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 .
Claims 1-18 are pending and have been examined.
Claim Interpretation
Claims 1 and 10 recite the phrases of “via a machine learning model” and “via at least one computer processor.” For examination purposes, the word “via” is interpreted as “by.” Therefore, the phrases are interpreted as “by a machine learning model” and “by at least one computer processor.”
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 1-18 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claims 1-18 are directed to a system or method, which are statutory categories of invention. (Step 1: YES).
The Examiner has identified method Claim 10 as the claim that represents the claimed invention for analysis and is similar to system Claim 1.
Claim 10 recites the limitations of:
A method for determining structured data, the method comprising:
receiving, in real-time and via at least one computer processor, a plurality of medical concepts and a corresponding plurality of labels, wherein each medical concept from the plurality of medical concepts is associated with a corresponding label from the plurality of labels;
receiving, in real-time and via the at least one computer processor, a plurality of target attributes associated with a plurality of fields of one or more structured data, wherein each field from the plurality of fields is associated with one or more target attributes from the plurality of target attributes;
receiving, in real-time and via the at least one computer processor, a transcript of a conversation between a physician and a patient;
determining, in real-time and via the at least one computer processor, a plurality of transcript concepts in the transcript based on the plurality of medical concepts, where in each transcript concept from the plurality of transcript concepts is a corresponding medical concept from the plurality of medical concepts;
assigning, in real-time and via the at least one computer processor, each transcript concept with the label associated with the corresponding medical concept;
determining, in real-time and via the at least one computer processor, a plurality of concept combinations by combining the plurality of transcript concepts, wherein each concept combination from the plurality of concept combinations is a combination of two or more transcript concepts from the plurality of transcript concepts, such that the labels of the two or more transcript concepts associate with one or more target attributes from the plurality of target attributes of a same field from the plurality of fields;
determining, in real-time and via a machine learning model, a combination label for each concept combination, the combination label being a valid label or an invalid label; and
generating, in real-time and via the at least one computer processor, an output structured data based on the concept combinations having the valid label.
These above limitations, under their broadest reasonable interpretation, cover performance of the limitation as mental processes. The claim recites elements, in non-bold above, which covers performance of the limitation that can be concepts performed in the mind of a person or with pen and paper. A person read and comprehend mentally (receive in real time) medical concepts with labels, read and comprehend (receive in real time) target attributes associated with fields, read and comprehend (receive in real time) a transcript of a conversation between a physician and a patient. A person can analyze in their mind (determining in real time) by reading transcript concepts in a transcript based on medical concepts, assign mentally and with pen and paper each transcript concept with a label corresponding to the medical concept, analyze in their mind (determining in real time) a plurality of concept combinations by combining a plurality of concepts from a plurality of transcript concepts, mentally create (determining) a combination label for concept combinations and analyze (determine) the combination label is valid or invalid label, and with pen and paper generate an output of structured data based on concept combinations having a valid label. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation as a mental process, then it falls within the “Mental Processes” grouping of abstract ideas. Also, see MPEP 2106.04(a)(2) III C where using a generic computer was used in mental processes. Accordingly, the claim recites an abstract idea. Claim 1 is also abstract for similar reasons. (Step 2A-Prong 1: YES. The claims are abstract)
In that the claim recites a conversation between a physician and patient and provide an output to a user, the claim is also abstract as managing interactions or relationships between people, therefore abstract under Certain Methods of Organizing Human Activity.
This judicial exception is not integrated into a practical application. In particular, the claims only recite: computer-readable storage medium, computer processor, machine learning model (Claim 1); computer processor, machine learning model (Claim 10). The computer hardware is recited at a high-level of generality (i.e., as a generic processor performing a generic computer function) such that it amounts no more than mere instructions to apply the exception using a generic computer component. Also, the machine learning model is recited at a high-level of generality. See Applicant’s specification pg. 5, lines 10-17, where machine learning can be various models including decision tree and regression model, which can be performed with pen and paper (a person can create and analyze a decision tree and perform regression analysis). Accordingly, these additional elements, when considered separately and as an ordered combination, do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. Therefore claims 1 and 10 are directed to an abstract idea without a practical application. (Step 2A-Prong 2: NO. The additional claimed elements are not integrated into a practical application)
The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, when considered separately and as an ordered combination, they do not add significantly more (also known as an “inventive concept”) to the exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional element of using a computer hardware amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. Accordingly, these additional elements, when considered separately and as an ordered combination, do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. Steps such as receiving and output (transmitting) are steps that are considered insignificant extra solution activity and mere instructions to apply the exception using general computer components (see MPEP 2106.05(d), II). Thus claims 1 and 10 are not patent eligible. (Step 2B: NO. The claims do not provide significantly more)
Dependent claims 2-9 and 11-18 further define the abstract idea that is present in their respective independent claims 1 and 10 and thus correspond to Mental Processes and Certain Methods of Organizing Human Activity and hence are abstract for the reasons presented above. The dependent claims do not include any additional elements that integrate the abstract idea into a practical application or are sufficient to amount to significantly more than the judicial exception when considered both individually and as an ordered combination. Claims 2 and 11 recite text encoding, tensor, and concatenating at a high level of generality, and tensor is abstract as a mathematical concept. Claims 3 and 12 recite tensor and machine learning at a high level of generality, and tensor is a mathematical concept. Claims 4 and 14 limit machine learning to a binary classifier, which is using existing classifier models at a high level of generality. Claims 5 and 14 recites using training data by a machine learning model at a high level of generality. Claims 6 and 15 recite annotate the transcript which is abstract as a mental process, but also recited at a high level of generality. Claims 7-9 and 16-17 are also abstract as mental processes, further limit abstract concepts and also recite output or highlight data at a high level of generality. Therefore, the claims 2-9 and 11-18 are directed to an abstract idea. Thus, the claims 1-18 are not patent-eligible.
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 1-9 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.
The term “near real-real time” in claim 1 (pg. 1, lines 7-8) is a relative term which renders the claim indefinite. The term “near” is not defined by the claim, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention. Near real-time would be subjectively determined and therefore vary based on various claim interpretations. For examination purposes, this is interpreted as any time.
Claims 2-9 are further rejected as they depend from Claim 1.
Examiner Request
The Applicant is requested to indicate where in the specification there is support for amendments to claims should Applicant amend. The purpose of this is to reduce potential 35 U.S.C. §112(a) or §112 1st paragraph issues that can arise when claims are amended without support in the specification. The Examiner thanks the Applicant in advance.
Claim Rejections - 35 USC § 103
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 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 1, 4-10, and 13-18 are rejected under 35 U.S.C. 103 as being unpatentable over Pub. No. US 2020/0185102 to Leventhal et al. in view of Pub. No. US 2009/0182580 to Martin et al.
Regarding claim 1
A system for determining structured data, the system comprising:
at least one non-transitory computer-readable storage medium having instructions stored thereon; and
Leventhal et al. teaches:
Server (computer) system with instructions stored on memory…
“As shown in FIG. 33, the server system 200 can include one or more additional hardware components as desired. Exemplary additional hardware components include, but are not limited to, a memory 202 (alternatively referred to herein as a non-transitory computer readable medium). Exemplary memory 202 can include, for example, random access memory (RAM), static RAM, dynamic RAM, read-only memory (ROM), programmable ROM, erasable programmable ROM, electrically erasable programmable ROM, flash memory, secure digital (SD) card, and/or the like. Instructions for implementing the server system 200 can be stored on the memory 202 to be executed by the processor 201.” [0383]
at least one computer processor coupled to the at least one non-transitory computer- readable storage medium and configured to execute the instructions, in real-time or near real-time, to:
Instructions executed by the processor…
“As shown in FIG. 33, the server system 200 can include one or more additional hardware components as desired. Exemplary additional hardware components include, but are not limited to, a memory 202 (alternatively referred to herein as a non-transitory computer readable medium). Exemplary memory 202 can include, for example, random access memory (RAM), static RAM, dynamic RAM, read-only memory (ROM), programmable ROM, erasable programmable ROM, electrically erasable programmable ROM, flash memory, secure digital (SD) card, and/or the like. Instructions for implementing the server system 200 can be stored on the memory 202 to be executed by the processor 201.” [0383] Inherent with instructions executed by a processor is instructions executed in real-time.
