CN114722213A - Knowledge graph construction and application method of multi-disease multi-guideline clinical assistant decision support system - Google Patents
Knowledge graph construction and application method of multi-disease multi-guideline clinical assistant decision support system Download PDFInfo
- Publication number
- CN114722213A CN114722213A CN202210237471.9A CN202210237471A CN114722213A CN 114722213 A CN114722213 A CN 114722213A CN 202210237471 A CN202210237471 A CN 202210237471A CN 114722213 A CN114722213 A CN 114722213A
- Authority
- CN
- China
- Prior art keywords
- tumor
- treatment
- model
- index
- establishing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 201000010099 disease Diseases 0.000 title claims abstract description 22
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 22
- 238000000034 method Methods 0.000 title claims description 20
- 238000010276 construction Methods 0.000 title claims description 9
- 238000011282 treatment Methods 0.000 claims abstract description 154
- 206010028980 Neoplasm Diseases 0.000 claims abstract description 151
- 239000003814 drug Substances 0.000 claims abstract description 118
- 229940079593 drug Drugs 0.000 claims abstract description 83
- 206010067484 Adverse reaction Diseases 0.000 claims abstract description 31
- 230000006838 adverse reaction Effects 0.000 claims abstract description 31
- 238000012795 verification Methods 0.000 claims description 14
- 230000002265 prevention Effects 0.000 claims description 11
- 206010013700 Drug hypersensitivity Diseases 0.000 claims description 8
- 201000005311 drug allergy Diseases 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000012790 confirmation Methods 0.000 claims description 2
- 238000009795 derivation Methods 0.000 claims description 2
- 230000004927 fusion Effects 0.000 claims description 2
- 238000012545 processing Methods 0.000 claims description 2
- 230000001960 triggered effect Effects 0.000 claims 1
- 208000032612 Glial tumor Diseases 0.000 description 8
- 206010018338 Glioma Diseases 0.000 description 8
- 238000003745 diagnosis Methods 0.000 description 8
- JMANVNJQNLATNU-UHFFFAOYSA-N oxalonitrile Chemical compound N#CC#N JMANVNJQNLATNU-UHFFFAOYSA-N 0.000 description 7
- 238000001959 radiotherapy Methods 0.000 description 7
- 238000011269 treatment regimen Methods 0.000 description 7
- GQYIWUVLTXOXAJ-UHFFFAOYSA-N Lomustine Chemical compound ClCCN(N=O)C(=O)NC1CCCCC1 GQYIWUVLTXOXAJ-UHFFFAOYSA-N 0.000 description 5
- 229960002247 lomustine Drugs 0.000 description 5
- 238000002560 therapeutic procedure Methods 0.000 description 5
- AOJJSUZBOXZQNB-TZSSRYMLSA-N Doxorubicin Chemical compound O([C@H]1C[C@@](O)(CC=2C(O)=C3C(=O)C=4C=CC=C(C=4C(=O)C3=C(O)C=21)OC)C(=O)CO)[C@H]1C[C@H](N)[C@H](O)[C@H](C)O1 AOJJSUZBOXZQNB-TZSSRYMLSA-N 0.000 description 4
- 229940041181 antineoplastic drug Drugs 0.000 description 4
- 238000005259 measurement Methods 0.000 description 4
- CPTBDICYNRMXFX-UHFFFAOYSA-N procarbazine Chemical compound CNNCC1=CC=C(C(=O)NC(C)C)C=C1 CPTBDICYNRMXFX-UHFFFAOYSA-N 0.000 description 4
- 229960000624 procarbazine Drugs 0.000 description 4
- 238000013515 script Methods 0.000 description 4
- 230000009885 systemic effect Effects 0.000 description 4
- OGWKCGZFUXNPDA-XQKSVPLYSA-N vincristine Chemical compound C([N@]1C[C@@H](C[C@]2(C(=O)OC)C=3C(=CC4=C([C@]56[C@H]([C@@]([C@H](OC(C)=O)[C@]7(CC)C=CCN([C@H]67)CC5)(O)C(=O)OC)N4C=O)C=3)OC)C[C@@](C1)(O)CC)CC1=C2NC2=CC=CC=C12 OGWKCGZFUXNPDA-XQKSVPLYSA-N 0.000 description 4
- 229960004528 vincristine Drugs 0.000 description 4
- OGWKCGZFUXNPDA-UHFFFAOYSA-N vincristine Natural products C1C(CC)(O)CC(CC2(C(=O)OC)C=3C(=CC4=C(C56C(C(C(OC(C)=O)C7(CC)C=CCN(C67)CC5)(O)C(=O)OC)N4C=O)C=3)OC)CN1CCC1=C2NC2=CC=CC=C12 OGWKCGZFUXNPDA-UHFFFAOYSA-N 0.000 description 4
- 201000010133 Oligodendroglioma Diseases 0.000 description 3
- 239000002246 antineoplastic agent Substances 0.000 description 3
- 238000012217 deletion Methods 0.000 description 3
- 230000037430 deletion Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 229960002949 fluorouracil Drugs 0.000 description 3
- 230000002980 postoperative effect Effects 0.000 description 3
- 206010073128 Anaplastic oligodendroglioma Diseases 0.000 description 2
- GHASVSINZRGABV-UHFFFAOYSA-N Fluorouracil Chemical compound FC1=CNC(=O)NC1=O GHASVSINZRGABV-UHFFFAOYSA-N 0.000 description 2
