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 .
Priority
Receipt is acknowledged that application is a National Stage application of PCT PCT/EP2023/066342. Priority to EP22187507.3 with a priority date of 7/28/2022 is acknowledged under 35 USC 119(e) and 37 CFR 1.78.
Receipt is acknowledged that application is a National Stage application of PCT PCT/EP2023/066342. Priority to PCT/CN2022/100008 with a priority date of 6/21/2021 is acknowledged under 35 USC 119(e) and 37 CFR 1.78.
Information Disclosure Statement
The IDS dated 12/19/2024 has been considered and placed in the application file.
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 1-15 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claims 1, 14, and 15, Further claim 1 recites A data processing method for analysis of medical records, comprising: obtaining a text of a data record; obtaining a questionnaire including at least one question, the at least one question being associated with a data type of data in the data record; selecting, based on the data type, a data extraction module from a plurality of data extraction modules being collectively configured to extract data of a plurality of data types, the selected data extraction module being configured to extract data of the data type from the text; and extracting, by the selected data extraction module, data from the text of the data record.
Further claim 14 states An electronic device, comprising: at least one processing unit; and a memory coupled to the at least one processing unit and storing computer program instructions therein, the instructions, when executed by the at least one processing unit,
Further claim 15 states A computer-readable storage medium having program code stored thereon
The limitation of “obtaining …”, “obtaining…”, “selecting…”, and “extracting…” , as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind. For example, a person receiving documents and questions. After receiving the questions classifying them into types whether the answer would be numerical value or text. Further going through the document and finding the answer to the questions received.
If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
This judicial exception is not integrated into a practical application. In particular, claim 1 does not recite any additional elements however claim 14 and 15 recite additional elements that are computer components “processor” (page 18 lines 25-35) and “memory” (page 16 lines 15-30) recited at a high-level of generality such that it amounts no more than mere instructions to apply the exception using a generic computer component. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The claims are directed to an abstract idea.
The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional element of using the computer components 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. The claims are not patent eligible.
Claims 2 additionally recite The method of claim 1, wherein obtaining a questionnaire includes: determining a category of the data record, and selecting, based on the category of the data record, a questionnaire from a plurality of candidate questionnaires. However, this limitation does not prevent a human from performing the steps mentally as described above. Further, the person categorizing the documents received, further selecting the questions based on the type of document received. Thus, these claims are directed towards a mental process. Similar to above, no additional limitations are provided that provide a practical application, or amount to significantly more than the abstract idea. Therefore, the claims are not patent eligible.
Claims 3 additionally recites The method of claim 2, wherein determining the category of the data record includes: obtaining a word library including a plurality of words associated with categories of the data record; searching the text of the data record based on the word library; and in response to determining that a word in the word library is found in the text, determining the category of the data record based on the word. However, these limitations encompass a person categorizing the document by going through it and looking for key words such as X-Ray, MRI and etc. to be assorted with body images. Thus, these claims are directed towards a mental process. Similar to above, no additional limitations are provided that provide a practical application, or amount to significantly more than the abstract idea. Therefore, the claims are not patent eligible.
Claims 4 additionally recites The method of claim 2, wherein selecting a questionnaire from a plurality of candidate questionnaires includes: in response to determining that the category of the data record is a data record including at least one table, selecting a questionnaire including a question associated with a table; and in response to determining that the category of the data record is a data record including at least one paragraph, selecting a questionnaire including a question associated with a paragraph. However, these limitations encompass a person categorizing the document by going through it and looking to see if the document is structured data or unstructured data and selecting a question based on the document type. Thus, these claims are directed towards a mental process. Similar to above, no additional limitations are provided that provide a practical application, or amount to significantly more than the abstract idea. Therefore, the claims are not patent eligible.
Claims 5 additionally recites The method of claim 4, wherein the question in association with the data record including at least one table is associated with one or more of: an item of a medical laboratory examination, an object value for the item, a unit for the object value, a reference value for the item, and an abnormal indication; and the question in association with the data record including at least one paragraph is associated with one or more of: a datetime data type, a choice data type, a Boolean data type, a string data type, and a quantity data type. However, these limitations encompass a person categorizing the document by going through it and looking to see if the document is structured data or unstructured data and selecting a question based on the document type. Furthermore, the structured data would be in regards to medical equipment and the unstructured data would be in regards to date time, multiple choice, or free response. Thus, these claims are directed towards a mental process. Similar to above, no additional limitations are provided that provide a practical application, or amount to significantly more than the abstract idea. Therefore, the claims are not patent eligible.
Claims 6 additionally The method of claim 4, wherein the data record including at least one table includes a medical laboratory examination report having the at least one table; and the data record including at least one paragraph includes a medical imaging report having the at least one paragraph. However, these limitations encompass a person categorizing the document by going through it and looking to see if the document is structured data or unstructured data and selecting a question based on the document type. Furthermore, the structured data would be in regards to medical equipment report and the unstructured data would be in regards to a medical image. Thus, these claims are directed towards a mental process. Similar to above, no additional limitations are provided that provide a practical application, or amount to significantly more than the abstract idea. Therefore, the claims are not patent eligible.
Claims 7 additionally recites The method of claim 1, wherein the at least one question includes at least a first question associated with a first data type and a second question associated with a second data type different from the first data type, and the method further comprises: selecting, based on the first data type, a first data extraction module associated with the first data type from the plurality of data extraction modules; extracting, by the first data extraction module, data from the text of the data record; selecting, based on the second data type, a second data extraction module associated with the second data type from the plurality of data extraction modules; and extracting, by the second data extraction module, data from the text of the data record. However, these limitations encompass a person receiving questions in different types and answering them based on the documents. Thus, the claim is directed towards a mental process. Similar to above, no additional limitations are provided that provide a practical application, or amount to significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claims 8 additionally recites The method of claim 1, wherein the plurality of data extraction modules comprises at least two of the following: a datetime module associated with a datetime data type, the datetime module being configured to extract data of the datetime data type from the text; a text classifier associated with a choice data type or a Boolean data type, the text classifier being configured to extract data of the choice data type or the Boolean data type from a paragraph of the text; a question and answers module associated with a string data type or a quantity data type, the question and answers module being configured to extract data of the string data type or the quantity data type from the text as an answer to the question; and a named entity recognition module associated with a table, the named entity recognition module being configured to extract an entity corresponding to an entity category from a table area of the text. However, these limitations encompass a person going through the documents in a certain few such as looking for time, keywords, numbers, and entities to find the answer. Thus, the claim is directed towards a mental process. Similar to above, no additional limitations are provided that provide a practical application, or amount to significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claims 9 additionally recites The method of claim 8, wherein the text classifier is configured to select a paragraph in the text of the data record; segment the paragraph into multiple sentences; determine a data attribute based on the question; determine a semantical relationship between each of the multiple sentences and the data attribute; an determine an answer to the question based on the semantical relationship. However, these limitations encompass a person going through the documents finding the best paragraph within the document, separating it into sentences and ranking those sentences. Further answering the question based on the best sentence. Thus, the claim is directed towards a mental process. Similar to above, no additional limitations are provided that provide a practical application, or amount to significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claims 10 additionally recites The method of claim 8, wherein the named entity recognition module includes a conditional random field model, and the conditional random field model is configured to detect an entity in the table area; determine that the detected entity matches an entity category of a plurality of entity categories, wherein the plurality of entity categories includes one or more of: an item of a medical laboratory examination, an object value for the item, a unit for the object value, a reference value for the item, and an abnormal indication; and generate a label indicating the detected entity and the matched entity category. However, these limitations encompass a person going through the documents finding an entity and matching it to a entity category within the medical field, further making a label of that entity. Thus, the claim is directed towards a mental process. Similar to above, no additional limitations are provided that provide a practical application, or amount to significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claims 11 additionally recites the method of claim 8, wherein the question and answers module is configured to determine a data attribute based on the question; determine a location of one or more characters related to the data attribute in the text; and extract data of the string data type or the quantity data type from the text based on the location. However, these limitations encompass a person going through the documents to find a data attribute. Further determining its location within the text and extracting it. Thus, the claim is directed towards a mental process. Similar to above, no additional limitations are provided that provide a practical application, or amount to significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claims 12 additionally recites the method of claim 1, further comprising outputting the questionnaire filled with the extracted data. However, these limitations encompass a person going through the documents to find the answer to a question and giving the answer and question to the individual. Thus, the claim is directed towards a mental process. Similar to above, no additional limitations are provided that provide a practical application, or amount to significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claims 13 additionally recites The method of any of claims 1-12 claim 1,wherein the selected data extraction module is generated by training a machine learning model with a training data set of the data type, the training data set includes standard data obtained from a plurality of data records. However, these limitations encompass a person training Somone on the plurality of document categories and data types he will encounter. Thus, the claim is directed towards a mental process. Similar to above, no additional limitations are provided that provide a practical application, or amount to significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1, 14, 15 are rejected under 35 U.S.C. 102(a)(1) and (a)(2) as being anticipated by US 20180196920 A1, (Liang; Jennifer J.).