See Real-time below.
receive a plurality of medical concepts and a corresponding plurality of labels, wherein each medical concept from the plurality of medical concepts is associated with a corresponding label from the plurality of labels;
{
From Applicant’s specification on label (from their Pub. No. US 2025/0308653)…
“In some embodiments, the plurality of labels 114 may also be stored in the storage medium 102. In some embodiments, each label 116 may define a type of the corresponding medical concept 112. For example, “neck” or “knee” would be labeled as a “body structure”, “five” would be labelled as a “number”, etc. Specifically, each label 116 may encode the corresponding medical concept 112.” [0040]
From Applicant’s specification on medical concept…
“The at least one computer processor 104 is configured to execute the instructions to receive a plurality of medical concepts 110 and a corresponding plurality of labels 114. In some embodiments, the plurality of medical concepts 110 include multiple medical concepts 112. In some embodiments, the plurality of labels 114 include multiple labels 116. Each medical concept 112 from the plurality of medical concepts 110 is associated with a corresponding label 116 from the plurality of labels 114. The medical concepts 112 may include, e.g., anatomical locations, signs and symptoms, diagnoses, medications, referrals, investigations and therapies, reasons for visit, severity, location, frequency, time of onset of a symptom entity, etc. Such a vocabulary may be stored in the storage medium 102.” [0039]
Therefore, a label may be body structure, number, etc. (classification) and medical concept may be anatomical locations, diagnoses, etc. Therefore, a label is similar to a genus (body structure, number) that encompasses species (shoulder, five) and medical concept is information/data related to medicine.
So, medical concept associated with a label could be flu or broken arm (medical concepts) associated with virus or body structure (label).
Therefore, label is just a high-level classification/category of a medical concept.
}
Receive notes with symptom, condition, etc. objects (plurality of medical concepts) with tags (plurality of labels)…
“The server system 200 can receive a large sample of doctor-patient notes covering thousands of different conditions. The server system 200 can tag the notes to establish a plurality of objects (or entities, items, or the like). These objects are used to establish an ontology (or Symptom-Attribute-Value (SAV) ontology) 230 that is of a large size and includes the plurality of objects established by tagging. For example, when applied to text notes relating about 200 common conditions, the server system 200 can construct an ontology with 15,000+ medical features. The SAV ontology 230 describes symptoms, their attribute, and all the possible values of all attributes. A particular case vector notes the symptom that is present in the case vector and the value of the attributes of those symptoms. Exemplary objects can include a symptom object, a condition object, a severity object, a time period object, an attribute object, a value object, and so on. The severity object and the time period object can each be an attribute object. Exemplary value objects for the severity object can include ‘mild,’ ‘severe,’ or ‘awaking at night.’ Exemplary value objects for the time period object can include ‘one day,’ ‘one week,’ etc. Additionally and/or alternatively, exemplary objects can include a negation object (e.g., representing a headache without dizziness), a link object (e.g., diarrhea after eating suspicious food), and/or other attributes of symptoms such as location and type (pulsating headache above the left ear), etc. Other examples of objects that can be part of the ontology 230 can include medication object, treatment object, and/or profile features (that is, the features that are not symptom, attribute or value, but that summarize the medical history or personal information). Each profile feature can be associated with a canonical name designated for each ontology object. Exemplary profile feature can refer to some characteristics of a person (such as gender, year of birth, etc), not related to a particular case.” [0066]
receive a plurality of target attributes associated with a plurality of fields of one or more structured data, wherein each field from the plurality of fields is associated with one or more target attributes from the plurality of target attributes;
{
From Applicant’s specification on “target attributes”…
“In some embodiments, the plurality of fields 120 may be associated with the EMR. In some embodiments, each field 122 may be defined by the one or more target attributes 126 associated with it. Specifically, the one or more target attributes 126 may determine data that may be populated in the corresponding field 122. For example, the EMR may include a SOAP note (Subjective, Objective, Assessment, and Plan note). The SOAP note is a widely used format for taking medical notes or summaries. The summaries collected in the SOAP note are digitized and stored in the EMR. The fields 122 associated with the SOAP note may include subjective information (S), objective observations (O), assessments data (A), and plan data (P). Subjective information (S) reported by a patient may include one or more of patient behavior, patient complaint, symptoms, progress from last encounter, problem, medical issues impacting or influencing patient's day-to-day routine, family history, medical history, social history, and so forth.” [0042]
So, target attributes define data that would be contained in a medical record (in a data field of the record). A symptom could be a target attribute and fever temperature data in a field.
Therefore, receiving a medical record is receiving target attributes associated with a field, as most medical records would contain such information.
}
Example of extracting (receiving) of data in fields with hemoglobin or insulin (attributes), where fields are in medical records (field associated with target attributes)…
“4. Above named entity extraction is comprised of: a. Using pre-tagged medical texts with hierarchical SAV ontology objects, and expressing negations, and relations between parts of sentences. Such tagging is performed on a representative subset of the text records. Such tagging will result in a tag database 800 (shown in FIG. 35 and being optionally part of the server system 200) for storing a list of possible tags, or strings linked to the different SAV entities. Two possible ways to tag text are: i. Tagging could be human labeling of specific words in the text itself, from which a hierarchical SAV ontology can be constructed along sample case or visit vectors. ii. Tagging can be achieved by using the structured (non-textual) fields in the electronic medical records accompanying the text, as contextual labels for the texts. For example, with this approach, the SAV of a diabetic patient can be learned by associating with patients that had high levels of Hemoglobin HbA1c or patients that were prescribed with Insulin. Stated somewhat differently, even when a medical condition is not explicitly recorded in the medical notes, the server system 200 can identify that the current patient has a specific medical condition via the structured fields (e.g., lab test). The server system 200 can use the SAV objects from data of other patients with the same medical condition to find the SAV objects to extract for the current patient.” [0149] – [0152]
receive a transcript of a conversation between a physician and a patient;
Run through (receive) doctor-patient notes (transcript of a conversation)…
“As there are many potential Symptom-Attribute-Value (SAV) combinations that can result in a certain condition, artificial intelligence (AI) can be used to weigh the different likelihoods. The server system 200 can be taught the main language and use natural language processing (NLP) to run through millions of doctor-patient notes to start finding these entities in the text. For example, name entity recognition (NER) can be used. Given the complexity of healthcare (HC), a set of heuristic rules based on physician behavior can be used to increase the accuracy of the NER capture.” [0070]
determine a plurality of transcript concepts in the transcript based on the plurality of medical concepts, wherein each transcript concept from the plurality of transcript concepts is a corresponding medical concept from the plurality of medical concepts;
{
From Applicant’s specification on “transcript concepts”…
“Referring to FIGS. 1 and 2, the at least one computer processor 104 is further configured to determine a plurality of transcript concepts 130 in the transcript T based on the plurality of medical concepts 110. The plurality of transcript concepts 130 may include multiple transcript concepts 132-1, 132-2, . . . , 132-N (collectively, transcript concepts 132), where N is a positive integer corresponding to a total number of the transcript concepts 132 in the plurality of transcript concepts 130.”
From Fig. 2 of Applicant’s specification…
PNG
media_image1.png
228
428
media_image1.png
Greyscale
Therefore, a “transcript concept” is just some type of medical concept from a transcript. A shoulder is a medical concept, a range of motion is a medical concept, etc.