- 230000003474 anti-emetic effect Effects 0.000 description 2
- 239000002111 antiemetic agent Substances 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 229930003827 cannabinoid Natural products 0.000 description 2
- 239000003557 cannabinoid Substances 0.000 description 2
- 210000000349 chromosome Anatomy 0.000 description 2
- 239000000890 drug combination Substances 0.000 description 2
- 238000002651 drug therapy Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 102000015694 estrogen receptors Human genes 0.000 description 2
- 108010038795 estrogen receptors Proteins 0.000 description 2
- 230000002489 hematologic effect Effects 0.000 description 2
- 208000029824 high grade glioma Diseases 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 201000011614 malignant glioma Diseases 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000001575 pathological effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000002271 resection Methods 0.000 description 2
- 230000003319 supportive effect Effects 0.000 description 2
- WJFKNYWRSNBZNX-UHFFFAOYSA-N 10H-phenothiazine Chemical compound C1=CC=C2NC3=CC=CC=C3SC2=C1 WJFKNYWRSNBZNX-UHFFFAOYSA-N 0.000 description 1
- 208000026310 Breast neoplasm Diseases 0.000 description 1
- 102100021899 Cyclin-L2 Human genes 0.000 description 1
- 208000030453 Drug-Related Side Effects and Adverse reaction Diseases 0.000 description 1
- 101000897452 Homo sapiens Cyclin-L2 Proteins 0.000 description 1
- 102100037845 Isocitrate dehydrogenase [NADP], mitochondrial Human genes 0.000 description 1
- 206010047700 Vomiting Diseases 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 229940009456 adriamycin Drugs 0.000 description 1
- 201000010134 adult oligodendroglioma Diseases 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 229940125644 antibody drug Drugs 0.000 description 1
- 210000001130 astrocyte Anatomy 0.000 description 1
- 239000003693 atypical antipsychotic agent Substances 0.000 description 1
- 229940127236 atypical antipsychotics Drugs 0.000 description 1
- 239000000090 biomarker Substances 0.000 description 1
- 229960000106 biosimilars Drugs 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 229940065144 cannabinoids Drugs 0.000 description 1
- 239000000969 carrier Substances 0.000 description 1
- 210000003169 central nervous system Anatomy 0.000 description 1
- 208000025997 central nervous system neoplasm Diseases 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000002512 chemotherapy Methods 0.000 description 1
- 239000003246 corticosteroid Substances 0.000 description 1
- 229960001334 corticosteroids Drugs 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 229960004679 doxorubicin Drugs 0.000 description 1
- 238000001647 drug administration Methods 0.000 description 1
- 238000009261 endocrine therapy Methods 0.000 description 1
- 229940034984 endocrine therapy antineoplastic and immunomodulating agent Drugs 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000012458 free base Substances 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 229940125721 immunosuppressive agent Drugs 0.000 description 1
- 239000003018 immunosuppressive agent Substances 0.000 description 1
- 238000010253 intravenous injection Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 239000000820 nonprescription drug Substances 0.000 description 1
- 230000000474 nursing effect Effects 0.000 description 1
- 229940000673 orphan drug Drugs 0.000 description 1
- 239000002859 orphan drug Substances 0.000 description 1
- 238000007911 parenteral administration Methods 0.000 description 1
- 229950000688 phenothiazine Drugs 0.000 description 1
- 150000002990 phenothiazines Chemical class 0.000 description 1
- 230000035935 pregnancy Effects 0.000 description 1
- 238000004393 prognosis Methods 0.000 description 1
- 239000003369 serotonin 5-HT3 receptor antagonist Substances 0.000 description 1
- 238000009120 supportive therapy Methods 0.000 description 1
- 238000001356 surgical procedure Methods 0.000 description 1
- 238000002626 targeted therapy Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 210000001519 tissue Anatomy 0.000 description 1
- 238000011277 treatment modality Methods 0.000 description 1