Claim 1, 14, and 15
Regarding Claim 1, 14, and 15 , Liang teach
1. (Original) A data processing method for analysis of medical records, comprising:
(paragraph 9 "In one illustrative embodiment, a method is provided in a data processing system comprising at least one processor and at least one memory, the at least one memory comprising instructions which are executed by the at least one processor and configure the processor to implement a patient information extractor. The method comprises receiving, by the patient information extractor, a query specification for executing a query on a patient electronic medical record (EMR). The query specification provides parameters indicating a methodology for extracting search results from the patient EMR. The method further comprises retrieving, by the patient information extractor, the patient EMR from a patient registry. The method further comprises automatically executing, by the patient information extractor, the query specification on the retrieved patient EMR to thereby extract the search results from the patient EMR in accordance with the parameters of the query specification. The method further comprises automatically processing, by the patient information extractor, the extracted search results to generate a patient indicator value. The patient indicator value represents an answer to a question about the patient. The method further comprises performing a patient evaluation operation based on the patient indicator value.")
obtaining a text of a data record;
(paragraph 105 "FIG. 4 is a block diagram of a patient information extractor in accordance with an illustrative embodiment. Each patient is represented by an electronic medical record (EMR) 440. This record consists of (1) a number of clinical notes 441, which are free-text descriptions of patient encounters with medical staff (e.g., office visits), surgical procedures, and interactions between medical professionals concerning the patient, and (2) structured data 442 detailing ordered medications, tests and procedures, and patient demographic and possibly other static information (e.g., lists of allergies). It is likely and desirable, but not required for the purposes of this disclosure, that the EMR has been processed by analytic tools, such as MetaMap, which can detect free-text medical concepts and associate with them concepts in ontology, such as Unified Medical Language System (UM LS) dictionary, for example. The UMLS is a compendium of many controlled vocabularies in the biomedical sciences. It provides a mapping structure among these vocabularies and, thus, allows one to translate among the various terminology systems; it may also be viewed as a comprehensive thesaurus and ontology of biomedical concepts. UMLS further provides facilities for natural language processing. It is intended to be used mainly by developers of systems in medical informatics. Other useful but not required analytic tools would be tools that determine the type of clinical note (e.g., Office Visit, Operative Report, Telephone Encounter, Patient Instructions), and the sections within a note (e.g., Past Medical History, Current Medications, Assessment and Plan).")
obtaining a questionnaire including at least one question, the at least one question being associated with a data type of data in the data record;
(paragraph 20-27, 29-30 "In this disclosure, the term “patient indicator” refers to a value, which may be Boolean, numeric, date, categorical, or any other type, that represents the answer to a question about the patient, the patient's health, or the patient's medical, social, or family history present in an electronic medical record (EMR). Typical such questions include:
[0021] What is the patient's age?
[0022] What is the patient's ethnicity?
[0023] Does the patient have a family history of cancer?
[0024] Has the patient been on tamoxifen within the last year?
[0025] Did the patient ever have blood clots following surgery?
[0026] What kind of hysterectomy did the patient have?
[0027] When was the patient's last colonoscopy?
[0029] An indicator question falls into one of a small number of classes: yes/no, temporal, categorical, etc. The illustrative embodiments provide a dedicated search engine for each of these classes, which given one or more search terms and possible constraints can generate a list of search results. The illustrative embodiments also provide post-processors that take these search results and any constraints not expressible to the search engine, perform filtering and sorting, and extract indicator values.
[0030] All of the parameters for this search and extraction process can be laid out in a machine-readable attribute-value format, such as the JavaScript™ Object Notation (JSON), in a query specification data structure or document. The query specification contains entries for each indicator of interest. The process is as follows: the user writes a query specification for each patient indicator of interest; the query specification is run against the EMR, generating search results; the search results are post-processed to generate the indicator value; the indicator values are written out. This process may then be repeated for other patients, if desired. These indicators subsequently may be used to determine whether the patient qualifies for a trial, to predict likely future medical concerns, to assess whether appropriate care has been given, etc.")
selecting, based on the data type, a data extraction module
(paragraph 29 " An indicator question falls into one of a small number of classes: yes/no, temporal, categorical, etc. The illustrative embodiments provide a dedicated search engine for each of these classes, which given one or more search terms and possible constraints can generate a list of search results. The illustrative embodiments also provide post-processors that take these search results and any constraints not expressible to the search engine, perform filtering and sorting, and extract indicator values."
Paragraph 78 "In one embodiment, patient information extractor 120 provides one or more logical search engines (LSE), which find matches for search terms in the EMR according to specific logic, and one or more search result processors (SRP), which take the output of the LSEs, filter the results according to specified constraints, and return a result. Several LSEs may be implemented by the same physical search engine. The LSEs and SRPs are shown in FIG. 6 and described in further detail below."
Paragraph 120 "" FIG. 6 is a flowchart illustrating operation of a patient information extractor in accordance with an illustrative embodiment. Operation begins for a current query specification and patient EMR (block 600), and the patient information extractor determines whether more query specification blocks (QSB) exist (block 601). Because a query specification has at least one QSB, the first iteration will result in a determination that more QSB exist. The patient information extractor then gets the next QSB in the query specification and extracts the specified search term, logical search engine (LSE), and search result processor (SRP) (block 602). The patient information extractor also extracts from the query specification one or more constraints, such as a date restriction, a note type restriction, a section type restriction, a provider type restriction, or a department/specialty restriction.)
from a plurality of data extraction modules
(Paragraph 78 "In one embodiment, patient information extractor 120 provides one or more logical search engines (LSE), which find matches for search terms in the EMR according to specific logic, and one or more search result processors (SRP), which take the output of the LSEs, filter the results according to specified constraints, and return a result. Several LSEs may be implemented by the same physical search engine. The LSEs and SRPs are shown in FIG. 6 and described in further detail below."
paragraph 108 "In processing a QSB, query specification processor 410 calls the specified logical search engine (LSE) 420 with the specified search term. A “literal” LSE finds instances of the search term. For example, given a search term “heart attack,” the LSE would match the term “heart attack.” A “literal” LSE is similar to a standard find function and may or may not observe capitalization. A “semantic” LSE finds conceptual matches. For example, a “semantic” LSE may match the search term “heart attack” with the matched term “myocardial infarction,” because both terms are mapped to the same concept in UMLS. A “more specific” LSE finds conceptual matches via “ISA” (is a) relations. In knowledge representation, object-oriented programming and design, “ISA” (is_a or is-a) is a subsumption relationship between abstractions (e.g., types, classes), where one class A is a subclass of another class B (and so B is a superclass of A). For example, a “more specific” LSE may match the search term “cancer” with “leukemia,” because in UMLS, leukemia ISA cancer. An “associative” LSE finds terms that co-occur in external corpora. For example, an “associative” LSE may match the search term “asthma” with the matched term “wheezing.” Standard technologies such as latent semantic analysis (LSA) do this. A “logical” LSE finds conceptual matches via relations other than ISA. For example, a “logical” LSE may match the search term “headache” with the matched term “Tylenol,” because Tylenol® TREATS headache, and match the search term “Tylenol” to the matched term “acetaminophen,” because Tylenol is a BRAND_NAME_OF acetaminophen.")
being collectively configured to extract data of a plurality of data types,
(paragraph 111 "A “SpecificIndicator” SRP receives an indicator value as a parameter and performs the following operation: If the hit list qualifies (by size and date range), return the indicator value, else null. A “YesNo” SRP performs the following operation: If the hit list qualifies, return Yes (or 1), else No (or 0). A “FirstLast” SRP receives a flag indicating First/Last as a parameter and performs the following operation: The qualifying hit list is sorted by date and the date of the first/last entry is returned. A “Temporal” SRP can receive as parameters one or two dates, each date with a flag indicating before/after. With this revision, the SRP can return yes/no depending on if the match falls within two specified dates. An “AnswerType” SRP receives an answer type indicating a property of the patient (e.g., age, weight) and performs the following operation: either the answer is looked up in the data structure part of the EMR, or a question-answering system is run on the clinical notes; the specified quantity is returned.")
the selected data extraction module being configured to extract data of the data type from the text; and
(paragraph 105 "FIG. 4 is a block diagram of a patient information extractor in accordance with an illustrative embodiment. Each patient is represented by an electronic medical record (EMR) 440. This record consists of (1) a number of clinical notes 441, which are free-text descriptions of patient encounters with medical staff (e.g., office visits), surgical procedures, and interactions between medical professionals concerning the patient, and (2) structured data 442 detailing ordered medications, tests and procedures, and patient demographic and possibly other static information (e.g., lists of allergies). It is likely and desirable, but not required for the purposes of this disclosure, that the EMR has been processed by analytic tools, such as MetaMap, which can detect free-text medical concepts and associate with them concepts in ontology, such as Unified Medical Language System (UM LS) dictionary, for example. The UMLS is a compendium of many controlled vocabularies in the biomedical sciences. It provides a mapping structure among these vocabularies and, thus, allows one to translate among the various terminology systems; it may also be viewed as a comprehensive thesaurus and ontology of biomedical concepts. UMLS further provides facilities for natural language processing. It is intended to be used mainly by developers of systems in medical informatics. Other useful but not required analytic tools would be tools that determine the type of clinical note (e.g., Office Visit, Operative Report, Telephone Encounter, Patient Instructions), and the sections within a note (e.g., Past Medical History, Current Medications, Assessment and Plan)."
Paragraph 111" A “SpecificIndicator” SRP receives an indicator value as a parameter and performs the following operation: If the hit list qualifies (by size and date range), return the indicator value, else null. A “YesNo” SRP performs the following operation: If the hit list qualifies, return Yes (or 1), else No (or 0). A “FirstLast” SRP receives a flag indicating First/Last as a parameter and performs the following operation: The qualifying hit list is sorted by date and the date of the first/last entry is returned. A “Temporal” SRP can receive as parameters one or two dates, each date with a flag indicating before/after. With this revision, the SRP can return yes/no depending on if the match falls within two specified dates. An “AnswerType” SRP receives an answer type indicating a property of the patient (e.g., age, weight) and performs the following operation: either the answer is looked up in the data structure part of the EMR, or a question-answering system is run on the clinical notes; the specified quantity is returned.")
extracting, by the selected data extraction module, data from the text of the data record.