}
Construct (determine) ontology of medical features (concepts) from text notes on common conditions (based on medical concepts)…
“The server system 200 can receive a large sample of doctor-patient notes covering thousands of different conditions. The server system 200 can tag the notes to establish a plurality of objects (or entities, items, or the like). These objects are used to establish an ontology (or Symptom-Attribute-Value (SAV) ontology) 230 that is of a large size and includes the plurality of objects established by tagging. For example, when applied to text notes relating about 200 common conditions, the server system 200 can construct an ontology with 15,000+ medical features. The SAV ontology 230 describes symptoms, their attribute, and all the possible values of all attributes. A particular case vector notes the symptom that is present in the case vector and the value of the attributes of those symptoms. Exemplary objects can include a symptom object, a condition object, a severity object, a time period object, an attribute object, a value object, and so on. The severity object and the time period object can each be an attribute object. Exemplary value objects for the severity object can include ‘mild,’ ‘severe,’ or ‘awaking at night.’ Exemplary value objects for the time period object can include ‘one day,’ ‘one week,’ etc. Additionally and/or alternatively, exemplary objects can include a negation object (e.g., representing a headache without dizziness), a link object (e.g., diarrhea after eating suspicious food), and/or other attributes of symptoms such as location and type (pulsating headache above the left ear), etc. Other examples of objects that can be part of the ontology 230 can include medication object, treatment object, and/or profile features (that is, the features that are not symptom, attribute or value, but that summarize the medical history or personal information). Each profile feature can be associated with a canonical name designated for each ontology object. Exemplary profile feature can refer to some characteristics of a person (such as gender, year of birth, etc), not related to a particular case.” [0066]
assign each transcript concept with the label associated with the corresponding medical concept;
{
This would just be linking the medical concept (provided by a transcript) to a high-level classification/category (label).
Therefore, shoulder (transcript concept) could be assigned to body structure (label).
}
Ontology (transcript concept) with tagging (labeling)…
“The server system 200 can receive a large sample of doctor-patient notes covering thousands of different conditions. The server system 200 can tag the notes to establish a plurality of objects (or entities, items, or the like). These objects are used to establish an ontology (or Symptom-Attribute-Value (SAV) ontology) 230 that is of a large size and includes the plurality of objects established by tagging. For example, when applied to text notes relating about 200 common conditions, the server system 200 can construct an ontology with 15,000+ medical features. The SAV ontology 230 describes symptoms, their attribute, and all the possible values of all attributes. A particular case vector notes the symptom that is present in the case vector and the value of the attributes of those symptoms. Exemplary objects can include a symptom object, a condition object, a severity object, a time period object, an attribute object, a value object, and so on. The severity object and the time period object can each be an attribute object. Exemplary value objects for the severity object can include ‘mild,’ ‘severe,’ or ‘awaking at night.’ Exemplary value objects for the time period object can include ‘one day,’ ‘one week,’ etc. Additionally and/or alternatively, exemplary objects can include a negation object (e.g., representing a headache without dizziness), a link object (e.g., diarrhea after eating suspicious food), and/or other attributes of symptoms such as location and type (pulsating headache above the left ear), etc. Other examples of objects that can be part of the ontology 230 can include medication object, treatment object, and/or profile features (that is, the features that are not symptom, attribute or value, but that summarize the medical history or personal information). Each profile feature can be associated with a canonical name designated for each ontology object. Exemplary profile feature can refer to some characteristics of a person (such as gender, year of birth, etc), not related to a particular case.” [0066]
Another example of tagging (labeling)…
“3. Performing automated NLP based feature extraction (for example, named entity extraction) using medical ontology to provide the entity space from the textual parts of the data, such as: a. Doctor visit notes and summary b. Hospital discharge summary 4. Above named entity extraction is comprised of: a. Using pre-tagged medical texts with hierarchical SAV ontology objects, and expressing negations, and relations between parts of sentences. Such tagging is performed on a representative subset of the text records. Such tagging will result in a tag database 800 (shown in FIG. 35 and being optionally part of the server system 200) for storing a list of possible tags, or strings linked to the different SAV entities. Two possible ways to tag text are: i. Tagging could be human labeling of specific words in the text itself, from which a hierarchical SAV ontology can be constructed along sample case or visit vectors. ii. Tagging can be achieved by using the structured (non-textual) fields in the electronic medical records accompanying the text, as contextual labels for the texts. For example, with this approach, the SAV of a diabetic patient can be learned by associating with patients that had high levels of Hemoglobin HbA1c or patients that were prescribed with Insulin. Stated somewhat differently, even when a medical condition is not explicitly recorded in the medical notes, the server system 200 can identify that the current patient has a specific medical condition via the structured fields (e.g., lab test). The server system 200 can use the SAV objects from data of other patients with the same medical condition to find the SAV objects to extract for the current patient. b. Building, from the tagged examples, an ontology of symptoms (e.g., headache, abdominal pain), their attribute (e.g., severity, when did it start, where does it hurt, etc.) and the attribute values (severity=severe, started 3 days ago, etc.) c. Automatically applying named entity extraction on non-tagged text notes, applying several NLP algorithms such as semantic and phrase analysis, edit distance matching to the ontology objects to extract the ontology objects from the non-tagged texts generating feature vectors per user visit. 5. Building, as part of NLP and/or manually, ontologies for describing demographic, medical history, lab tests, medications to standardize all features of a person/medical case. 6. Combining such ontologies with industry standard ontologies (for example, medications based on Anatomical Therapeutic Chemical (ATC), National Drug Code (NDC), and RxNORM). 7. Translating such ontologies, and their corresponding tag vectors to be easily used in extraction of SAV values into case vectors from doctor notes in different languages.” [0146] – [0157]
determine a plurality of concept combinations by combining the plurality of transcript concepts, wherein each concept combination from the plurality of concept combinations is a combination of two or more transcript concepts from the plurality of transcript concepts, such that the labels of the two or more transcript concepts associate with one or more target attributes from the plurality of target attributes of a same field from the plurality of fields;
{
Therefore, the above could be…
Combine shoulder + range of motion (transcript/medical concepts);
Body structure + percent/number (labels or classification/category of transcript concepts);
Associate labels (body structure + percent/number) with…
Patient evaluation such as objective observation (target attributes) where shoulder has range of motion 5% (objective observation, therefore target attributes)
}
Combined symptom and value objects, where headache is a same field for target attributes of pain and head, also unilateral-left for left side (therefore, plurality of concept combinations)…
“Following those tagging iterations, a “post-tagging” process is run in a semi manual mode where the end result is a combined “symptom, attribute, value OBJECTS model”—in which a canonical term is used to group multiple tags. For example, the symptom tag in example 4 above (“pain head”) is grouped with the symptom “headache” from example 1 to form one canonical symptom object named “Headache”. Similarly, the value tag “left side” from example 4 is grouped with the value tag “left” from example 2 to form one canonical value object named “unilateral—left”. Such canonization can allow training of the machine learning model 210 (shown in FIG. 3A).” [0273]
determine, via a machine learning model, a combination label for each concept combination, the combination label being a valid label or an invalid label; and
{
Combination label could be body structure (e.g., arm) with motion (rotate).
}
Combined tags “unilateral-left” (combination label) with machine learning…
“Following those tagging iterations, a “post-tagging” process is run in a semi manual mode where the end result is a combined “symptom, attribute, value OBJECTS model”—in which a canonical term is used to group multiple tags. For example, the symptom tag in example 4 above (“pain head”) is grouped with the symptom “headache” from example 1 to form one canonical symptom object named “Headache”. Similarly, the value tag “left side” from example 4 is grouped with the value tag “left” from example 2 to form one canonical value object named “unilateral—left”. Such canonization can allow training of the machine learning model 210 (shown in FIG. 3A).” [0273]
Example of training with attributes and label and using machine learning and error (valid or invalid)…
“Turning to FIG. 12B, an exemplary diagram illustrating another alternative embodiment of a process of machine learning is shown. Training data 401 can include a plurality of case vectors 228. Each case vector 228 can combine a feature vector 224 (for example, including profile features such as history, and symptoms, attributes) and a label 226. The label 226 can be any suitable category labels for model training. Exemplary labels 226 can include medical insights, including condition diagnosis, for example. A selected case vector 228 can be inputted into the model 210 being trained, to output a model prediction 412 of a current iteration of model training. An error 414 can be based on a difference between the label 226 and the model prediction 412. For example, the error 414 can be equal to (label 226-model prediction 412).” [0306]
generate an output structured data based on the concept combinations having the valid label.
{
From Applicant’s specification on “structured data”…
“In recent years, extensive research and development effort in healthcare industry has been focused on automating a process of generating the clinical note based on doctor-patient conversations (audio or transcript). Typically, in case of doctor-patient conversation transcripts, pieces of information (e.g., words, phrases, etc.) may need to be extracted from the transcript. Such pieces of information may be combined with other pieces to form a target structed data. Structured data is data organized into specific fields (or categories) as part of a schema, with each field having a defined purpose. Examples of such fields may include numbers, anatomical structures, laterality, etc. Structured data categories may include patient history, family history, past surgeries, medications, exam results, vital signs, and more.” [0003]
Therefore a “field” can be a category, such as numbers, anatomical structures, etc.