- 239000000439 tumor marker Substances 0.000 description 1
- 230000008673 vomiting Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Animal Behavior & Ethology (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Pathology (AREA)
- Physics & Mathematics (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The tumor knowledge map applied to a multi-disease and multi-guideline clinical assistant decision support system is provided aiming at the current situation that the knowledge map aiming at the tumor diseases is lacked in the current medical field, a tumor index body model, a tumor treatment scheme model, a tumor drug and adverse reaction model and a tumor treatment rule model are respectively established aiming at the particularity and complexity of the tumor diseases, the four models are integrated, through the linkage of the four models, a knowledge source and assistant judgment are provided for assistant decision of the tumor diseases, and a user is helped to make a decision quickly.
Description
Technical Field
The invention relates to the field of medical artificial intelligence application, in particular to a construction method and specific application of a tumor knowledge graph.
Background
With the development of the idea of everything interconnection and artificial intelligence technology, the concept of Knowledge Graph (Knowledge Graph) was proposed by google in 2012 and successfully applied to search engines. In the medical field, useful medical knowledge is extracted from massive medical documents and data, inference logic is completed, and doctors are assisted to make the fastest and best decision without leaving the graph searching and graph calculating capacity of knowledge graphs. There are also many knowledge maps in the medical field at home and abroad, such as large knowledge maps with reasoning ability like Freebase and DBpedia; domestic knowledge graphs for multiple diseases such as CMeKG, Chinese medicine knowledge graph and the like. Related products based on medical knowledge maps, such as International Business Machines (IBM) Waton abroad and medical intelligence benefits of hundreds of degrees in China, which appear in the market at present are based on medical knowledge map technologies, clinical assistant decision-making systems are established, and the medical level of primary hospitals is improved. But most are decision systems designed for one disease. The knowledge graph in the medical field mainly takes concepts as main parts, has single relation and incomplete structure, and is in a research and exploration stage.
The construction of the knowledge graph in the application is to carry out modeling, data combing and importing on the knowledge graph according to the concept of combining the body and the service requirement on the content of guidelines, specifications, documents and the like related to the medical field, develop a knowledge graph interface and assist clinical auxiliary decision-making to support service query.
Disclosure of Invention
In view of the above, the present application aims to construct a set of medical knowledge maps suitable for various tumors and an application method in practical scenes.
In one aspect, the present application provides a method for constructing a tumor knowledge graph for multiple diseases and multiple guidelines, comprising:
step 1, designing a tumor-related knowledge map framework. The method comprises 4 tumor subgraph model designs: (1) designing a tumor index body model framework; mainly refers to the tumor-related indexes, the index logic, the index mutual-deduction rule, the structure of the indexes and the mapping between the indexes and the international standard terms. (2) Designing a tumor treatment scheme model framework; mainly relates to the relationship among treatment schemes, guidelines, documents, drugs, medication cycles, medication doses, and the like. (3) Designing a tumor drug and adverse reaction model framework; mainly refers to the drugs, drug structures, incompatibility caused by drugs and corresponding prevention and treatment schemes related to tumor treatment. (4) Designing a tumor treatment rule model framework; expressions relating primarily to indices have evolved different treatment regimes under different treatment conditions.