(paragraph 120 "FIG. 6 is a flowchart illustrating operation of a patient information extractor in accordance with an illustrative embodiment. Operation begins for a current query specification and patient EMR (block 600), and the patient information extractor determines whether more query specification blocks (QSB) exist (block 601). Because a query specification has at least one QSB, the first iteration will result in a determination that more QSB exist. The patient information extractor then gets the next QSB in the query specification and extracts the specified search term, logical search engine (LSE), and search result processor (SRP) (block 602). The patient information extractor also extracts from the query specification one or more constraints, such as a date restriction, a note type restriction, a section type restriction, a provider type restriction, or a department/specialty restriction."
paragraph 121 "The patient information extractor processes the search term with the specified LSE to search the patient EMR and generate search results (block 603). The patient information extractor then processes the search results with the specified SRP to generate an indicator value (block 604). The patient information extractor determines whether the indicator value is null (i.e., the search results were empty or fail to meet the one or more constraints) (block 605). If the value is null, then the current QSB did not result in generating an indicator value, and operation returns to block 601 to determine whether more QSB exist. The patient information extractor then repeats blocks 602-605 for the next QSB if one exists." )
Regarding Claim 14, Liang further teach
14. (Currently amended) An electronic device, comprising:
at least one processing unit;
(paragraph 33 "Moreover, it should be appreciated that the use of the term “engine,” if used herein with regard to describing embodiments and features of the invention, is not intended to be limiting of any particular implementation for accomplishing and/or performing the actions, steps, processes, etc., attributable to and/or performed by the engine. An engine may be, but is not limited to, software, hardware and/or firmware or any combination thereof that performs the specified functions including, but not limited to, any use of a general and/or specialized processor in combination with appropriate software loaded or stored in a machine readable memory and executed by the processor. Further, any name associated with a particular engine is, unless otherwise specified, for purposes of convenience of reference and not intended to be limiting to a specific implementation. Additionally, any functionality attributed to an engine may be equally performed by multiple engines, incorporated into and/or combined with the functionality of another engine of the same or different type, or distributed across one or more engines of various configurations.")
and a memory coupled to the at least one processing unit and storing computer program instructions therein, the instructions, when executed by the at least one processing unit, causing the electronic device to perform the method of any of claims 1-13 claim 1.
(paragraph 33 "Moreover, it should be appreciated that the use of the term “engine,” if used herein with regard to describing embodiments and features of the invention, is not intended to be limiting of any particular implementation for accomplishing and/or performing the actions, steps, processes, etc., attributable to and/or performed by the engine. An engine may be, but is not limited to, software, hardware and/or firmware or any combination thereof that performs the specified functions including, but not limited to, any use of a general and/or specialized processor in combination with appropriate software loaded or stored in a machine readable memory and executed by the processor. Further, any name associated with a particular engine is, unless otherwise specified, for purposes of convenience of reference and not intended to be limiting to a specific implementation. Additionally, any functionality attributed to an engine may be equally performed by multiple engines, incorporated into and/or combined with the functionality of another engine of the same or different type, or distributed across one or more engines of various configurations."
Paragraph 81 "In the depicted example, local area network (LAN) adapter 212 connects to SB/ICH 204. Audio adapter 216, keyboard and mouse adapter 220, modem 222, read only memory (ROM) 224, hard disk drive (HDD) 226, CD-ROM drive 230, universal serial bus (USB) ports and other communication ports 232, and PCI/PCIe devices 234 connect to SB/ICH 204 through bus 238 and bus 240. PCI/PCIe devices may include, for example, Ethernet adapters, add-in cards, and PC cards for notebook computers. PCI uses a card bus controller, while PCIe does not. ROM 224 may be, for example, a flash basic input/output system (BIOS).")
Regarding Claim 15, Liang further teach
15. (Currently amended) A computer-readable storage medium having program code stored thereon, the program code configured, upon execution, to cause an apparatus to perform the method of any of claims 1-13 claim 1.
(paragraph 33 "Moreover, it should be appreciated that the use of the term “engine,” if used herein with regard to describing embodiments and features of the invention, is not intended to be limiting of any particular implementation for accomplishing and/or performing the actions, steps, processes, etc., attributable to and/or performed by the engine. An engine may be, but is not limited to, software, hardware and/or firmware or any combination thereof that performs the specified functions including, but not limited to, any use of a general and/or specialized processor in combination with appropriate software loaded or stored in a machine readable memory and executed by the processor. Further, any name associated with a particular engine is, unless otherwise specified, for purposes of convenience of reference and not intended to be limiting to a specific implementation. Additionally, any functionality attributed to an engine may be equally performed by multiple engines, incorporated into and/or combined with the functionality of another engine of the same or different type, or distributed across one or more engines of various configurations.")
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.
Claims 2 are rejected under 35 U.S.C. 103 as obvious over US 20180196920 A1, (Liang; Jennifer J.) in view of US 20190228848 A1, (Saliman; Justin).
Claim 2
Regarding Claim 2, Liang teach
2. (Original) The method of claim 1, wherein obtaining a questionnaire includes:
determining a category of the data record, and
(paragraph 105 "FIG. 4 is a block diagram of a patient information extractor in accordance with an illustrative embodiment. Each patient is represented by an electronic medical record (EMR) 440. This record consists of (1) a number of clinical notes 441, which are free-text descriptions of patient encounters with medical staff (e.g., office visits), surgical procedures, and interactions between medical professionals concerning the patient, and (2) structured data 442 detailing ordered medications, tests and procedures, and patient demographic and possibly other static information (e.g., lists of allergies). It is likely and desirable, but not required for the purposes of this disclosure, that the EMR has been processed by analytic tools, such as MetaMap, which can detect free-text medical concepts and associate with them concepts in ontology, such as Unified Medical Language System (UM LS) dictionary, for example. The UMLS is a compendium of many controlled vocabularies in the biomedical sciences. It provides a mapping structure among these vocabularies and, thus, allows one to translate among the various terminology systems; it may also be viewed as a comprehensive thesaurus and ontology of biomedical concepts. UMLS further provides facilities for natural language processing. It is intended to be used mainly by developers of systems in medical informatics. Other useful but not required analytic tools would be tools that determine the type of clinical note (e.g., Office Visit, Operative Report, Telephone Encounter, Patient Instructions), and the sections within a note (e.g., Past Medical History, Current Medications, Assessment and Plan).")
Liang do not explicitly teach all of selecting, based on the category of the data record, a questionnaire from a plurality of candidate questionnaires.
However, Saliman teach selecting, based on the category of the data record, a questionnaire from a plurality of candidate questionnaires.
(paragraph 53 "Finally, reporting module 122 may report the scores to one or more of user device 124, patient device 128 and treatment facility computer system 134. In addition to the request for a treatment, there are other events that may prompt an OMD to be administered to a patient. In one example, the scheduling of an initial appointment (e.g., a consultation) for a patient to discuss a medical condition with a healthcare professional may prompt an OMD to be administered to the patient. Administering an OMD to the patient prior to this initial appointment may be useful for establishing a baseline state of health for the patient, but the selection of the OMD may have some complexity, as no treatment code, treatment name or diagnostic code may yet be available when the initial appointment is scheduled. In many instances, all that the patient will provide is a brief description of the symptoms he/she may be experiencing (e.g., shortness of breath, fever, etc.) and/or a chief complaint. In one embodiment, such symptoms may be provided to OMD selector 106, which attempts to match the symptoms with one or more treatment codes, treatment names, or diagnostic codes."
paragraph 54 "Such matching by OMD selector 106 may be performed using a learning machine. For instance, matches between, for example, symptoms and treatment codes; symptom and treatment names; and/or symptoms and diagnostic codes) may be provided by a healthcare professional when treating patients, and such matches may be used to train a model that can then be used to determine treatment codes, treatment names or diagnostic codes based on, for example, a patient's symptoms and/or treatment provider notes. Upon determining a treatment code, treatment name, or a diagnostic code, OMD selector 106 may select one or more OMDs based on matches provided in matched treatment code and OMD database 104 (as described above). It is anticipated that the determination of a treatment code, treatment name or diagnostic code by OMD selector 106 may be, in some instances, an imperfect process, so a healthcare provider, or other expert, may be asked to make any necessary adjustments to the treatment code, treatment name and/or diagnostic code determination, before OMD selector 106 selects the one or more OMDs."
Paragraph 47 "OMD and/or medical questionnaire responses received by server 102 may be communicated to a medical note interface generator 140 by, for example, communications interface 114. Medical note interface generator 140 may then generate one or more medical note interfaces by, for example, executing one or more process(es) described herein. Generated medical note interfaces may be stored in, for example, a medical note interface database 142 which may, in some instances, index a patient and/or a patient account to a medical note interface. Additionally, or alternatively, medical note interfaces may be stored in, a medical note interface database 144 that is external to server 102 which may, in some instances, be operated by a treatment facility computer system 134 and/or a user. The medical note interfaces stored on medical note interface 144 may be associated with index a patient, a patient's EMR (stored in patient EMR database 130), and/or a patient account."
Paragraph 55 "As another example, the billing for a medical appointment during which a patient discusses a medical condition with a healthcare professional may prompt an OMD to be administered to the patient. More specifically, billing program 138 may notify server 102 when a bill (or invoice) is generated. If the billing for the appointment occurs after the appointment has concluded, the bill may be associated with a diagnostic code (which may be determined by the healthcare professional during the medical appointment). OMD selector 106 may use the diagnostic code to locate one or more appropriate OMDs for the patient with the assistance of matched treatment code and OMD database 104."
The classification of diagnostic to the patient is the category of record.
The OMD selector selects a questionnaire from a plurality of questionnaires based on the classification.)
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Liang to incorporate the teachings of Saliman to provide a “selecting, based on the category of the data record, a questionnaire from a plurality of candidate questionnaires.” Doing so would Establish a baseline state of health for the patient, as recognized by Saliman. (Paragraph 54).
Claims 3 are rejected under 35 U.S.C. 103 as obvious over US 20180196920 A1, (Liang; Jennifer J.) in view of US 20190228848 A1, (Saliman; Justin) in further view of US 20160210426 A1, (Robinson; Samuel A).