Structured data is data organized into categories as part of a schema (outline/model), each category has a defined purpose.
}
Example of combined vector with category label and used as input, therefore, output the vector (structured data)…
“Turning to FIG. 12B, an exemplary diagram illustrating another alternative embodiment of a process of machine learning is shown. Training data 401 can include a plurality of case vectors 228. Each case vector 228 can combine a feature vector 224 (for example, including profile features such as history, and symptoms, attributes) and a label 226. The label 226 can be any suitable category labels for model training. Exemplary labels 226 can include medical insights, including condition diagnosis, for example. A selected case vector 228 can be inputted into the model 210 being trained, to output a model prediction 412 of a current iteration of model training. An error 414 can be based on a difference between the label 226 and the model prediction 412.
Real-time
Leventhal et al. teaches server and processor to perform computerized functions of providing health information. They do not literally teach real-time.
Martin et al. also in the business of providing health information teaches:
Real-time processing….
“In some embodiments, the data aggregator 110 and/or dissemination engine 120 process the patient data in real-time. The real-time processing by the data aggregator 110 of some embodiments includes parsing and tagging the collected data as the data aggregator 110 receives new patient data. For instance, the data aggregator 110 regularly pulls data from patient monitoring devices. Alternatively, the data aggregator 110 may operate in an on-demand basis, whereby whenever new data is submitted to the data aggregator 110, the data aggregator 110 performs the parsing and tagging of the data. The real-time processing by the dissemination engine 120 of some embodiments includes generating the various data tuples based on functions performed by each receiving destination immediately after the data aggregator 110 tags the data. The dissemination engine 120 either directly receives the data from the data aggregator 110 in real-time or directly retrieves the data from the data warehouse 130 in real-time. For example, as the data aggregator 110 populates the data warehouse 130, a flag is set whereby the dissemination engine 120 identifies the newly collected data. Additionally, the real-time processing by the dissemination engine 120 includes disseminating the various data tuples to the corresponding destinations once the data tuples are generated.” [0034]
“In some embodiments, the aggregated data for a patient includes data components of a subjective, objective, assessment, plan (SOAP) note. FIG. 2 presents a data record containing data components of a SOAP note in accordance with some embodiments. The various data components 220 of the data record 210 are automatically populated using the data aggregated by the data aggregator 110. It should be apparent to one of ordinary skill in the art that a data record may include a conceptual data record that is distributed across multiple tables of the data warehouse 130.” [0038]
It would have been obvious to one of ordinary skill in the art before the effective filing date to include in the method and system of Leventhal et al. the ability to perform real-time processing as taught by Martin et al. since the claimed invention is merely a combination of old elements and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. Further motivation is provided by Martin et al. who teaches the advantages of real time processing and Leventhal et al. benefits as they teach computers for processing information, and real-time processing allows for up to the minute analysis of medical notes.
Regarding claim 10
A method for determining structured data, the method comprising:
receiving, in real-time and via at least one computer processor, a plurality of medical concepts and a corresponding plurality of labels, wherein each medical concept from the plurality of medical concepts is associated with a corresponding label from the plurality of labels;
{
From Applicant’s specification on label (from Pub. No. US 20250308653)…
“In some embodiments, the plurality of labels 114 may also be stored in the storage medium 102. In some embodiments, each label 116 may define a type of the corresponding medical concept 112. For example, “neck” or “knee” would be labeled as a “body structure”, “five” would be labelled as a “number”, etc. Specifically, each label 116 may encode the corresponding medical concept 112.” [0040]
From Applicant’s specification on medical concept…
“The at least one computer processor 104 is configured to execute the instructions to receive a plurality of medical concepts 110 and a corresponding plurality of labels 114. In some embodiments, the plurality of medical concepts 110 include multiple medical concepts 112. In some embodiments, the plurality of labels 114 include multiple labels 116. Each medical concept 112 from the plurality of medical concepts 110 is associated with a corresponding label 116 from the plurality of labels 114. The medical concepts 112 may include, e.g., anatomical locations, signs and symptoms, diagnoses, medications, referrals, investigations and therapies, reasons for visit, severity, location, frequency, time of onset of a symptom entity, etc. Such a vocabulary may be stored in the storage medium 102.” [0039]
Therefore, a label may be body structure, number, etc. (classification) and medical concept may be anatomical locations, diagnoses, etc. Therefore, a label is similar to a genus (body structure, number) that encompasses species (shoulder, five) and medical concept is information/data related to medicine.
So, medical concept associated with a label could be flu or broken arm (medical concepts) associated with virus or body structure (label).
Therefore, label is just a high-level classification/category of a medical concept.
}
Leventhal et al. teaches:
Receive notes with symptom, condition, etc. objects (medical concepts) with tags (labels)…
“The server system 200 can receive a large sample of doctor-patient notes covering thousands of different conditions. The server system 200 can tag the notes to establish a plurality of objects (or entities, items, or the like). These objects are used to establish an ontology (or Symptom-Attribute-Value (SAV) ontology) 230 that is of a large size and includes the plurality of objects established by tagging. For example, when applied to text notes relating about 200 common conditions, the server system 200 can construct an ontology with 15,000+ medical features. The SAV ontology 230 describes symptoms, their attribute, and all the possible values of all attributes. A particular case vector notes the symptom that is present in the case vector and the value of the attributes of those symptoms. Exemplary objects can include a symptom object, a condition object, a severity object, a time period object, an attribute object, a value object, and so on. The severity object and the time period object can each be an attribute object. Exemplary value objects for the severity object can include ‘mild,’ ‘severe,’ or ‘awaking at night.’ Exemplary value objects for the time period object can include ‘one day,’ ‘one week,’ etc. Additionally and/or alternatively, exemplary objects can include a negation object (e.g., representing a headache without dizziness), a link object (e.g., diarrhea after eating suspicious food), and/or other attributes of symptoms such as location and type (pulsating headache above the left ear), etc. Other examples of objects that can be part of the ontology 230 can include medication object, treatment object, and/or profile features (that is, the features that are not symptom, attribute or value, but that summarize the medical history or personal information). Each profile feature can be associated with a canonical name designated for each ontology object. Exemplary profile feature can refer to some characteristics of a person (such as gender, year of birth, etc), not related to a particular case.” [0066]
See Real-time below.
receiving, in real-time and via the at least one computer processor, a plurality of target attributes associated with a plurality of fields of one or more structured data, wherein each field from the plurality of fields is associated with one or more target attributes from the plurality of target attributes;
{
From Applicant’s specification on “target attributes”…
“In some embodiments, the plurality of fields 120 may be associated with the EMR. In some embodiments, each field 122 may be defined by the one or more target attributes 126 associated with it. Specifically, the one or more target attributes 126 may determine data that may be populated in the corresponding field 122. For example, the EMR may include a SOAP note (Subjective, Objective, Assessment, and Plan note). The SOAP note is a widely used format for taking medical notes or summaries. The summaries collected in the SOAP note are digitized and stored in the EMR. The fields 122 associated with the SOAP note may include subjective information (S), objective observations (O), assessments data (A), and plan data (P). Subjective information (S) reported by a patient may include one or more of patient behavior, patient complaint, symptoms, progress from last encounter, problem, medical issues impacting or influencing patient's day-to-day routine, family history, medical history, social history, and so forth.” [0042]
So target attributes defines data that would be contained in a medical record (in a data field of the record). A symptom could be a target attribute and fever temperature data in a field.
Therefore, receiving a medical record is receiving target attributes associated with a field.