And 2, acquiring and expressing tumor related knowledge. Extracting entities from reference data such as guidelines, specifications, documents and the like related to the medical field by combining manpower with NLP (non-line segment) and establishing a triple relation;
and 3, fusing knowledge. Based on the structures of multiple guidelines and multiple disease species, carrying out knowledge deduplication and merging on the 4 tumor subgraphs, and integrating into a tumor-related knowledge map model;
and 4, storing the knowledge graph. Selecting a suitable knowledge graph database to store the knowledge graph in step 3.
In another aspect of the present application, a method for applying a multi-disease multi-guideline tumor knowledge graph is provided, which comprises:
step 1, checking knowledge of a knowledge graph in a mode of combining manpower and Python scripts;
and 2, developing a knowledge graph query interface, deploying and releasing.
Has the advantages that:
the tumor knowledge map applied to a multi-disease and multi-guideline clinical assistant decision support system is provided aiming at the current situation that the knowledge map aiming at the tumor diseases is lacked in the current medical field, a tumor index body model, a tumor treatment scheme model, a tumor drug and adverse reaction model and a tumor treatment rule model are respectively established aiming at the particularity and complexity of the tumor diseases, the four models are integrated, through the linkage of the four models, a knowledge source and assistant judgment are provided for assistant decision of the tumor diseases, and a user is helped to make a decision quickly.
Drawings
FIG. 1 is a diagram of an index ontology;
FIG. 2 is a diagram of the treatment protocol body;
FIG. 3 is a view showing the structure of the main body of the medicine;
FIG. 4 is a diagram of a rule ontology structure.
Detailed Description
The invention will now be described in detail by way of exemplary embodiments with reference to the accompanying drawings. It should be understood, however, that the described embodiments are merely one application scenario for the present invention, and not all. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present invention without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the invention provides a construction method of an assistant decision support system, which is used for assistant diagnosis decision of diseases based on a tumor knowledge graph. Wherein the tumor knowledge map comprises: the tumor index body model, the tumor treatment scheme model, the tumor medicament and adverse reaction model and the tumor treatment rule model are constructed to obtain the tumor knowledge map based on the fusion of the four models.
Step one, selecting a disease species and a reference guideline. Taking glioma as an example, 5 guidelines are referred to, namely "clinical practice guideline for tumor of central nervous system of NCCN", "diagnosis and treatment guideline for astrocyte and oligodendroglioma of adults" of European neurooncology Association (EANO) "," clinical practice guideline for treating diffuse glioma of adults with CGCG "," diagnosis and treatment guideline for glioma of central nervous system of China (CCNS) ", and" diagnosis and treatment norm for glioma of brain (NHC) ", respectively.
And step two, constructing a tumor knowledge map.
And step three, checking the accuracy of the tumor knowledge graph. And performing data content verification on the knowledge graph content stored in the graph database in a mode of combining manual verification and Python script verification, wherein the data content verification comprises entity verification, attribute verification, relationship verification, spelling verification and the like.
And step four, developing, deploying and releasing an interface of the knowledge graph. And compiling a query script of clinical knowledge content by using Python and graph database query languages according to the service requirements of different scenes, and developing a knowledge query interface. Such as fuzzy search scheme interface, scheme detail interface, medicine fuzzy matching interface, atlas scheme comparison, disease course abstract, index recommendation, treatment mode fuzzy query, etc.
Another embodiment of the present invention will be described in detail with respect to the construction of a tumor knowledge map.
Step 1, constructing a tumor index ontology model.
And S101, index acquisition. And (4) finishing a set of glioma index system by using the guide file mentioned in the step one in a mode of combining manual work and NLP. The relationship between the indices forms index logic, shown in the spreadsheet as IF { index A and value and/or index B and value } THEN { index C and value }. For example, one index in breast tumors, the "gestation period", exists on the premise that the value of the other index, the "sex", is "female".
