputting a second plurality of datasets to the second machine-learning model, the second plurality of datasets corresponding to the second modality of medical data and utilizing the second machine-learning model to encode the second plurality of datasets into the second vector representation (See at least [Page 2, lines 28-32] via: “...receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type... generating a vector for each of the plural patient data records by processing each patient data record using a corresponding encoding algorithm, wherein the encoding algorithm used to generate the vector is selected based on the type of patient data record and wherein the vectors are for use by a machine learning model...”; in addition see at least [Page 5, lines 6-8] via: “... plurality of patient data records comprises: ... (ii) a second patient data record comprising a non-image medical data comprising a physiological signal..”; in addition see at least [Page 6, lines 43-46 and Page 7, lines 1-2] via: “...generating a second vector 120' for the non-medical imaging data received in the first step 110'. In this case, generating the vector 120' comprises processing each patient data record (i.e. the non-medical imaging data in this case) using an encoding algorithm which is optimized and specific to the specific data type. This is a different encoding algorithm to that used to encode the first vector such that the data extraction is optimized for each data type thereby providing an output that is especially accurate and that key aspects and information in each particular data type is retained and taken into account in subsequent processing...”)
Nevertheless Liu, Yamamoto, Ribbing and Will are silent the following limitation that is taught by Molero:
wherein a dimension of the second vector representation is reduced with respect to a dimension of the second plurality of datasets. (See at least [0028] via: “...instead of generating a vector to numerically represent each data element of a subject record, techniques may be executed to reduce the dimensionality of the subject record by identifying and selecting a subset of data elements from the set of data elements. The subset of data elements may represent the “important” data elements, where “importance” of a data element is determined based on a prediction using feature extraction techniques, such as Singular Value Decomposition (SVD). For example, transforming a subject record into a transformed representation that is consumable by machine-learning and artificial-intelligence models may include performing one or more feature extraction techniques on the non-numerical values included in the data elements of a subject record to generate a feature vector that numerically represents a decomposed version of the non-numerical values. In some implementations, feature extraction techniques may include, for example, reducing the dimensionality of a set of data elements of a subject record (e.g., each data element representing a feature or dimension of a subject) into an optimal subset of features that can be used to, for example, predict an outcome or event. Reducing the dimensionality of the set of data elements may include reducing N data elements into a subset of M elements, where M is smaller than N. In these implementations, each element of the subset of M elements may be transformed into a numerical value. In some implementations, a feature vector may be generated to represent the N data elements of a subject record. The feature vector may include a vector for each data element of the set of data elements...”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify Liu, Yamamoto, Ribbing and Will with Molero. Liu teaches collating patient data for analysis comprising receiving a set of input data comprising a plurality of patient data records, wherein the plural patient data records comprise medical imaging data and at least one other patient data type; and generating a vector for each of the plural patient data records by processing each patient data record using a corresponding encoding algorithm, wherein the encoding algorithm used to generate the vector is selected based on the type of patient data record and wherein the vectors are for use by a machine learning model. However, Liu fails to disclose techniques to reduce the dimensionality of the subject record by identifying and selecting a subset of data elements from the set of data elements, whereby reducing the dimensionality of the set of data elements may include reducing N data elements into a subset of M elements, where M is smaller than N as taught by Molero. Combining Liu and Molero is helpful reducing vector dimensionality which enhances processing by improving computational efficiency, reducing storage costs, and mitigating the "curse of dimensionality". It speeds up machine learning training and inference, reduces noise, and prevents overfitting by focusing on key, informative features, ultimately enhancing model accuracy and performance.
Prior Art Made of Record
The prior art made of record and not relied upon is considered pertinent to Applicant's disclosure, and is listed in the attached form PTO-892 (Notice of References Cited). Unless expressly noted otherwise by the Examiner, all documents listed on form PTO-892 are cited in their entirety.
Kunz (US 20250299802 A1) - SYSTEMS AND METHODS TO PROCESS ELECTRONIC IMAGES FOR DETERMINING TREATMENT - teaches: processing digital pathology images, the method including receiving a plurality of digital pathology images of at least one pathology specimen, the pathology specimen being associated with a patient. The method may further include determining receiving metadata corresponding to the plurality of digital pathology images, the metadata comprising data regarding previous medical treatment of the patient. Next, the method may include providing the medical images and metadata as input to a machine learning system, the machine learning system having been trained by receiving as input historical treatment information and digital images labeled with a predicted treatment regimen. Lastly, the method may include outputting, by the machine learning system, a treatment effectiveness assessment.
Response to Arguments
Applicant's arguments filed 6-24-2026, have been fully considered but not found
persuasive.
Applicant amended independent claims 1, 15 and dependent claims 9-10, 13-14 as posted in the above analysis with additions underlined and deletions as .
In response to applicant's arguments regarding claim rejection under 35 U.S.C § 101.
Several steps are taken in the analysis as to whether an invention is rejected under 101. The first step is to determine if the claim falls within a statutory category. In this case it does for claims 1 and 15 since the claims recite a method and system, for “multi-modal patient representation”.