}
Example of extracting (receiving) of data in fields with hemoglobin or insulin (attributes), where fields are in medical records (field associated with target)…
“4. Above named entity extraction is comprised of: a. Using pre-tagged medical texts with hierarchical SAV ontology objects, and expressing negations, and relations between parts of sentences. Such tagging is performed on a representative subset of the text records. Such tagging will result in a tag database 800 (shown in FIG. 35 and being optionally part of the server system 200) for storing a list of possible tags, or strings linked to the different SAV entities. Two possible ways to tag text are: i. Tagging could be human labeling of specific words in the text itself, from which a hierarchical SAV ontology can be constructed along sample case or visit vectors. ii. Tagging can be achieved by using the structured (non-textual) fields in the electronic medical records accompanying the text, as contextual labels for the texts. For example, with this approach, the SAV of a diabetic patient can be learned by associating with patients that had high levels of Hemoglobin HbA1c or patients that were prescribed with Insulin. Stated somewhat differently, even when a medical condition is not explicitly recorded in the medical notes, the server system 200 can identify that the current patient has a specific medical condition via the structured fields (e.g., lab test). The server system 200 can use the SAV objects from data of other patients with the same medical condition to find the SAV objects to extract for the current patient.” [0149] – [0152]
See Real-time below.
receiving, in real-time and via the at least one computer processor, a transcript of a conversation between a physician and a patient;
Run through (receive) doctor-patient notes (conversation)…
“As there are many potential Symptom-Attribute-Value (SAV) combinations that can result in a certain condition, artificial intelligence (AI) can be used to weigh the different likelihoods. The server system 200 can be taught the main language and use natural language processing (NLP) to run through millions of doctor-patient notes to start finding these entities in the text. For example, name entity recognition (NER) can be used. Given the complexity of healthcare (HC), a set of heuristic rules based on physician behavior can be used to increase the accuracy of the NER capture.” [0070]
See Real-time below.
determining, in real-time and via the at least one computer processor, a plurality of transcript concepts in the transcript based on the plurality of medical concepts, where in each transcript concept from the plurality of transcript concepts is a corresponding medical concept from the plurality of medical concepts;
{
From Applicant’s specification on “transcript concepts”…
“Referring to FIGS. 1 and 2, the at least one computer processor 104 is further configured to determine a plurality of transcript concepts 130 in the transcript T based on the plurality of medical concepts 110. The plurality of transcript concepts 130 may include multiple transcript concepts 132-1, 132-2, . . . , 132-N (collectively, transcript concepts 132), where N is a positive integer corresponding to a total number of the transcript concepts 132 in the plurality of transcript concepts 130.”
From Fig. 2…
PNG
media_image1.png
228
428
media_image1.png
Greyscale
Therefore, a “transcript concept” is just some type of medical concept from a transcript.
}
Construct (determine) ontology of medical features (concepts)…
“The server system 200 can receive a large sample of doctor-patient notes covering thousands of different conditions. The server system 200 can tag the notes to establish a plurality of objects (or entities, items, or the like). These objects are used to establish an ontology (or Symptom-Attribute-Value (SAV) ontology) 230 that is of a large size and includes the plurality of objects established by tagging. For example, when applied to text notes relating about 200 common conditions, the server system 200 can construct an ontology with 15,000+ medical features. The SAV ontology 230 describes symptoms, their attribute, and all the possible values of all attributes. A particular case vector notes the symptom that is present in the case vector and the value of the attributes of those symptoms. Exemplary objects can include a symptom object, a condition object, a severity object, a time period object, an attribute object, a value object, and so on. The severity object and the time period object can each be an attribute object. Exemplary value objects for the severity object can include ‘mild,’ ‘severe,’ or ‘awaking at night.’ Exemplary value objects for the time period object can include ‘one day,’ ‘one week,’ etc. Additionally and/or alternatively, exemplary objects can include a negation object (e.g., representing a headache without dizziness), a link object (e.g., diarrhea after eating suspicious food), and/or other attributes of symptoms such as location and type (pulsating headache above the left ear), etc. Other examples of objects that can be part of the ontology 230 can include medication object, treatment object, and/or profile features (that is, the features that are not symptom, attribute or value, but that summarize the medical history or personal information). Each profile feature can be associated with a canonical name designated for each ontology object. Exemplary profile feature can refer to some characteristics of a person (such as gender, year of birth, etc), not related to a particular case.” [0066]
See Real-time below.
assigning, in real-time and via the at least one computer processor, each transcript concept with the label associated with the corresponding medical concept;
{
This would just be linking the medical concept (provided by a transcript) to a high-level classification/category (label).
Therefore, shoulder (transcript concept) could be assigned to body structure (label).
}
Ontology (transcript concept) with tagging (labeling)…
“The server system 200 can receive a large sample of doctor-patient notes covering thousands of different conditions. The server system 200 can tag the notes to establish a plurality of objects (or entities, items, or the like). These objects are used to establish an ontology (or Symptom-Attribute-Value (SAV) ontology) 230 that is of a large size and includes the plurality of objects established by tagging. For example, when applied to text notes relating about 200 common conditions, the server system 200 can construct an ontology with 15,000+ medical features. The SAV ontology 230 describes symptoms, their attribute, and all the possible values of all attributes. A particular case vector notes the symptom that is present in the case vector and the value of the attributes of those symptoms. Exemplary objects can include a symptom object, a condition object, a severity object, a time period object, an attribute object, a value object, and so on. The severity object and the time period object can each be an attribute object. Exemplary value objects for the severity object can include ‘mild,’ ‘severe,’ or ‘awaking at night.’ Exemplary value objects for the time period object can include ‘one day,’ ‘one week,’ etc. Additionally and/or alternatively, exemplary objects can include a negation object (e.g., representing a headache without dizziness), a link object (e.g., diarrhea after eating suspicious food), and/or other attributes of symptoms such as location and type (pulsating headache above the left ear), etc. Other examples of objects that can be part of the ontology 230 can include medication object, treatment object, and/or profile features (that is, the features that are not symptom, attribute or value, but that summarize the medical history or personal information). Each profile feature can be associated with a canonical name designated for each ontology object. Exemplary profile feature can refer to some characteristics of a person (such as gender, year of birth, etc), not related to a particular case.” [0066]
Another example of tagging (labeling)…
“3. Performing automated NLP based feature extraction (for example, named entity extraction) using medical ontology to provide the entity space from the textual parts of the data, such as: a. Doctor visit notes and summary b. Hospital discharge summary 4. Above named entity extraction is comprised of: a. Using pre-tagged medical texts with hierarchical SAV ontology objects, and expressing negations, and relations between parts of sentences. Such tagging is performed on a representative subset of the text records. Such tagging will result in a tag database 800 (shown in FIG. 35 and being optionally part of the server system 200) for storing a list of possible tags, or strings linked to the different SAV entities. Two possible ways to tag text are: i. Tagging could be human labeling of specific words in the text itself, from which a hierarchical SAV ontology can be constructed along sample case or visit vectors. ii. Tagging can be achieved by using the structured (non-textual) fields in the electronic medical records accompanying the text, as contextual labels for the texts. For example, with this approach, the SAV of a diabetic patient can be learned by associating with patients that had high levels of Hemoglobin HbA1c or patients that were prescribed with Insulin. Stated somewhat differently, even when a medical condition is not explicitly recorded in the medical notes, the server system 200 can identify that the current patient has a specific medical condition via the structured fields (e.g., lab test). The server system 200 can use the SAV objects from data of other patients with the same medical condition to find the SAV objects to extract for the current patient. b. Building, from the tagged examples, an ontology of symptoms (e.g., headache, abdominal pain), their attribute (e.g., severity, when did it start, where does it hurt, etc.) and the attribute values (severity=severe, started 3 days ago, etc.) c. Automatically applying named entity extraction on non-tagged text notes, applying several NLP algorithms such as semantic and phrase analysis, edit distance matching to the ontology objects to extract the ontology objects from the non-tagged texts generating feature vectors per user visit. 5. Building, as part of NLP and/or manually, ontologies for describing demographic, medical history, lab tests, medications to standardize all features of a person/medical case. 6. Combining such ontologies with industry standard ontologies (for example, medications based on Anatomical Therapeutic Chemical (ATC), National Drug Code (NDC), and RxNORM). 7. Translating such ontologies, and their corresponding tag vectors to be easily used in extraction of SAV values into case vectors from doctor notes in different languages.” [0146] – [0157]
See Real-time below.