Claim 3
Regarding Claim 3, Liang in view of Saliman, further Liang teach
3. (Original) The method of claim 2, wherein determining the category of the data record includes:
obtaining a word library including a plurality of words associated with categories of the data record;
(paragraph 105 "FIG. 4 is a block diagram of a patient information extractor in accordance with an illustrative embodiment. Each patient is represented by an electronic medical record (EMR) 440. This record consists of (1) a number of clinical notes 441, which are free-text descriptions of patient encounters with medical staff (e.g., office visits), surgical procedures, and interactions between medical professionals concerning the patient, and (2) structured data 442 detailing ordered medications, tests and procedures, and patient demographic and possibly other static information (e.g., lists of allergies). It is likely and desirable, but not required for the purposes of this disclosure, that the EMR has been processed by analytic tools, such as MetaMap, which can detect free-text medical concepts and associate with them concepts in ontology, such as Unified Medical Language System (UM LS) dictionary, for example. The UMLS is a compendium of many controlled vocabularies in the biomedical sciences. It provides a mapping structure among these vocabularies and, thus, allows one to translate among the various terminology systems; it may also be viewed as a comprehensive thesaurus and ontology of biomedical concepts. UMLS further provides facilities for natural language processing. It is intended to be used mainly by developers of systems in medical informatics. Other useful but not required analytic tools would be tools that determine the type of clinical note (e.g., Office Visit, Operative Report, Telephone Encounter, Patient Instructions), and the sections within a note (e.g., Past Medical History, Current Medications, Assessment and Plan)."
Paragraph 108 "In processing a QSB, query specification processor 410 calls the specified logical search engine (LSE) 420 with the specified search term. A “literal” LSE finds instances of the search term. For example, given a search term “heart attack,” the LSE would match the term “heart attack.” A “literal” LSE is similar to a standard find function and may or may not observe capitalization. A “semantic” LSE finds conceptual matches. For example, a “semantic” LSE may match the search term “heart attack” with the matched term “myocardial infarction,” because both terms are mapped to the same concept in UMLS. A “more specific” LSE finds conceptual matches via “ISA” (is a) relations. In knowledge representation, object-oriented programming and design, “ISA” (is_a or is-a) is a subsumption relationship between abstractions (e.g., types, classes), where one class A is a subclass of another class B (and so B is a superclass of A). For example, a “more specific” LSE may match the search term “cancer” with “leukemia,” because in UMLS, leukemia ISA cancer. An “associative” LSE finds terms that co-occur in external corpora. For example, an “associative” LSE may match the search term “asthma” with the matched term “wheezing.” Standard technologies such as latent semantic analysis (LSA) do this. A “logical” LSE finds conceptual matches via relations other than ISA. For example, a “logical” LSE may match the search term “headache” with the matched term “Tylenol,” because Tylenol® TREATS headache, and match the search term “Tylenol” to the matched term “acetaminophen,” because Tylenol is a BRAND_NAME_OF acetaminophen.")
searching the text of the data record based on the word library; and
(paragraph 121 "The patient information extractor processes the search term with the specified LSE to search the patient EMR and generate search results (block 603). The patient information extractor then processes the search results with the specified SRP to generate an indicator value (block 604). The patient information extractor determines whether the indicator value is null (i.e., the search results were empty or fail to meet the one or more constraints) (block 605). If the value is null, then the current QSB did not result in generating an indicator value, and operation returns to block 601 to determine whether more QSB exist. The patient information extractor then repeats blocks 602-605 for the next QSB if one exists.")
Liang in view of Saliman do not explicitly teach all of in response to determining that a word in the word library is found in the text, determining the category of the data record based on the word.
However, Robinson teach
in response to determining that a word in the word library is found in the text, determining the category of the data record based on the word.
(paragraph 27 "Classifications data 132 may be stored by memory 114 and include various data (e.g., rules, features, instructions, algorithms) used to classify selected medical documents. For example, as will be described in more detail below, classifications data 132 may include features used in classifying selected medical documents. In some examples, these stored features may include words, phrases, characters, numbers, document titles, and other textual features that may be included in medical documents to be classified. Features may also include graphical components such as icons, symbols, or any other such identifiable items. In some examples, features may include various metadata related to medical documents. For example, features may include header information, text styles, page formatting, location of a document page relative to other document pages, position of various portions or sections within the medical document, and other metadata features. Classifications data 132 may additionally include various temporary data generated as a part of a document classification process, as will be explained in more detail below. In some examples, classifications data 132 may further include association information. The association information may be associations between various features and specific classifications that may be used as part of a classification process. For example, classifications data 132 may include an association between the phrase “history and physical” and the history and physical document classification. In some examples, the association may represent an addition or weighting factor that is used to correctly classify the portion of the medical document according to the identified features therein. Classifications data 132 may further include removable features. Removable features may include features, such as words characters, symbols, and/or phrases, that may be removed prior to or as part a classification process because the removable features may not assist the classification process. These removable features may be features of a medical document which do not affect the classification process or may interfere with correct classification of one or more portions of the document."
Paragraph 31 "Classification module 124 may then process the documents and the data generated by parsing module 122 to generate a classification for each document page. In some examples, classification module 124 may generate a classification from a list of predetermined classifications. The list of predetermined classifications may include categories or types of medical information that may limit the number of applicable medical codes relevant to each classification. In this manner, the accuracy of later coding of the medical information may be improved with the aid of the context of the classification. In at least one example, the predetermined classifications may include a history and physical classification, an operative report classification, an emergency room classification, a progress notes classification, and a discharge summary classification. In other examples, the exact names and number of predetermined classifications may vary. For example, the types of classifications may be adjusted for types of facilities, medical practices, medical professionals, patients, or any other situation. In this manner, as few as two or three classifications may be used or a many as ten, twenty, or more classifications may be relevant to the medical documents to be classified.")
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Liang in view of Saliman to incorporate the teachings of Robinson to provide a “in response to determining that a word in the word library is found in the text, determining the category of the data record based on the word.” Doing so would Improve the code of medical information , as recognized by Robinson. (Paragraph 31).
Claims 4 are rejected under 35 U.S.C. 103 as obvious over US 20180196920 A1, (Liang; Jennifer J.) in view of US 20190228848 A1, (Saliman; Justin) in view of US 20180157990 A1, (Allen; Corville O.) in further view of US 20230068338 A1, (Ramnani; Roshni Ramesh).
Claim 4
Regarding Claim 4,Liang in view of Saliman do not explicitly teach all of 4. (Original) The method of claim 2, wherein selecting a questionnaire from a plurality of candidate questionnaires includes: in response to determining that the category of the data record is a data record including at least one table, selecting a questionnaire including a question associated with a table; and
in response to determining that the category of the data record is a data record including at least one paragraph, selecting a questionnaire including a question associated with a paragraph.
However, Allen teach
4. (Original) The method of claim 2, wherein selecting a questionnaire from a plurality of candidate questionnaires includes:
in response to determining that the category of the data record is a data record including at least one table, selecting a questionnaire including a question associated with a table; and
(paragraph 94 "FIGS. 5a through 5c are a generalized flowchart of the performance of table-based groundtruth generation operations implemented in accordance with an embodiment of the invention. In this embodiment, table-based groundtruth generation operations are begun in step 502, followed by the receipt of a corpus of human-readable text, such as a collection of documents, in step 504. A target document within the corpus is selected in step 506, followed by the identification of any tables it may contain in step 508. A determination is then made in step 510 whether the document contains one or more tables."
Paragraph 96 "Questions for question-answer (QA) pairs are then automatically generated in step 522 by applying direct statement templates to the contents of the table. The resulting questions, the row/column label data associated with the table, and the contents of the table's cells are then processed in step 524 to generate associated QA pairs. A low initial score (e.g., score=0.1) is then assigned in step 526 to each of the resulting QA pairs to signify their respective significance, as they are generated from direct statements associated with the contents of the table and may not necessarily reflect typical end-user questions.")
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Liang in view of Saliman to incorporate the teachings of Allen to provide a “The method of claim 2, wherein selecting a questionnaire from a plurality of candidate questionnaires includes: in response to determining that the category of the data record is a data record including at least one table, selecting a questionnaire including a question associated with a table; and ” Doing so would Improve coverage of tables in a corpus, as recognized by Allen. (Paragraph 1).
Liang in view of Saliman in further view of Allen do not explicitly teach all of
in response to determining that the category of the data record is a data record including at least one paragraph, selecting a questionnaire including a question associated with a paragraph.
However, Ramnani teach
in response to determining that the category of the data record is a data record including at least one paragraph, selecting a questionnaire including a question associated with a paragraph.
(paragraph 22 "The input into the system may include input documents in the form of written text and/or audio transcripts. The input documents may include learning materials used to teach learners. The input documents may cover a wide variety of topics and may each include differing levels of detail depending on teaching goals. The input documents may be in various formats (e.g., MICROSOFT WORD, MICROSOFT POWERPOINT, PDF, etc.)."
paragraph 24 "referring back to FIG. 1, test builder 116 may include six main modules that may carry out the disclosed method: a preprocessor 118, a content selector 120, a question generator 122, a distractor generator 124, a question scorer 126, and a question organizer 128. The content selector may decide from which paragraphs/sentences of the original document to generate questions. The question generator module may generate questions. In some embodiment, one or more of the following types of questions may be generated: descriptive, factoid, fill in the blank, Boolean, and true-false. In some embodiments, the question generator module may include multiple question generators. For example, multiple question generators may be specific to the type of question generated. In some embodiments, the question generator module may include a single question generator that can generate multiple types of questions. The question generator(s) may include pre-trained models or templates containing common words and/or phrases used in questions, depending on the question type, and may include slots for words and/or phrases specific to the input documents (e.g., words and/or phrases extracted from input documents using disclosed techniques). The distractor generator may generate distractors (i.e., alternate confusing answers) for factoid and fill in the blank questions, which both include single word and/or single phrase answers. The answer scorer may score answers for description type questions. As part of scoring, the given answers may be compared with the ideal answer and the user may be provided feedback on the closeness to the answer. Answer scoring may be used to assess a user's performance and give feedback in real time. The question organizer may select questions and determine a particular order in which the questions are to be asked. In some embodiments, the question organizer may both select test questions from those generated by question generator and determine the order of the selected test questions based at least partially on answer scoring from the past performance of a user. The question organizer may allow a user (e.g., learner role) to opt for a shuffled assessment in which specific questions are chosen from the overall set and presented to the user based on one or more of question complexity, variety, and previous performance of the user.")