S102, setting derivation rules. There is a relationship that is derived between the tumor markers, for example, the markers "percent estrogen receptor < 1" and "estrogen receptor status negative" are sufficient requirements.
S103, setting an index structure. Different tumors have partially identical indexes and partially different indexes. It is necessary to establish a set of index structures suitable for all tumor diseases. Referring to SNOMED CT, LOINC, ICD and other standards, tumor indexes are divided into nine major categories, such as personal information, disease/diagnosis/scale, clinical manifestation, operation/operation, treatment and nursing, adverse reaction, curative effect/follow-up/prognosis, management data, quantifier/limit value and the like. There are also multilayer structural relationships in different broad categories, such as: clinical presentation-assisted examination-imaging examination-MRI imaging-MRI findings.
And S104, setting the attribute of the index. The attributes of the indexes are mainly divided into index names, Chinese and English, code IDs, data types, units, value ranges of the indexes, index display logics, index mutual-deduction rules and the like according to ontology knowledge and business requirements.
And S105, standard processing of indexes. Since the index names in different guidelines are different and the measurement units of the index are also different due to measuring instruments and the like, it is necessary to standardize the index when creating the index body model. The index terminology set used in this application refers primarily to LOINC (Logical Observation Identifiers Names and Codes), including laboratory (Lab), Clinical (Clinical), information attachment (Attachments), and questionnaire (survey), and encompasses hematology, serology, vital signs, radiology reports, tumor marker Codes, and other types of observations. The term set of the measurement Units used for the index value is UCUM (Unified Code for Units of Measure), which is also the term set of the measurement Units used in LOINC. And mapping the indexes in the S101 to a standard term set one by one, and mapping the measurement units of the index values to a UCUM term set.
And S106, establishing a relation between the index and the index structure. The index is the index acquired in S101, and the index structure is the index structure designed according to the medical logic in S103. The index is connected to the upper structure of the index using the relationship "has _ belongTo". For example, the index "MRI diagnosis" is related to the last layer index structure "MRI examination finding" mentioned in the example of S103: clinical presentation-assisted examination-imaging examination-MRI imaging-MRI findings- (has _ belongTo) -MRI diagnosis.
And 2, constructing a tumor treatment scheme model.
S201, obtaining a tumor treatment scheme. Structured documents of treatment protocols are acquired using manual interpretation guidelines to comb knowledge and NLP extraction entities.
S202, establishing a relation between the treatment scheme and the medicine. And establishing a ternary relation among the treatment scheme, the usage of the medicine and the dosage of the medicine. Specifically, since the same drug may be used in different regimens, the amount of drug used may be the same, or may be different. Therefore, it is necessary to establish a single-dose regimen first and then to connect with the regimen mentioned in the manual. For example, there is one scheme in NCCN: AC-T, firstly establishing a single-medicine scheme: scheme a, scheme C and scheme T. The AC-T protocol was then linked to the 3 single drugs described above using "has _ subtreatmentAnd". "-" represents sequential, meaning that after the simultaneous use of drug A and drug C is finished, drug T is used. Adding a group to the edge attribute can interpret the sequence.
S203, establishing a multi-phalangeal multi-disease structure. The same regimen may appear in different guidelines for different disease species, and may also correspond to different stages of treatment, documentation, recommended levels, rules, etc. In order to express the structure more clearly, an entity of 'treatment condition' needs to be established at the step, the entity attribute comprises codes of guide, disease, literature, and the like, and the relationship is established with the entity corresponding to the codes. For example, 5 guidelines for glioma refer to a systemic treatment regimen: "standard radiotherapy combined with lomustine + procarbazine + vincristine", as shown in the following table:
for the same protocol, each line is a "treatment condition", and the implementation protocol corresponds to the description in the guideline. The above examples are only a part of the display of "treatment conditions", and information such as disease codes, evidence levels, guideline remarks and the like can be included in the practical application process.