determining, in real-time and via the at least one computer processor, a plurality of concept combinations by combining the plurality of transcript concepts, wherein each concept combination from the plurality of concept combinations is a combination of two or more transcript concepts from the plurality of transcript concepts, such that the labels of the two or more transcript concepts associate with one or more target attributes from the plurality of target attributes of a same field from the plurality of fields;
{
Combine shoulder + range of motion (transcript/medical concepts);
Body structure + percent/number (labels or classification/category of transcript concepts);
Associate labels (body structure + percent/number) with…
Patient evaluation such as objective observation (target attributes) where shoulder has range of motion 5% (objective observation, therefore target attributes)
}
Combined symptom and value objects, where headache is a same field for target attributes of pain and head…
“Following those tagging iterations, a “post-tagging” process is run in a semi manual mode where the end result is a combined “symptom, attribute, value OBJECTS model”—in which a canonical term is used to group multiple tags. For example, the symptom tag in example 4 above (“pain head”) is grouped with the symptom “headache” from example 1 to form one canonical symptom object named “Headache”. Similarly, the value tag “left side” from example 4 is grouped with the value tag “left” from example 2 to form one canonical value object named “unilateral—left”. Such canonization can allow training of the machine learning model 210 (shown in FIG. 3A).” [0273]
See Real-time below.
determining, in real-time and via a machine learning model, a combination label for each concept combination, the combination label being a valid label or an invalid label; and
{
Combination label could be body structure (e.g., arm) with motion (rotate).
}
Combined tags “unilateral-left” (combination label) with machine learning…
“Following those tagging iterations, a “post-tagging” process is run in a semi manual mode where the end result is a combined “symptom, attribute, value OBJECTS model”—in which a canonical term is used to group multiple tags. For example, the symptom tag in example 4 above (“pain head”) is grouped with the symptom “headache” from example 1 to form one canonical symptom object named “Headache”. Similarly, the value tag “left side” from example 4 is grouped with the value tag “left” from example 2 to form one canonical value object named “unilateral—left”. Such canonization can allow training of the machine learning model 210 (shown in FIG. 3A).” [0273]
Example of training with attributes and label and using machine learning and error (valid or invalid)…
“Turning to FIG. 12B, an exemplary diagram illustrating another alternative embodiment of a process of machine learning is shown. Training data 401 can include a plurality of case vectors 228. Each case vector 228 can combine a feature vector 224 (for example, including profile features such as history, and symptoms, attributes) and a label 226. The label 226 can be any suitable category labels for model training. Exemplary labels 226 can include medical insights, including condition diagnosis, for example. A selected case vector 228 can be inputted into the model 210 being trained, to output a model prediction 412 of a current iteration of model training. An error 414 can be based on a difference between the label 226 and the model prediction 412. For example, the error 414 can be equal to (label 226-model prediction 412).” [0306] Inherent with machine learning is real-time processing.
See Real-time below.
generating, in real-time and via the at least one computer processor, an output structured data based on the concept combinations having the valid label.
{
From Applicant’s specification on “structured data”…
“In recent years, extensive research and development effort in healthcare industry has been focused on automating a process of generating the clinical note based on doctor-patient conversations (audio or transcript). Typically, in case of doctor-patient conversation transcripts, pieces of information (e.g., words, phrases, etc.) may need to be extracted from the transcript. Such pieces of information may be combined with other pieces to form a target structed data. Structured data is data organized into specific fields (or categories) as part of a schema, with each field having a defined purpose. Examples of such fields may include numbers, anatomical structures, laterality, etc. Structured data categories may include patient history, family history, past surgeries, medications, exam results, vital signs, and more.” [0003]
Therefore a “field” can be a category, such as numbers, anatomical structures, etc.
Structured data is data organized into categories as part of a schema (outline/model), each category has a defined purpose.
}
Example of combined vector with category label and used as input, therefore, output the vector (structured data)…
“Turning to FIG. 12B, an exemplary diagram illustrating another alternative embodiment of a process of machine learning is shown. Training data 401 can include a plurality of case vectors 228. Each case vector 228 can combine a feature vector 224 (for example, including profile features such as history, and symptoms, attributes) and a label 226. The label 226 can be any suitable category labels for model training. Exemplary labels 226 can include medical insights, including condition diagnosis, for example. A selected case vector 228 can be inputted into the model 210 being trained, to output a model prediction 412 of a current iteration of model training. An error 414 can be based on a difference between the label 226 and the model prediction 412. For example, the error 414 can be equal to (label 226-model prediction 412).” [0306]
See Real-time below.
Real-time
Leventhal et al. teaches server and processor to perform computerized functions of providing health information. They do not literally teach real-time.
Martin et al. also in the business of providing health information teaches:
Real-time processing….
“In some embodiments, the data aggregator 110 and/or dissemination engine 120 process the patient data in real-time. The real-time processing by the data aggregator 110 of some embodiments includes parsing and tagging the collected data as the data aggregator 110 receives new patient data. For instance, the data aggregator 110 regularly pulls data from patient monitoring devices. Alternatively, the data aggregator 110 may operate in an on-demand basis, whereby whenever new data is submitted to the data aggregator 110, the data aggregator 110 performs the parsing and tagging of the data. The real-time processing by the dissemination engine 120 of some embodiments includes generating the various data tuples based on functions performed by each receiving destination immediately after the data aggregator 110 tags the data. The dissemination engine 120 either directly receives the data from the data aggregator 110 in real-time or directly retrieves the data from the data warehouse 130 in real-time. For example, as the data aggregator 110 populates the data warehouse 130, a flag is set whereby the dissemination engine 120 identifies the newly collected data. Additionally, the real-time processing by the dissemination engine 120 includes disseminating the various data tuples to the corresponding destinations once the data tuples are generated.” [0034]
“In some embodiments, the aggregated data for a patient includes data components of a subjective, objective, assessment, plan (SOAP) note. FIG. 2 presents a data record containing data components of a SOAP note in accordance with some embodiments. The various data components 220 of the data record 210 are automatically populated using the data aggregated by the data aggregator 110. It should be apparent to one of ordinary skill in the art that a data record may include a conceptual data record that is distributed across multiple tables of the data warehouse 130.” [0038]
It would have been obvious to one of ordinary skill in the art before the effective filing date to include in the method and system of Leventhal et al. the ability to perform real-time processing as taught by Martin et al. since the claimed invention is merely a combination of old elements and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. Further motivation is provided by Martin et al. who teaches the advantages of real time processing and Leventhal et al. benefits as they teach computers for processing information, and real-time processing allows for up to the minute analysis of medical notes.
Regarding claims 4 and 13
(claim 4) The system of claim 1, wherein the machine learning model is a binary classifier model.
Leventhal et al. teaches:
Machine learning classification…
“Turning to FIG. 11A, an exemplary diagram illustrating an embodiment of a process of machine learning is shown. The classification server (or condition classifier) 268 (shown in FIG. 3C) can include one or more classifiers. When using more than one classifier, there is an additional process that implements some “voting” or other decision heuristic to pick the best classification result.” [0295]
Where classifiers can be binary…
“Classifiers can be binary—such a classifier considers one class at a time and provide the probability of the feature vector as belonging to this class. Multi-class classifiers work on a collection of classes, and produce a probability vector for this collection, where each item in this vector is a number between 0 and 1 that can be interpreted as representing the probability of the feature vector to belong to this class. The sum of all those probabilities is normalized to 1.” [0297]
Regarding claims 5 and 14
(claim 5) The system of claim 1, wherein the at least one computer processor is further configured to train the machine learning model using a training data comprising a training transcript, a plurality of training concept combinations, and a plurality of training combination labels corresponding to the plurality of training concept combinations.