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Liang in view of Saliman in further view of Allen to incorporate the teachings of Ramnani to provide a “in response to determining that the category of the data record is a data record including at least one paragraph, selecting a questionnaire including a question associated with a paragraph.” Doing so would Select the best question suitable to the subject, as recognized by Ramnani. (Paragraph 5).
Claims 5 are rejected under 35 U.S.C. 103 as obvious over US 20180196920 A1, (Liang; Jennifer J.) in view of US 20190228848 A1, (Saliman; Justin) in view of US 20180157990 A1, (Allen; Corville O.) in view of US 20230068338 A1, (Ramnani; Roshni Ramesh) in further view of US 20200279622 A1, (Heywood; James A.).
Claim 5
Regarding Claim 5, Liang in view of Saliman in view of Allen in further view of Ramnani, further Liang teaches
the question in association with the data record including at least one paragraph
(paragraph 105 "FIG. 4 is a block diagram of a patient information extractor in accordance with an illustrative embodiment. Each patient is represented by an electronic medical record (EMR) 440. This record consists of (1) a number of clinical notes 441, which are free-text descriptions of patient encounters with medical staff (e.g., office visits), surgical procedures, and interactions between medical professionals concerning the patient, and (2) structured data 442 detailing ordered medications, tests and procedures, and patient demographic and possibly other static information (e.g., lists of allergies). It is likely and desirable, but not required for the purposes of this disclosure, that the EMR has been processed by analytic tools, such as MetaMap, which can detect free-text medical concepts and associate with them concepts in ontology, such as Unified Medical Language System (UM LS) dictionary, for example. The UMLS is a compendium of many controlled vocabularies in the biomedical sciences. It provides a mapping structure among these vocabularies and, thus, allows one to translate among the various terminology systems; it may also be viewed as a comprehensive thesaurus and ontology of biomedical concepts. UMLS further provides facilities for natural language processing. It is intended to be used mainly by developers of systems in medical informatics. Other useful but not required analytic tools would be tools that determine the type of clinical note (e.g., Office Visit, Operative Report, Telephone Encounter, Patient Instructions), and the sections within a note (e.g., Past Medical History, Current Medications, Assessment and Plan).")
is associated with one or more of: a datetime data type, a choice data type, a Boolean data type, a string data type, and a quantity data type.
(paragraph 20-27, 29-30 "In this disclosure, the term “patient indicator” refers to a value, which may be Boolean, numeric, date, categorical, or any other type, that represents the answer to a question about the patient, the patient's health, or the patient's medical, social, or family history present in an electronic medical record (EMR). Typical such questions include:
[0021] What is the patient's age?
[0022] What is the patient's ethnicity?
[0023] Does the patient have a family history of cancer?
[0024] Has the patient been on tamoxifen within the last year?
[0025] Did the patient ever have blood clots following surgery?
[0026] What kind of hysterectomy did the patient have?
[0027] When was the patient's last colonoscopy? ")
Liang in view of Saliman in view of Allen in further view of Ramnani do not explicitly teach all of 5. (Original) The method of claim 4, wherein the question in association with the data record including at least one table is associated with one or more of: an item of a medical laboratory examination, an object value for the item, a unit for the object value, a reference value for the item, and an abnormal indication; and
However, Heywood teach
5. (Original) The method of claim 4, wherein the question in association with the data record including at least one table is associated with one or more of: an item of a medical laboratory examination, an object value for the item, a unit for the object value, a reference value for the item, and an abnormal indication; and
(paragraph 73 - 74 "A medical object type is a category of thing (or a class, in programming terms: a type of thing that has multiple instances, all with potentially different attributes), such as a condition, symptom, treatment, procedure, provider type, and lab test. A medical object is an instance of that type (e.g. pain, swelling, and redness are all medical objects of the “symptom” type). Each medical object can have a variety of different attributes and relationships assigned to it, including but not limited to other medical objects (e.g. two different conditions may each have different sets of symptoms assigned to them). Different medical object types can be mapped to standardized ontologies for ease of interpretation. FIG. 12 and FIG. 13 illustrate an exemplary embodiment of standard questions and responses that can be built as medical objects. For example, FIG. 12 illustrates various standard questions that can be presented to a user that are associated with database tables of medical objects relating to symptoms, abilities, and experiences that can have a plurality of attributes associated therewith. For example, in FIG. 13 medical objects relating to questions and responses can include questions/items 240, scoring rules 244 related to the questions/items 240, questionnaire scoring rules 246, patient report outcomes 248, and features 250 that relate to tracking symptoms and optimizing treatments.
[0074] Question templates include a question stem (which can contain one or more text variables, one of which is usually a medical object type) and a set of response options (which can be hard-coded into the template OR populated from a table based on the value of the text variable).")
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Liang in view of Saliman in view of Allen in further view of Ramnani to incorporate the teachings of Heywood to provide a “5. (Original) The method of claim 4, wherein the question in association with the data record including at least one table is associated with one or more of: an item of a medical laboratory examination, an object value for the item, a unit for the object value, a reference value for the item, and an abnormal indication; and” Doing so would Fill gaps in their historical medical profile, as recognized by Heywood. (Paragraph 31).
Claims 6 are rejected under 35 U.S.C. 103 as obvious over US 20180196920 A1, (Liang; Jennifer J.) in view of US 20190228848 A1, (Saliman; Justin) in view of US 20180157990 A1, (Allen; Corville O.) in view of US 20230068338 A1, (Ramnani; Roshni Ramesh) in view of US 20220284722 A1, (PERIYAKARUPPAN; Nandhinee.) in further view of US 20210057069 A1, (WANG; Chenyu.).
Claim 6
Regarding Claim 6, Liang in view of Saliman in view of Allen in view of Ramnani, do not explicitly teach all of 6. (Original) The method of claim 4, wherein the data record including at least one table includes a medical laboratory examination report having the at least one table; and
the data record including at least one paragraph includes a medical imaging report having the at least one paragraph.
However, PERIYAKARUPPAN teach
The method of claim 4, wherein the data record including at least one table includes a medical laboratory examination report having the at least one table; and
(paragraph 35 " A lab reports is created by technicians/radiologists and contains information about diagnosis/results in tabular structure for easy interpretation by humans. A medical invoice contains information present in the table like format with varying size and length. The information is present in descriptions, cost associated with each transaction, procedures, drugs, duration, dosage, units etc. However, it is hard for computers to read the lab reports and/or the medical invoices and extract the relevant information automatically because the traditional approaches are mostly applicable to a fixed structure of documents, i.e. laboratory reports/medical invoice with a fixed structure and with fixed number of tables.")
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified , Liang in view of Saliman in view of Allen in view of Ramnani, to incorporate the teachings of PERIYAKARUPPAN to provide a “The method of claim 4, wherein the data record including at least one table includes a medical laboratory examination report having the at least one table; and” Doing so would Show diagnosis/ results in a tabular structure for easy interpretation by humans, as recognized by PERIYAKARUPPAN. (Paragraph 35).
Liang in view of Saliman in view of Allen in view of Ramnani in further view of PERIYAKARUPPAN, do not explicitly teach all of
the data record including at least one paragraph includes a medical imaging report having the at least one paragraph.
However, WANG teach
the data record including at least one paragraph includes a medical imaging report having the at least one paragraph.
(paragraph 35 "At S104, a paragraph for describing each of the diagnosis items is respectively constructed based on an expanded model of the diagnosis items."
Paragraph 39 "At S105, the medical report of the medical image is generated based on the paragraphs, the keyword sequence, and the diagnosis items."
Paragraph 40 "In this embodiment, the medical report of the medical image may be created after the device for generating the medical report determines the diagnosis items included in the medical image, the paragraphs for describing the diagnosis items, and the keywords corresponding to the diagnosis items. It should be noted that, since the paragraphs of the diagnosis items are sufficiently readable, the medical report may be divided into modules based on the diagnosis items, and each of the module is filled in the corresponding paragraph, that is, the medical report visible to the actual user may only contain the contents of the paragraphs and do not directly reflect the diagnosis items and the keywords. Of course, the generating device may associatedly display the diagnosis items, the keywords, and the paragraphs, so that the user may quickly determine the specific contents of the medical report from the short and refined keyword sequence, and determine his/her own health status through the diagnosis items, and then learn about the evaluation of the health status in detail through the paragraphs, and quickly understand the contents of the medical report from different perspectives, thereby improving the readability of the medical report and the efficiency of information acquisition.")
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified , Liang in view of Saliman in view of Allen in view of Ramnani in further view of PERIYAKARUPPAN, to incorporate the teachings of WANG to provide a “the data record including at least one paragraph includes a medical imaging report having the at least one paragraph.” Doing so would Improve the readability of the medical report, as recognized by WANG. (Paragraph 36).
Claims 7 and 8, and 10 and 11 are rejected under 35 U.S.C. 103 as obvious over US 20180196920 A1, (Liang; Jennifer J.) in view of US 20180157990 A1, (Allen; Corville O.).