And S204, establishing a non-drug treatment scheme structure. The tumor treatment scheme can be divided into six categories, namely, systemic treatment (chemotherapy, targeted therapy and endocrine therapy), operation, radiotherapy, other local minimally invasive and physical treatment, interventional therapy, local drug therapy, clinical experiment scheme and the like. Except for systemic and local drug therapy, other treatment regimens are drug-free. The structure of a non-drug treatment regimen mainly includes the treatment modality, the treatment site, and the treatment technique. For radiotherapy, a non-drug treatment plan, information such as radiotherapy target area, radiotherapy period and dose is added in the treatment plan structure.
And 3, constructing a tumor medicament and an adverse reaction model.
S301, obtaining the tumor medicine. More than 600 drugs for tumor therapy and supportive therapy are collected from multiple tumor therapy guidelines (e.g., NCCN, ASCO, CSCO, ESMO, etc.) and drug websites (e.g., FDA, NCI pharmacopoeia, medical insurance drug catalog, etc.).
S302, establishing a medicine structure. Drugs were classified in the following 22 classification schemes: antibody drugs, antineoplastic drug treatment categories, approved drugs, biomarker specific drugs, biosimilar drugs, classical hematologic drugs, labor cessation drugs, local action drugs, carriers for drugs, cessation drugs, hematologic drugs, immunosuppressive agents, test drugs, orphan drugs, over-the-counter drugs, pediatric oncology drugs, radiotherapy drugs, routed drugs, site-independent drugs, site-specific drugs, supportive drugs, defense tissue based drugs, and the like. Each class comprises a multilayered drug structure, e.g. the antineoplastic drug treatment class-cytotoxic chemotherapy-antimetabolites-pyrimidine analogs-fluoropyrimidines.
S303, setting the drug property. The Chinese and English, the alias and the abbreviation of the medicine are involved, and whether the medicine is included in the medical insurance information or not is also involved.
S304, standardizing the medicines. Drugs are widely known by various names, and in different guidelines, different hospitals may have different names, especially drug names, such as doxorubicin, also known as "adriamycin". To unify drug names, the present application uniformly maps drug names to the international drug nomenclature standard RxNorm.
S305, establishing a drug and adverse reaction entity. The adverse reaction of the drugs mainly refers to the vomiting reaction caused by the dosage of the drugs, and is mainly divided into two parts, namely adverse reaction risk caused by single drug dosage and caused by incompatibility of a plurality of drugs. According to the first edition of 'NCCN clinical practice antiemetic guideline for tumor' in 2019, the risk of adverse reaction generated by tumor drugs is divided into 7 categories of high, medium, low, micro, high, medium, low, micro and the like, and the drug administration mode mainly refers to oral administration and intravenous injection. Combining the medicines, the adverse reaction risks and the medicine administration modes to obtain all adverse reaction entities of the medicines. The attributes include limit types, limit values, units, and the like, and the metering taboos of the drugs are structured.
And S306, establishing a prevention and control scheme. Aiming at the risk of adverse reaction caused by the medicine in S305, a prevention and treatment scheme is provided in 'NCCN clinical practice antiemetic guideline', so that the risk of adverse reaction is reduced. Namely, the prevention and treatment scheme is only related to the risk of adverse reaction caused by the medicament, and has no direct relation with the disease diagnosis and the treatment scheme used by the patient. The prevention and treatment scheme is also composed of drug combinations (most of the drug combinations are supportive drugs), and information such as the usage amount of the drug is also available. In fact, the structure of the prevention and treatment scheme is the same as that of the treatment scheme in step 2, and the construction can refer to the treatment scheme structure in the tumor treatment scheme model.
S307, establishing a relation among the medicines, the medicine types, the adverse drug reactions and the prevention and treatment schemes. For example, the drug "fluorouracil", can be related to the example mentioned in S302: antineoplastic drug treatment categories-cytotoxic chemotherapy-antimetabolites-pyrimidine analogs-fluoropyrimidine-fluorouracil. As long as "fluorouracil" is infused intravenously in a treatment regimen, adverse effects "low risk of parenteral administration" are produced, and for this risk, corresponding control regimens such as "corticosteroids" or "cannabinoids" or "phenothiazines" are specified.