Leventhal et al. teaches:
Training data with doctor notes (transcript)…
“Turning to FIG. 3B, a more detailed view of the training system 240 is shown. The training system 240 can include one or more integrated health records 250A. The integrated health records 250A can include structured data 251A (including, for example, age, gender, lab results, diagnosis, etc.), and unstructured data 252A (including, for example, text data such as doctor notes). Optionally, the training system 240 can include one or more integrated health records 250B that is from a source that is different from the source of the integrated health records 250A, and can include structured data 251B (similar to structured data 251A) and/or unstructured data 252B (similar to unstructured data 252A).” [0059]
Training models using concept of reference population with combined medical symptoms and attributes (concepts)…
“12. Applying calculation (e.g., using Equation(1)) on the trained model to use it to extract a relevant sub model based on a reference population. The concept of reference population can be dynamic and based on both demographic information as well as particular combination of current symptoms, past medical history, medications taken, lab results, etc.” [0110]
“13. Predicting possible conditions based on user demographic, medical history, current symptoms and their attributes.” [0111]
Combined tags (labels) and training….
“Following those tagging iterations, a “post-tagging” process is run in a semi manual mode where the end result is a combined “symptom, attribute, value OBJECTS model”—in which a canonical term is used to group multiple tags. For example, the symptom tag in example 4 above (“pain head”) is grouped with the symptom “headache” from example 1 to form one canonical symptom object named “Headache”. Similarly, the value tag “left side” from example 4 is grouped with the value tag “left” from example 2 to form one canonical value object named “unilateral—left”. Such canonization can allow training of the machine learning model 210 (shown in FIG. 3A).” [0273]
Regarding claims 6 and 15
(claim 6) The system of claim 1, wherein the at least one computer processor is further configured to annotate the transcript with the plurality of transcript concepts.
Leventhal et al. teaches:
Marks (annotates) texts and format a tag…
“As shown in FIG. 7, a person tagging the text (or a human tagger, usually a medical doctor, medical student, etc.) marks texts as being a symptom (throat ache), attribute (severity) and value (light), forming a symptom tag 221 and an attribute tag 222 integrated with a value tag 223. Note that negative symptoms (that is, symptoms that the patient does not suffer from) are market as negative (e.g., no fever).” [0259]
Auto Tagger to mark-up records…
“Once the first subset of records is used for manual tagging (at least few thousand records), an automated process (“Auto Tagger”) can take place and cover a much larger set of visit records. The Auto Tagger uses the tagged visit records database, having symptom tags 221, attribute tags 222 and value tags 223, and use those to search the text fields of a much larger set of (non-tagged) visit records.” [0260]
Regarding claims 7 and 16
(claim 7) The system of claim 1, wherein the at least one computer processor is further configured to output the output structured data to at least one of a user interface and the at least one non-transitory computer-readable storage medium.
Leventhal et al. teaches:
Example of database (storage medium) for storing objects in tables (structured data)…
“Turning to FIG. 9B, an exemplary diagram illustrating an alternative embodiment of the tagging process is shown. The SAV objects of the objects in the ontology 230 (shown in FIG. 3C) can be stored in a relational database or a graph database. In a Graph database, the relationship between a symptom, possible attributes thereof, and the possible values for the attributes can be represented naturally in a graph. In one embodiment, The SAV objects can be stored in a relational database. In the relational database, the SAV objects can be stored in tables, such as Symptom table, Attribute table and Value table, and then with relationship tables, such as SymptomAttribute table, and SymptomAttributeValue tables as shown in FIG. 9C.”[0275]
Regarding claims 8 and 17
(clam 8) The system of claim 1, wherein the at least one computer processor is further configured to link the output structured data with the plurality of transcript concepts in the transcript.
Leventhal et al. teaches:
Example of SAV (Symptom-attribute-value) object (transcript concept) via (link) structured fields, where objects can be plural…
“ii. Tagging can be achieved by using the structured (non-textual) fields in the electronic medical records accompanying the text, as contextual labels for the texts. For example, with this approach, the SAV of a diabetic patient can be learned by associating with patients that had high levels of Hemoglobin HbA1c or patients that were prescribed with Insulin. Stated somewhat differently, even when a medical condition is not explicitly recorded in the medical notes, the server system 200 can identify that the current patient has a specific medical condition via the structured fields (e.g., lab test). The server system 200 can use the SAV objects from data of other patients with the same medical condition to find the SAV objects to extract for the current patient.” [0152]
Another example of relationship (link) between symptom (transcript concept) and attribute and attribute values (structured output)…
“Turning to FIG. 9B, an exemplary diagram illustrating an alternative embodiment of the tagging process is shown. The SAV objects of the objects in the ontology 230 (shown in FIG. 3C) can be stored in a relational database or a graph database. In a Graph database, the relationship between a symptom, possible attributes thereof, and the possible values for the attributes can be represented naturally in a graph. In one embodiment, The SAV objects can be stored in a relational database. In the relational database, the SAV objects can be stored in tables, such as Symptom table, Attribute table and Value table, and then with relationship tables, such as SymptomAttribute table, and SymptomAttributeValue tables as shown in FIG. 9C.”[0275]
Regarding claims 9 and 18
(claim 9) The system of claim 8, wherein the at least one computer processor is further configured to visually highlight the plurality of transcript concepts in the transcript linked to the output structured data.
Leventhal et al. teaches:
Example of mark as highlight…
“As shown in FIG. 7, a person tagging the text (or a human tagger, usually a medical doctor, medical student, etc.) marks texts as being a symptom (throat ache), attribute (severity) and value (light), forming a symptom tag 221 and an attribute tag 222 integrated with a value tag 223. Note that negative symptoms (that is, symptoms that the patient does not suffer from) are market as negative (e.g., no fever).” [0259]
Fig. 9 and highlight…
PNG
media_image2.png
158
634
media_image2.png
Greyscale
Claims 2, 3, 11, and 12 are rejected under 35 U.S.C. 103 as being unpatentable over the combined references in section (7) above in further view of Pub. No. US 2024/0145050 to Thompson IV et al.
Regarding claims 2 and 11
(claim 2) The system of claim 1, wherein, for each concept combination, the at least one computer processor is further configured to:
generate one or more text encodings, wherein each of the one or more text encodings is generated by encoding at least a portion of the transcript comprising one or more transcript concepts from the plurality of transcript concepts;
{
From Applicant’s specification on encoding…
“In some embodiments, the one or more text encodings 160 may be generated by a text encoder 162 using a pretrained language model, e.g., Word2vec, Long Short-Term Memory (LSTM), Transformer, etc. For example, the text encoder 162 may encode the portion of the transcript T including the one or more transcript concepts 132 in the form of the one or more text encodings 160. Specifically, the text encoder 162 may transform the portion of the transcript T (text) into a numerical representation. As shown in FIG. 2, the concept combination 142-1 incudes the transcript concepts 132-1, 132-2, i.e., “range or motion” and “shoulder”. Further, the concept combination 142-2 includes the transcript concepts 132-1, 132-3, i.e., “range or motion” and “wrist”.” [0052]
Therefore, using existing models to transform text.
}
Leventhal et al. teaches:
Encoded…
“The system can be based the type of data as described in the following. Some of the data are provided for illustrative purposes only, and the system can be extended to use and support additional data sources to provide wider and richer information. Additionally, a complete set of parallel records can be obtained for another population from a different source and be combined into the training data set using the method disclosed….” [0238]
“…(e) Potential or actual diagnosis—typically encoded in some standard such as ICD-9” [0244]
Example of translating (encoding) into case vectors from doctor notes…
“7. Translating such ontologies, and their corresponding tag vectors to be easily used in extraction of SAV values into case vectors from doctor notes in different languages.” [0157]
determine a plurality of slices of a tensor representation of the one or more text encodings corresponding to the one or more transcript concepts, wherein each slice from the plurality of slices contains a corresponding transcript concept from the one or more transcript concepts; and
Feature vector (1-dimensional tensor) with symptoms and profile (slices corresponding to transcript concepts)…
“Turning to FIG. 3C, an exemplary diagram illustrating an alternative embodiment of the environment 100 is shown. The online digital health information system 265 can include the app server 266, the conversation engine 264, the reference population calculator 1001, and the classifier server 268. The reference population (also referred to as RP, PopuLation Matched (PLM)) can include a subset of medical cases in the population medical database. Reference population information can include any information about the reference population and/or a reference population index which can be, and/or be based on, a number of the feature vectors having one or more relevant symptoms and one or more relevant profile features that are in the current feature vector.” [0061]
See Tensor below.
determine a tensor by concatenating the plurality of slices corresponding to the one or more transcript concepts.