Claim 7
Regarding Claim 7, Liang further teaches
selecting, based on the first data type, a first data extraction module associated with the first data type from the plurality of data extraction modules;
(paragraph 29 " An indicator question falls into one of a small number of classes: yes/no, temporal, categorical, etc. The illustrative embodiments provide a dedicated search engine for each of these classes, which given one or more search terms and possible constraints can generate a list of search results. The illustrative embodiments also provide post-processors that take these search results and any constraints not expressible to the search engine, perform filtering and sorting, and extract indicator values."
Paragraph 78 "In one embodiment, patient information extractor 120 provides one or more logical search engines (LSE), which find matches for search terms in the EMR according to specific logic, and one or more search result processors (SRP), which take the output of the LSEs, filter the results according to specified constraints, and return a result. Several LSEs may be implemented by the same physical search engine. The LSEs and SRPs are shown in FIG. 6 and described in further detail below."
Paragraph 120 "" FIG. 6 is a flowchart illustrating operation of a patient information extractor in accordance with an illustrative embodiment. Operation begins for a current query specification and patient EMR (block 600), and the patient information extractor determines whether more query specification blocks (QSB) exist (block 601). Because a query specification has at least one QSB, the first iteration will result in a determination that more QSB exist. The patient information extractor then gets the next QSB in the query specification and extracts the specified search term, logical search engine (LSE), and search result processor (SRP) (block 602). The patient information extractor also extracts from the query specification one or more constraints, such as a date restriction, a note type restriction, a section type restriction, a provider type restriction, or a department/specialty restriction.)
extracting, by the first data extraction module, data from the text of the data record;
(paragraph 120 "FIG. 6 is a flowchart illustrating operation of a patient information extractor in accordance with an illustrative embodiment. Operation begins for a current query specification and patient EMR (block 600), and the patient information extractor determines whether more query specification blocks (QSB) exist (block 601). Because a query specification has at least one QSB, the first iteration will result in a determination that more QSB exist. The patient information extractor then gets the next QSB in the query specification and extracts the specified search term, logical search engine (LSE), and search result processor (SRP) (block 602). The patient information extractor also extracts from the query specification one or more constraints, such as a date restriction, a note type restriction, a section type restriction, a provider type restriction, or a department/specialty restriction."
paragraph 121 "The patient information extractor processes the search term with the specified LSE to search the patient EMR and generate search results (block 603). The patient information extractor then processes the search results with the specified SRP to generate an indicator value (block 604). The patient information extractor determines whether the indicator value is null (i.e., the search results were empty or fail to meet the one or more constraints) (block 605). If the value is null, then the current QSB did not result in generating an indicator value, and operation returns to block 601 to determine whether more QSB exist. The patient information extractor then repeats blocks 602-605 for the next QSB if one exists." )
selecting, based on the second data type, a second data extraction module associated with the second data type from the plurality of data extraction modules; and
(paragraph 29 " An indicator question falls into one of a small number of classes: yes/no, temporal, categorical, etc. The illustrative embodiments provide a dedicated search engine for each of these classes, which given one or more search terms and possible constraints can generate a list of search results. The illustrative embodiments also provide post-processors that take these search results and any constraints not expressible to the search engine, perform filtering and sorting, and extract indicator values."
Paragraph 78 "In one embodiment, patient information extractor 120 provides one or more logical search engines (LSE), which find matches for search terms in the EMR according to specific logic, and one or more search result processors (SRP), which take the output of the LSEs, filter the results according to specified constraints, and return a result. Several LSEs may be implemented by the same physical search engine. The LSEs and SRPs are shown in FIG. 6 and described in further detail below."
Paragraph 120 "" FIG. 6 is a flowchart illustrating operation of a patient information extractor in accordance with an illustrative embodiment. Operation begins for a current query specification and patient EMR (block 600), and the patient information extractor determines whether more query specification blocks (QSB) exist (block 601). Because a query specification has at least one QSB, the first iteration will result in a determination that more QSB exist. The patient information extractor then gets the next QSB in the query specification and extracts the specified search term, logical search engine (LSE), and search result processor (SRP) (block 602). The patient information extractor also extracts from the query specification one or more constraints, such as a date restriction, a note type restriction, a section type restriction, a provider type restriction, or a department/specialty restriction.)
extracting, by the second data extraction module, data from the text of the data record.
(paragraph 120 "FIG. 6 is a flowchart illustrating operation of a patient information extractor in accordance with an illustrative embodiment. Operation begins for a current query specification and patient EMR (block 600), and the patient information extractor determines whether more query specification blocks (QSB) exist (block 601). Because a query specification has at least one QSB, the first iteration will result in a determination that more QSB exist. The patient information extractor then gets the next QSB in the query specification and extracts the specified search term, logical search engine (LSE), and search result processor (SRP) (block 602). The patient information extractor also extracts from the query specification one or more constraints, such as a date restriction, a note type restriction, a section type restriction, a provider type restriction, or a department/specialty restriction."
paragraph 121 "The patient information extractor processes the search term with the specified LSE to search the patient EMR and generate search results (block 603). The patient information extractor then processes the search results with the specified SRP to generate an indicator value (block 604). The patient information extractor determines whether the indicator value is null (i.e., the search results were empty or fail to meet the one or more constraints) (block 605). If the value is null, then the current QSB did not result in generating an indicator value, and operation returns to block 601 to determine whether more QSB exist. The patient information extractor then repeats blocks 602-605 for the next QSB if one exists." )
Liang do not explicitly teach all of 7. (Original) The method of claim 1, wherein the at least one question includes at least a first question associated with a first data type and a second question associated with a second data type different from the first data type, and the method further comprises:
However, Allen teach
7. (Original) The method of claim 1, wherein the at least one question includes at least a first question associated with a first data type and a second question associated with a second data type different from the first data type, and the method further comprises:
(see figure 4 and 5 shows different data types and the questions flow
Paragraph 69 " FIG. 4 is an exemplary table used in accordance with an embodiment of the invention for automating the generation of table-based groundtruth. In this embodiment, a document containing an exemplary table 400 is received by a table-based groundtruth generation system. Parsing operations are performed on the document to extract the table 400, followed by performing additional parsing operations on the extracted table 400 to parse its associated column 410 and row 412 labels. The type of content (e.g., numerical, categorical, percentage, date, time, keywords, entities, etc.) associated with the column 410 and row 412 labels is then identified and assigned as metadata to each cell 414 within the table 400 and to the column 410 and row 412 labels."
Paragraph 70 - 76 "Questions for QA pairs are then automatically generated by applying direct statement templates, as described in greater detail herein, to the contents of the table 400. In various embodiments, use of the direct statement template automatically generates a question for a QA pair by combining words or phrases such as “who,” “what,” “when,” “where,” “how many,” “what percentage,” and so forth, with information contained in column 410 and row 412 labels. For example, based upon the contents of the table 400 shown in FIG. 4, such questions may include: [0071] “What is the mean Age of subjects in the Control (C) group?” [0072] “What is the mean Age of subjects in the Psoriatic (P) group?” [0073] “How many Male subjects are in the Control (C) group?” [0074] “How many Male subjects are in the Psoriatic (P) group?” [0075] “How many White subjects are in the Mild Psoriasis (MP) group?” [0076] “How many Hispanic subjects are in the Severe Psoriasis (SP) group?”")
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Liang to incorporate the teachings of Allen to provide a “7. (Original) The method of claim 1, wherein the at least one question includes at least a first question associated with a first data type and a second question associated with a second data type different from the first data type, and the method further comprises:” Doing so would Improve the code of medical information , as recognized by Allen. (Paragraph 31).
Claim 8
Regarding Claim 8, Liang further teaches
a text classifier associated with a choice data type or a Boolean data type, the text classifier being configured to extract data of the choice data type or the Boolean data type from a paragraph of the text;
(paragraph 29 "An indicator question falls into one of a small number of classes: yes/no, temporal, categorical, etc. The illustrative embodiments provide a dedicated search engine for each of these classes, which given one or more search terms and possible constraints can generate a list of search results. The illustrative embodiments also provide post-processors that take these search results and any constraints not expressible to the search engine, perform filtering and sorting, and extract indicator values."
Paragraph 105 "FIG. 4 is a block diagram of a patient information extractor in accordance with an illustrative embodiment. Each patient is represented by an electronic medical record (EMR) 440. This record consists of (1) a number of clinical notes 441, which are free-text descriptions of patient encounters with medical staff (e.g., office visits), surgical procedures, and interactions between medical professionals concerning the patient, and (2) structured data 442 detailing ordered medications, tests and procedures, and patient demographic and possibly other static information (e.g., lists of allergies). It is likely and desirable, but not required for the purposes of this disclosure, that the EMR has been processed by analytic tools, such as MetaMap, which can detect free-text medical concepts and associate with them concepts in ontology, such as Unified Medical Language System (UM LS) dictionary, for example. The UMLS is a compendium of many controlled vocabularies in the biomedical sciences. It provides a mapping structure among these vocabularies and, thus, allows one to translate among the various terminology systems; it may also be viewed as a comprehensive thesaurus and ontology of biomedical concepts. UMLS further provides facilities for natural language processing. It is intended to be used mainly by developers of systems in medical informatics. Other useful but not required analytic tools would be tools that determine the type of clinical note (e.g., Office Visit, Operative Report, Telephone Encounter, Patient Instructions), and the sections within a note (e.g., Past Medical History, Current Medications, Assessment and Plan)."