The second step under 2A prong one is to determine if the claims recite an abstract idea, which would be the case if the invention can be grouped as either: a) mathematical concepts; (b) mental processes; or (c) certain methods of organizing human activity (encompassing (i) fundamental economic principles, (ii) commercial or legal interactions or (iii) managing personal behavior or relationships or interactions between people). The current invention is classified as an abstract idea since it may be grouped to belong to certain methods of organizing human activity under managing personal behavior or relationships or interactions between people as it recites “multi-modal patient representation”..
The third step under 2A Prong Two is to determine if additional elements in the claim imposes a meaningful limit on the abstract idea in order to integrate it into a practical idea. The current invention does not represent a practical idea since the additional elements amount to mere instructions to implement an abstract idea on a computer, or merely use a generic computer as a tool to implement the abstract idea.
the fourth step under 2B is to determine if additional elements of the claim provide an inventive concept. An invention may be classified as an inventive concept if a computer-implemented processes is determined to be significantly more than an abstract idea (and thus eligible), where generic computer components are able in combination to perform functions that are not merely generic, and non-conventional even if generic computer operations on a generic computing device is used to implement the abstract idea. The current invention does not represent an inventive concept since the additional elements amount to mere instructions to implement an abstract idea on a computer, or merely use a generic computer as a tool to implement the abstract idea.
Step 2A Prong ONE
The Applicant argues that independent claims 1 and 15 do not belong to an abstract idea, since it is not directed to a method of organizing human activity. Instead the Applicant argues that the claims describe a technical architecture for processing medical data.
The Examiner disagrees since the Applicant’s argument is not persuasive. The method to select the abstract idea is to strip the additional elements from the claims. As seen below the recited boldened words constitute the abstract idea after stripping the un-boldened additional elements of amended limitation of claims 1 and 15:
accessing a set of medical data associated with a patient, wherein the set of medical data includes a plurality of modalities of medical data, wherein each of the modalities, consists of one data type and is associated with one data source;
inputting one or more of the modalities each having a data type of laboratory testing data into a first machine-learning model trained to generate a first vector representation;
inputting another one of the modalities of medical data, into a second machine- learning model trained to generate a second vector representation of the second modality of medical data, wherein the second modality of medical data consists of a second data type;
generating by a third machine-learning model, a combined vector representation based on the first vector representation and the second vector representation;
storing the combined vector representation to a database
retrieving, based on the patient, the combined vector representation from the database;
identifying, by an ensemble model, a cohort of patients similar to the patient based on a patient similarity inference using the retrieved combined vector representation; and
generating, by the ensemble model, a clinical prediction for the patient based on the retrieved combined vector representation.
The selected abstract idea (boldened limitations) of claims 1 and 15 belong to certain methods of organizing human activity under managing personal behavior or relationships or interactions between people as it recites “multi-modal patient representation”. The application relates to patient data that is often spread across many modalities and sources, such as EHRs, omics datasets, and medical images. These data types differ in scale, sparsity, missingness, and timing, making them difficult to analyze together in a standardized way. The application states that without a unified representation, personalized healthcare tasks such as survival prediction and cohort matching remain difficult. It is in this sense that the application relates to certain methods of organizing human activity under managing personal behavior or relationships or interactions between people. (refer to MPP 2106.04(a)(2)). Accordingly independent claims 1 and 15 recite an abstract idea.
Step 2A Prong TWO
The Applicant argues that even assuming that amended claims 1, 15 recites a judicial exception, the claims covers a technical solution to a technical problem, and thus, as a whole, integrates the recited judicial exception into a practical application of the exception. Specifically, the Applicant argues that claims 1, 15, considered as a whole, covers a particular technical solution to the technical problem of integrating heterogeneous, multi-modal medical data into a combined patient representation suitable for enabling an ensemble model to perform multiple predictions (patient cohort identification and clinical prediction) in a technically efficient manner. Thus, the technical problem identified is the inability of conventional computing systems to integrate heterogeneous, multi-modal medical data into a unified representation suitable for performing multiple personalized healthcare tasks. Claims 1, 15 provide a technical solution to this problem: the system accesses a plurality of modalities of medical data, inputs each modality into a respective machine-learning model to generate a vector representation, combines those representations via a third machine-learning model into a combined vector representation, stores it to a database, and then, after retrieval, uses an ensemble model to both identify a cohort of similar patients based on a patient similarity inference and generate a clinical prediction, each using the same retrieved combined vector representation. This architecture enables the system to perform multiple predictions with a single call to the database, thereby reducing total database calls and decreasing processing time. Thus, the combined vector representation enables an ensemble model to perform two downstream tasks: (1) identifying a cohort of patients similar to the patient, based on a patient similarity inference, and (2) generating a clinical prediction for the patient. Both tasks are performed using the same retrieved combined vector representation, which reduces the number of database calls required and decreases overall processing time. Moreover, as both predictions are more accurate when performed using the combined vector representation than when performed using any individual, disaggregated vector representation. This improvement is recited by claims 1, 15.