Example of add (concatenate) joint pain (sliced) to visit vector (tensor)…
“Another example is the ability to treat knee pain, elbow pain and ankle pain as instances of joint pain. The post-tagging process can add a new symptom “joint pain” to the tagged visit vectors.” [0291]
Example of combined (concatenate) vector (tensor) with medical condition and diagnosis (slices)…
“Turning to FIG. 10, an exemplary diagram illustrating an embodiment of the case vector 228 is shown. The case vector 228 is shown as being a combined vector including the feature vector 224 of the medical case 246 (shown in FIG. 6) and diagnosis of a medical condition of the medical case for the same medical case 246. Those records along with the other data sources provide a larger case vector 228 that includes information such as patient age and gender, indication of chronic conditions, smoking habits, weight or BMI, and symptom features (the symptoms, attributes, and values) extracted from the text and the diagnosis.” [0293]
See Tensor below.
Tensor
The combined references teach vector. They do not literally use the word tensor.
Thompson IV et al. also in the business of vector teaches:
Snippets (slices) with tensors…
“In some embodiments, the first portion of the classifier further includes a multi-headed intra-attention mechanism that aggregates, for each respective episodic record in the sub-plurality of episodic records, the corresponding plurality of corresponding contextualized token tensors for each respective snippet in the plurality of corresponding snippets to output a corresponding contextualized snippet tensor, thereby forming a corresponding plurality of corresponding contextualized snippet tensors for the respective episodic record.” [0031]
Episodic records with clinical condition or outcome…
“The memory 92 of the computer system 100 stores: an operating system 34 that includes procedures for handling various basic system services; an input output module 64 for obtaining in electronic form, episodic records that include corresponding unstructured clinical data from one or more electronic medical records (EMR) or electronic health records (EHR) for patients. In some embodiments, the input output module 64 labels episodic records predicted to represent an instance of the clinical condition to form a set of episodic records. In some embodiments, the input output module 64 trains a model to predict an outcome of the clinical condition using the episodic records that are labelled; “[0067] – [0069]
“…a language pattern recognition module 40 for filtering the episodic records 38 using language pattern recognition to identify episodic records that include an expression related to a clinical condition. In some embodiments, the language pattern recognition module 40 matches one or more regular expressions against corresponding unstructured clinical data. In some embodiments, the language pattern recognition includes a machine learning model trained to identify language related to the clinical condition;” [0072]
Concatenation of vectors (token/snippet) constitute a tensor…
“… a splitting module 44 that includes snippets 46 and tokens 48. The splitting module 44 splits unstructured clinical data for an episodic record into corresponding snippets. Each snippet includes a corresponding set of tokens, which may include lexical tokens, such as words. The individual token and snippet representations may include vectors and are sometimes referred to as embeddings. The cumulation or concatenation of these vectors or embeddings constitutes a tensor. The snippets and tokens may be referred to as tensors, because the snippets and/or tokens are typically batched and concatenated during training;” [0074]
Aggregation with machine learning models…
“…a classifier 50 that includes an aggregation module 52 (sometimes referred to as a first portion of the classifier 50) and an interpretation module 54 (sometimes referred to as the second portion of the classifier 50). The first portion includes an aggregation function that aggregates corresponding snippets for an episodic record to output a corresponding representation. The second portion interprets the corresponding representation to output a corresponding prediction for whether the episodic record represents an instance of a clinical condition. The aggregation module 52 and the interpretation module 54 include respective parameters (e.g., parameters obtained from training machine learning models);” [0075]
It would have been obvious to one of ordinary skill in the art before the effective filing date to include in the method and system of the combined references to use tensors as taught by Thompson IV et al. since the claimed invention is merely a combination of old elements and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. Further motivation is provided by Thompson IV et al. who teaches using tensors and that tensors can be vectors. The combined references benefit as they also teach using vectors.
Regarding claims 3 and 12
(clam 3) The system of claim 2, wherein the at least one computer processor is further configured to feed the tensors corresponding to the plurality of concept combinations to the machine learning model.
Leventhal et al. teaches:
Neural Network (machine learning) and input from vectors (tensors)…
“Turning to FIG. 12A, an exemplary diagram illustrating an alternative embodiment of a process of machine learning is shown. The health predictive model 210 can include an additional and/or alternative Neural Network model. The health predictive model 210 can be trained using back-propagation methods using the combined case vectors 228 and/or feature vectors 224 as input and expected output.” [0305]
Tensor
The combined references teach vector. They do not literally use the word tensor.
Thompson IV et al. also in the business of vector teaches:
Snippets (slices) with tensors…
“In some embodiments, the first portion of the classifier further includes a multi-headed intra-attention mechanism that aggregates, for each respective episodic record in the sub-plurality of episodic records, the corresponding plurality of corresponding contextualized token tensors for each respective snippet in the plurality of corresponding snippets to output a corresponding contextualized snippet tensor, thereby forming a corresponding plurality of corresponding contextualized snippet tensors for the respective episodic record.” [0031]
Episodic records with clinical condition or outcome…
“The memory 92 of the computer system 100 stores: an operating system 34 that includes procedures for handling various basic system services; an input output module 64 for obtaining in electronic form, episodic records that include corresponding unstructured clinical data from one or more electronic medical records (EMR) or electronic health records (EHR) for patients. In some embodiments, the input output module 64 labels episodic records predicted to represent an instance of the clinical condition to form a set of episodic records. In some embodiments, the input output module 64 trains a model to predict an outcome of the clinical condition using the episodic records that are labelled; “[0067] – [0069]
“…a language pattern recognition module 40 for filtering the episodic records 38 using language pattern recognition to identify episodic records that include an expression related to a clinical condition. In some embodiments, the language pattern recognition module 40 matches one or more regular expressions against corresponding unstructured clinical data. In some embodiments, the language pattern recognition includes a machine learning model trained to identify language related to the clinical condition;” [0072]
Concatenation of vectors (token/snippet) constitute a tensor…
“… a splitting module 44 that includes snippets 46 and tokens 48. The splitting module 44 splits unstructured clinical data for an episodic record into corresponding snippets. Each snippet includes a corresponding set of tokens, which may include lexical tokens, such as words. The individual token and snippet representations may include vectors and are sometimes referred to as embeddings. The cumulation or concatenation of these vectors or embeddings constitutes a tensor. The snippets and tokens may be referred to as tensors, because the snippets and/or tokens are typically batched and concatenated during training;” [0074]
Aggregation with machine learning models…
“…a classifier 50 that includes an aggregation module 52 (sometimes referred to as a first portion of the classifier 50) and an interpretation module 54 (sometimes referred to as the second portion of the classifier 50). The first portion includes an aggregation function that aggregates corresponding snippets for an episodic record to output a corresponding representation. The second portion interprets the corresponding representation to output a corresponding prediction for whether the episodic record represents an instance of a clinical condition. The aggregation module 52 and the interpretation module 54 include respective parameters (e.g., parameters obtained from training machine learning models);” [0075]
It would have been obvious to one of ordinary skill in the art before the effective filing date to include in the method and system of the combined references to use tensors as taught by Thompson IV et al. since the claimed invention is merely a combination of old elements and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. Further motivation is provided by Thompson IV et al. who teaches using tensors and that tensors can be vectors. The combined references benefit as they also teach using vectors.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
The following prior art teaches at least text labeling and machine learning:
US-11742092-B2; US-11810671-B2; US-11568997-B2; US-12142358-B2; US-11875883-B1; US-12573484-B2; US-12112839-B2; US-20150324535-A1; US-20200251225-A1; US-20170213007-A1; US-20170235906-A1; US-20200160985-A1; US-20200388396-A1; US-20240404669-A1; US-20250095802-A1; US-20140181128-A1; US-20140330586-A1; US-20220375605-A1; US-20230223016-A1; US-20240428958-A1; US-20250252261-A1; US-20200176098-A1; US-20210057068-A1
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KENNETH BARTLEY whose telephone number is (571)272-5230. The examiner can normally be reached Mon-Fri: 7:30 - 4:00 EST.
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, SHAHID MERCHANT can be reached at (571) 270-1360. 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.
/KENNETH BARTLEY/Primary Examiner, Art Unit 3684