Paragraph 108 "In processing a QSB, query specification processor 410 calls the specified logical search engine (LSE) 420 with the specified search term. A “literal” LSE finds instances of the search term. For example, given a search term “heart attack,” the LSE would match the term “heart attack.” A “literal” LSE is similar to a standard find function and may or may not observe capitalization. A “semantic” LSE finds conceptual matches. For example, a “semantic” LSE may match the search term “heart attack” with the matched term “myocardial infarction,” because both terms are mapped to the same concept in UMLS. A “more specific” LSE finds conceptual matches via “ISA” (is a) relations. In knowledge representation, object-oriented programming and design, “ISA” (is_a or is-a) is a subsumption relationship between abstractions (e.g., types, classes), where one class A is a subclass of another class B (and so B is a superclass of A). For example, a “more specific” LSE may match the search term “cancer” with “leukemia,” because in UMLS, leukemia ISA cancer. An “associative” LSE finds terms that co-occur in external corpora. For example, an “associative” LSE may match the search term “asthma” with the matched term “wheezing.” Standard technologies such as latent semantic analysis (LSA) do this. A “logical” LSE finds conceptual matches via relations other than ISA. For example, a “logical” LSE may match the search term “headache” with the matched term “Tylenol,” because Tylenol® TREATS headache, and match the search term “Tylenol” to the matched term “acetaminophen,” because Tylenol is a BRAND_NAME_OF acetaminophen."
Paragraph 111 "A “SpecificIndicator” SRP receives an indicator value as a parameter and performs the following operation: If the hit list qualifies (by size and date range), return the indicator value, else null. A “YesNo” SRP performs the following operation: If the hit list qualifies, return Yes (or 1), else No (or 0). A “FirstLast” SRP receives a flag indicating First/Last as a parameter and performs the following operation: The qualifying hit list is sorted by date and the date of the first/last entry is returned. A “Temporal” SRP can receive as parameters one or two dates, each date with a flag indicating before/after. With this revision, the SRP can return yes/no depending on if the match falls within two specified dates. An “AnswerType” SRP receives an answer type indicating a property of the patient (e.g., age, weight) and performs the following operation: either the answer is looked up in the data structure part of the EMR, or a question-answering system is run on the clinical notes; the specified quantity is returned.")
a question and answers module associated with a string data type or a quantity data type, the question and answers module being configured to extract data of the string data type or the quantity data type from the text as an answer to the question; and
(paragraph 38 "In accordance with yet another illustrative embodiment, which involves a question-answering (QA) pipeline (such as the IBM Watson™ cognitive system), logical search engines (LSEs), which take query terms from the query spec and produce hit lists, may include LSEs that accept query terms in the form of a natural language question. In this case, an LSE itself may be a QA system. Therefore, in this example embodiment, the QA system is called by the mechanisms of the illustrative embodiments."
Paragraph 111 "A “SpecificIndicator” SRP receives an indicator value as a parameter and performs the following operation: If the hit list qualifies (by size and date range), return the indicator value, else null. A “YesNo” SRP performs the following operation: If the hit list qualifies, return Yes (or 1), else No (or 0). A “FirstLast” SRP receives a flag indicating First/Last as a parameter and performs the following operation: The qualifying hit list is sorted by date and the date of the first/last entry is returned. A “Temporal” SRP can receive as parameters one or two dates, each date with a flag indicating before/after. With this revision, the SRP can return yes/no depending on if the match falls within two specified dates. An “AnswerType” SRP receives an answer type indicating a property of the patient (e.g., age, weight) and performs the following operation: either the answer is looked up in the data structure part of the EMR, or a question-answering system is run on the clinical notes; the specified quantity is returned.")
NOTE: a named entity recognition module associated with a table, the named entity recognition module being configured to extract an entity corresponding to an entity category from a table area of the text.
(Due to the claim stating "two of the following" there is no need to map the remaining limitation.)
Liang do not explicitly teach all of 8. (Original) The method of claim 1, wherein the plurality of data extraction modules comprises at least two of the following: a datetime module associated with a datetime data type, the datetime module being configured to extract data of the datetime data type from the text;
However, Allen teach a datetime module associated with a datetime data type, the datetime module being configured to extract data of the datetime data type from the text;
(paragraph 53 "In various embodiments, an input corpus 302 of human-readable text includes one or more documents 304, which in turn may contain unstructured text 306, one or more tables 308, and one or more bullet lists 328. In certain embodiments, the table 308 may include column 310 labels, row 312 labels, associated cells 314 containing structured data, or some combination thereof. In one embodiment, structured data within a table 308, and its corresponding column 310 and row 312 labels, is processed by a table-based groundtruth generation system 250 to generate a QA pair."
Paragraph 60 "The type of content (e.g., numerical, categorical, percentage, date, time, etc.) associated with the column 310 and row 312 labels is then identified and assigned as metadata to each cell 314 within the table 308 (and to the corresponding column 310 and row 312 labels). In certain embodiments, the type of content associated with the column 310 and row 312 labels (and assigned as metadata) can also include keyword and entity information. Entries in the cells can be used to aid identification of the type of content. An entity is a category, such as “city.” “New York” is a keyword and a specific example of the city entity. Questions for QA pairs are then automatically generated by applying direct statement templates to the contents of the table 308. As used herein, a direct statement broadly refers to a statement containing data that is directly stated within a document 304. In certain embodiments, the directly-stated data is contained in a table 308."
Paragraph 61 "] In various embodiments, use of the direct statement template automatically generates a question for a QA pair by combining words or phrases such as “who,” “what,” “when,” “where,” “how many,” “what percentage,” and so forth, with information contained in column 310 and row 312 labels. In certain embodiments, answers for the QA pairs are contained within one or more cells 314 of the table 308 respectively associated with their corresponding column 310 and row 312 labels. In these embodiments, a low initial ranking score (e.g., score=0.1) is assigned to each of the resulting QA pairs to signify their respective significance, as they are generated from direct statements associated with the contents of the table 308 and may not necessarily reflect typical end-user questions. The method by which the initial score is selected and assigned is a matter of design choice.")
See claim 7 for rationale.
Claim 10
Regarding Claim 10, (because claim 8 requires only two listed module types, there is no need to map the remaining alternatives once two are shown)
Claim 11
Regarding Claim 11, Liang in view of Allen, further Liang teaches
11. (Original) The method of claim 8, wherein the question and answers module is configured to determine a data attribute based on the question;
determine a location of one or more characters related to the data attribute in the text;
(paragraph 108 "In processing a QSB, query specification processor 410 calls the specified logical search engine (LSE) 420 with the specified search term. A “literal” LSE finds instances of the search term. For example, given a search term “heart attack,” the LSE would match the term “heart attack.” A “literal” LSE is similar to a standard find function and may or may not observe capitalization. A “semantic” LSE finds conceptual matches. For example, a “semantic” LSE may match the search term “heart attack” with the matched term “myocardial infarction,” because both terms are mapped to the same concept in UMLS. A “more specific” LSE finds conceptual matches via “ISA” (is a) relations. In knowledge representation, object-oriented programming and design, “ISA” (is_a or is-a) is a subsumption relationship between abstractions (e.g., types, classes), where one class A is a subclass of another class B (and so B is a superclass of A). For example, a “more specific” LSE may match the search term “cancer” with “leukemia,” because in UMLS, leukemia ISA cancer. An “associative” LSE finds terms that co-occur in external corpora. For example, an “associative” LSE may match the search term “asthma” with the matched term “wheezing.” Standard technologies such as latent semantic analysis (LSA) do this. A “logical” LSE finds conceptual matches via relations other than ISA. For example, a “logical” LSE may match the search term “headache” with the matched term “Tylenol,” because Tylenol® TREATS headache, and match the search term “Tylenol” to the matched term “acetaminophen,” because Tylenol is a BRAND_NAME_OF acetaminophen.")
and extract data of the string data type or the quantity data type from the text based on the location.
Paragraph 111 "A “SpecificIndicator” SRP receives an indicator value as a parameter and performs the following operation: If the hit list qualifies (by size and date range), return the indicator value, else null. A “YesNo” SRP performs the following operation: If the hit list qualifies, return Yes (or 1), else No (or 0). A “FirstLast” SRP receives a flag indicating First/Last as a parameter and performs the following operation: The qualifying hit list is sorted by date and the date of the first/last entry is returned. A “Temporal” SRP can receive as parameters one or two dates, each date with a flag indicating before/after. With this revision, the SRP can return yes/no depending on if the match falls within two specified dates. An “AnswerType” SRP receives an answer type indicating a property of the patient (e.g., age, weight) and performs the following operation: either the answer is looked up in the data structure part of the EMR, or a question-answering system is run on the clinical notes; the specified quantity is returned.")
Claims 9 are rejected under 35 U.S.C. 103 as obvious over US 20180196920 A1, (Liang; Jennifer J.) in view of US 20180157990 A1, (Allen; Corville O.) in further view US 20180341871 A1, (MAITRA; Anutosh)
Claim 9
Regarding Claim 9, Liang in view of Allen do not explicitly teach all of
The method of claim 8, wherein the text classifier is configured to select a paragraph in the text of the data record; segment the paragraph into multiple sentences; determine a data attribute based on the question; determine a semantical relationship between each of the multiple sentences and the data attribute; and determine an answer to the question based on the semantical relationship.
However, MAITRA teach
9. (Original) The method of claim 8, wherein the text classifier is configured to select a paragraph in the text of the data record;
(paragraph 57 "If the question is a list question type, the question answering platform may utilize a strategy to extract sufficient information as the answer for a list question type. For example, the question answering platform may filter candidate paragraphs based on list information and sizes of the candidate paragraphs, and may score each candidate paragraph based on the following equation:
S.sub.h(Q,A)=w.sub.1*SS+w.sub.2*DR+w.sub.3*TC+w.sub.4*NG+w.sub.5*LCS.
The question answering platform may select the candidate paragraph with the maximum score, and may extract the sentences from the selected paragraph using sentence segmentation. The question answering platform may score each sentence based on the following equation:
S(Q,A)=W×S.sub.c(Q,A)+V×S.sub.h(Q,A).