Accordingly, the Applicant argues that claims 1, 15 integrate the alleged judicial exception into a practical application because the claimed architecture provides a concrete, technical improvement to machine-learning-based medical data processing.
The Examiner disagrees since the Applicant’s arguments are not persuasive. The Examiner restates that claims 1 & 15 do not integrate the abstract idea into a practical application. Neither claim 1 or 15 recite additional elements that impose a meaningful limit on the abstract idea:
Claims 1, 15 recite:
first machine-learning model trained to generate a first vector representation;
second machine- learning model trained to generate a second vector representation;
generating by a third machine-learning model, a combined vector representation
storing the combined vector representation to a database;
retrieving, the combined vector representation from the database;
identifying, by an ensemble model, a cohort of patients;
generating, by the ensemble model, a clinical prediction.
Claim 15 recites:
one or more non-transitory computer-readable storage media including instructions;
one or more processors coupled to the one or more storage media, the one or more processors configured to execute the instructions and cause the system to access a set of medical data associated with a patient
The additional elements as recited above amount to mere instructions to implement an abstract idea on a computer, or merely use a computer as a tool to implement the abstract idea. (refer to MPEP 2106.05(f)). Support for this can be found in the specification, paragraphs [0009-0017]. Accordingly, the claim as a whole does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
The cited technological advantages argued by the Applicant, specifically that “the system accesses a plurality of modalities of medical data, inputs each modality into a respective machine-learning model to generate a vector representation, combines those representations via a third machine-learning model into a combined vector representation, stores it to a database, and then, after retrieval, uses an ensemble model to both identify a cohort of similar patients based on a patient similarity inference and generate a clinical prediction, each using the same retrieved combined vector representation, thus enabling the system to perform multiple predictions with a single call to the database, thereby reducing total database calls and decreasing processing time” are all accomplished with the mere use of a generic computer as a tool to implement the abstract idea. Hence, claims 1 and 15 as recited are not enough to classify the claims as integrated into a practical application. In order to integrate the abstract idea into a practical application the additional elements should be shown to impose a meaningful limit on the abstract idea which is not the case in this invention. A colloquial interpretation of a practical application as in this case is not enough.
In order to integrate the abstract idea into a practical idea the Applicant could demonstrate at least one of the conditions enumerated below applies:
Improvements to the functioning of a computer, or to any other technology or technical field - see MPEP 2106.05(a)
Applying the judicial exception with, or by use of, a particular machine - see MPEP 2106.05(b)
Effecting a transformation or reduction of a particular article to a different state or thing - see MPEP 2106.05(c)
Applying or using the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort designed to monopolize the exception - see MPEP 2106.05(e) and Vanda Memo
The Applicant has not demonstrated any of the above listed conditions. As a result the Examiner restates the rejection of the invention under 35 USC §101.
Step 2B
Although the Applicant offered no arguments related to inventiveness under Step 2B, The Examiner states, similar to the analysis under Step 2A Prong Two, that the additional elements amount to mere instructions to implement an abstract idea on a computer, or merely use a computer as a tool to implement the abstract idea. (refer to MPEP 2106.05(f)). The use of generic computer components, in combination, do not perform functions that are not merely generic, and non-conventional even if the generic computer operations on a generic computing device is used to implement the abstract idea. Accordingly, the claim does not provide an inventive concept (significantly more than the abstract idea) and hence the claim is ineligible.
In order evaluate whether the claim recites additional elements that amount to an inventive concept what could be shown is:
Adding a specific limitation (unconventional other than what is well-understood, routine, conventional (WURC) activity in the field - see MPEP 2106.05(d)
The Applicant has not demonstrated the above listed condition.
In response to applicant's arguments regarding claim rejection under 35 U.S.C § 101.
The Applicant argues that the following amended limitations of claims 1 and 15 are not taught by Liu or Rahman alone or in combination:
generating by a third machine-learning model, a combined vector representation based on the first vector representation and the second vector representation;
retrieving, based on the patient, the combined vector representation from the database;
identifying, by an ensemble model, a cohort of patients similar to the patient based on a patient similarity inference using the retrieved combined vector representation; and
generating, by the ensemble model, a clinical prediction for the patient based on the retrieved combined vector representation
Therefore, the references fail to disclose or suggest all of the claim limitations of claims 1 and 15.
The applicant's arguments with respect to claim Claims 1, 15 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument
For reasons of record and as set forth above, the examiner maintains the rejection of claims 1-11, 13-15 as being directed to a judicial exception without significantly more, and thereby being directed to non-statutory subject matter under 35 USC §101. In addition claims 1-11, 13-15 are rejected under 35 USC §103. In reaching this decision, the Examiner considered all evidence presented and all arguments
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 PIERRE L MACCAGNO whose telephone number is (571)270-5408. The examiner can normally be reached M-F 8:00 to 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, Mamon Obeid can be reached at (571)270-1813. 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.
/PIERRE L MACCAGNO/Examiner, Art Unit 3687
/KATHERINE KOLOSOWSKI-GAGER/Primary Examiner, Art Unit 3687
Prosecution Insights