The question answering platform may select sentences having a score greater than a predetermined threshold value, and may utilize the selected sentences to generate the answer to the question.")
segment the paragraph into multiple sentences;
(paragraph 57 "If the question is a list question type, the question answering platform may utilize a strategy to extract sufficient information as the answer for a list question type. For example, the question answering platform may filter candidate paragraphs based on list information and sizes of the candidate paragraphs, and may score each candidate paragraph based on the following equation:
S.sub.h(Q,A)=w.sub.1*SS+w.sub.2*DR+w.sub.3*TC+w.sub.4*NG+w.sub.5*LCS.
The question answering platform may select the candidate paragraph with the maximum score, and may extract the sentences from the selected paragraph using sentence segmentation. The question answering platform may score each sentence based on the following equation:
S(Q,A)=W×S.sub.c(Q,A)+V×S.sub.h(Q,A).
The question answering platform may select sentences having a score greater than a predetermined threshold value, and may utilize the selected sentences to generate the answer to the question.")
determine a data attribute based on the question;
(paragraph 21 "In some implementations, the question answering platform may extract a focus of the question. The focus of the question may include a word or a sequence of words that defines the question and disambiguates the question (e.g., indicates what the question is looking for). The focus of the question may be contained within a noun phrase of the question, and the noun phrase may indicate what the question is expecting an answer to do. In the case of an imperative question, a direct object of a question word may contain the focus. In the case of an interrogatory question, there may be certain natural language dependencies that capture a relation between a question word and the focus."
Paragraph 23 "In some implementations, the question answering platform may classify the question. The question answering platform may classify questions into decision questions and non-decision questions. Decision questions may include yes or no answers, while non-decision questions may require specific answers varying in length from possibly a single word to a few paragraphs. Decision questions may appear in different lexical constructs, such as “be” questions (e.g., is, are, was, were, and/or the like), “do” questions (e.g., do, does, did, and/or the like), modal questions (e.g., can, will, shall, and/or the like), has, have, had, or the like questions, and/or the like. Non-decision questions may be further classified into sub-categories, such as interrogatives (e.g., what, how, why, which, where, and/or the like), imperatives (e.g., describe, provide, justify, list, and/or the like), and/or the like. Further, each lexical construct can be sub-divided based on the answer types expected, such as time, person, location, descriptive, measure, and/or the like. A hierarchical structure in question taxonomy may then be evident.")
determine a semantical relationship between each of the multiple sentences and the data attribute; and
(paragraph 57 "If the question is a list question type, the question answering platform may utilize a strategy to extract sufficient information as the answer for a list question type. For example, the question answering platform may filter candidate paragraphs based on list information and sizes of the candidate paragraphs, and may score each candidate paragraph based on the following equation:
S.sub.h(Q,A)=w.sub.1*SS+w.sub.2*DR+w.sub.3*TC+w.sub.4*NG+w.sub.5*LCS.
The question answering platform may select the candidate paragraph with the maximum score, and may extract the sentences from the selected paragraph using sentence segmentation. The question answering platform may score each sentence based on the following equation:
S(Q,A)=W×S.sub.c(Q,A)+V×S.sub.h(Q,A).
The question answering platform may select sentences having a score greater than a predetermined threshold value, and may utilize the selected sentences to generate the answer to the question."
Paragraph 50 "The semantic similarity score (SS) technique may determine a semantic representation of the question (e.g., a word vector VEC(Q)) using word vector averaging, as follows: …"
"where q may represent the question, VEC(t.sub.i) may represent a word vector of word t.sub.i, and number of lookups may represent a number of words in the question for which word embeddings are available. The semantic similarity score technique may determine a word vector (e.g., VEC(A)) for the candidate answer in a similar manner. The semantic similarity score technique may calculate a cosine similarity between the question word vector and the candidate answer word vector as follows:
SS=cosine(VEC(Q),VEC(A)).")
determine an answer to the question based on the semantical relationship.
(paragraph 57 "If the question is a list question type, the question answering platform may utilize a strategy to extract sufficient information as the answer for a list question type. For example, the question answering platform may filter candidate paragraphs based on list information and sizes of the candidate paragraphs, and may score each candidate paragraph based on the following equation:
S.sub.h(Q,A)=w.sub.1*SS+w.sub.2*DR+w.sub.3*TC+w.sub.4*NG+w.sub.5*LCS.
The question answering platform may select the candidate paragraph with the maximum score, and may extract the sentences from the selected paragraph using sentence segmentation. The question answering platform may score each sentence based on the following equation:
S(Q,A)=W×S.sub.c(Q,A)+V×S.sub.h(Q,A).
The question answering platform may select sentences having a score greater than a predetermined threshold value, and may utilize the selected sentences to generate the answer to the question."
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Liang in view of Allen to incorporate the teachings of MAITRA to provide a “The method of claim 8, wherein the text classifier is configured to select a paragraph in the text of the data record; segment the paragraph into multiple sentences; determine a data attribute based on the question; determine a semantical relationship between each of the multiple sentences and the data attribute; an determine an answer to the question based on the semantical relationship.” Doing so would Improve speed and efficiency of the process and conserve computing resource, as recognized by MAITRA. (Paragraph 59).
Claims 12 are rejected under 35 U.S.C. 103 as obvious over US 20180196920 A1, (Liang; Jennifer J.) in view of US 20170249646 A1, (Krupa; Wojciech.)
Claim 12
Regarding Claim 12, Liang do not explicitly teach all of
12. (Original) The method of claim 1, further comprising outputting the questionnaire filled with the extracted data.
However, Krupa teach
The method of claim 1, further comprising outputting the questionnaire filled with the extracted data.
(paragraph 24 "In an example embodiment, a platform is provided that allows for automatic prepopulating of fields of a questionnaire for a job application. This platform may contain numerous components, which collectively may be known as the “easy apply” system. Questions from different questionnaires may be normalized into a single version of the question. These normalized questions may be stored in a data object in a database. Upon detecting an indication from a member of a social networking service, the normalized questions may be retrieved from the database and used to automatically extract information related to the normalized questions from the member's social networking profile, as well as extract information related to the normalized questions from other sources. This information may then be used to automatically prepopulate fields in the exact questionnaire corresponding to the job for which the member is interested in applying."
Paragraph 76 "At operation 1214, it is determined if this is the last question record in the questionnaire mapping record, and if not, the process loops back up to operation 1206 for the next question record in the questionnaire mapping record. If so, however, then the process proceeds to operation 1216, where the answers provided to the normalized questions are presented to the user via the easy apply front-end for validation. Here, the user may review the answers (many of which are likely automatically populated) and indicate that it is permissible to send this information to the third party providing the questionnaire. At operation 1218, a determination is made as to whether the answers provided to the normalized questions answered all of the questions in the questionnaire. If not, then at operation 1220, the member is prompted to answer any unanswered questions in the questionnaire via the easy apply front-end." )
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Liang to incorporate the teachings of Krupa to provide a “12. (Original) The method of claim 1, further comprising outputting the questionnaire filled with the extracted data.” Doing so would Reduce the amount of time to answer a questionnaire, as recognized by Krupa. (Paragraph 5).
Claims 13 are rejected under 35 U.S.C. 103 as obvious over US 20180196920 A1, (Liang; Jennifer J.) in further view US 20220309109 A1, (BENINCASA; Gregorio)
Claim 13
Regarding Claim 13,Liang do not explicitly teach all of 13. (Currently amended) The method of any of claims 1-12 claim 1,wherein the selected data extraction module is generated by training a machine learning model with a training data set of the data type, the training data set includes standard data obtained from a plurality of data records.
However, BENINCASA teach
The method of any of claims 1-12 claim 1,wherein the selected data extraction module is generated by training a machine learning model with a training data set of the data type, the training data set includes standard data obtained from a plurality of data records.
(paragraph 9 "In embodiments, the questions types may comprise two or more of: a point extraction question type, a section extraction question type and a table extraction question type."
Paragraph 101 " A blank (untrained) machine learning (ML) model is trained to extract “answers” to a single user-defined question. Such models may be referred to herein as data extraction models. When a user defines a new question, a new—and, at that point, blank—data extraction model is created and will be trained on the single task of answering that question. The user can assign a question identifier to each question and may choose to include information about the question in the question identifier. However, that information is for his own benefit (and, where applicable, the benefit of other users)—it is not required by the data extraction model to understand the question or how to answer it. Rather, the data extraction model learns how to answer the single question to which it relates from labelled examples in a training process. Each labelled example takes the form of a labelled token in a tokenized representation of a training document. That is, each token of the training document is associated with a label. In the examples that follow, the label is a binary label which identifies the token as either relevant to the specific question (positive example) or not relevant to the specific question (negative example). However, other labelling schemes can be used to indicate token relevance, such as BIO (beginning-inside-outside) or BILOU (beginning-inside-last-outside-unit). Labelling schemes such as BIO and BILOU can be used to additionally indicate the relative position of a token within a relevant chunk (and to distinguish chunks of multiple tokens from “unit”, i.e. single-token, chunks). Although the following description uses binary labelling as an example, the description applies equally to other labelling schemes. The document extraction system learns how to answer the specific question from multiple such labelled training documents.")
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Liang to incorporate the teachings of BENINCASA to provide a “13. (Currently amended) The method of any of claims 1-12 claim 1,wherein the selected data extraction module is generated by training a machine learning model with a training data set of the data type, the training data set includes standard data obtained from a plurality of data records.” Doing so would Achieve better accuracy and tend to somewhat more flexible range of document, as recognized by BENINCASA. (Paragraph 5).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ALI M HASSAN whose telephone number is (571)272-5331. The examiner can normally be reached Monday - Friday 8:00am - 4:00pm.
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, Paras Shah can be reached at (571)270-1650. 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.
/ALI M HASSAN/ Examiner, Art Unit 2653
/Paras D Shah/ Supervisory Patent Examiner, Art Unit 2653
08/03/2026
Prosecution Insights