It is further noted that "has _ ADRs" is used between the single drugs mentioned in S305 and adverse reactions; incompatibility exists among a plurality of medicines, the medicines are connected with a incompatibility node through has _ incompatibility, and then has _ ADRs are still used between the incompatibility and adverse reaction. The two adverse reactions are connected with a prevention and treatment scheme by using 'has _ identification treatment'.
And 4, constructing a tumor treatment rule model.
S401, obtaining a treatment rule. The rules are read from the guide and entered in a predetermined format. For example, the glioma NCCN guideline CLIO-1 contains the following rules: (image examination, flat scan MRI | | | image examination, enhanced MRI ═ prompt high-grade glioma & & multidisciplinary assessment whether maximum safe resection can be performed & & whether recurrence is equal to no, infer an operative plan: safe excision of the maximum range.
In the above rules, "image examination", "MRI image performance", "multidisciplinary evaluation whether or not maximum safe resection is possible", and "whether or not recurrence" are all indexes, "flat scan MRI", "enhanced MRI", "prompt high-grade glioma", "yes", and "no" are all values corresponding to the indexes, and both the indexes and the values of the indexes are from the index ontology model;
the inference is that the protocol, in the rules, is encoded with a unique identifier, and the specific protocol details are from the treatment protocol ontology model. The index and value, treatment regimen combination constitute different rules. For example, the first rule on GLIO-2 in glioma NCCN: (WHO 4+ anaplastic oligodendroglioma | | WHO 5 classification ═ oligodendroglioma, IDH mutation with combined deletion type of 1p/19q (pathological grade: grade III)) & &1 chromosome short arm (1p) ═ deletion & &19 chromosome long arm (19q) ═ deletion & & whether relapse ═ no, the reasoning result contains the systemic treatment regimen "lomustine + procarbazine + vincristine".
S402, adding treatment conditions. The "treatment conditions" in the tumor treatment protocol model also need to be added in the "tumor treatment rules model". Because the same rule addition plus different treatment phases may lead to different treatment regimes. Each rule entity corresponds to different 'treatment conditions', and mainly comprises a treatment stage, a recommendation level, an evidence level and the like. The following differences exist from the "treatment conditions" in the tumor treatment protocol model: (1) disease information does not exist, each guide corresponds to a set of rules, the guide and the disease are in many-to-one relationship, and if the guide is determined, the disease is uniquely determined; (2) the remarks are different. The remarks in the tumor treatment scheme are mainly directed at the dosage of the medicament; the remarks in the treatment rules are mainly the attention points of the inferred treatment scheme used in combination with other operations (such as radiotherapy and surgery). (3) The treatment stages differ. The tumor treatment protocol is mainly based on knowledge storage, all suitable treatment stages are stored, and indexes in the treatment rules imply the treatment stages, for example, "WHO 4+ classification ═ anaplastic oligodendroglioma" indicates that the pathological result is postoperative, then the rules comprising "WHO 4+ classification" should be postoperative, and the inferred treatment protocol is only "postoperative" in the above-mentioned treatment rule model comprising "WHO 4+ classification" even if stored "preoperatively and postoperatively" in the treatment protocol model.
And step 5, combining the tumor index body model, the tumor rule model and the tumor treatment scheme model with the tumor medicament and adverse reaction model to generate a set of knowledge maps with complete functions, namely the tumor knowledge maps. The tumor treatment rules model closely links the other three models:
(1) the indexes and index values in the tumor treatment rule model are derived from the tumor index ontology model. As an example of the treatment rule given in S401, the index involved in the rule and the index value are both from the tumor index ontology model. (2) The reasoning result in the tumor treatment rule model is derived from the tumor treatment scheme model. The reasoning result of the formula "lomustine + procarbazine + vincristine" is from tumor treatment scheme models. (3) Drugs in tumor treatment regimen models, such as "lomustine", "procarbazine", "vincristine", come from tumor drugs and adverse reaction models, where the use of "lomustine" has adverse reactions "oral administration is of moderate to high risk", for which 16 prevention regimens are included, such as "breakthrough therapy, 5-HT3 antagonist + atypical antipsychotic", "breakthrough therapy, cannabinoid + phenothiazine", etc.
And step 6, storing the tumor knowledge map. The tumor knowledge map is mainly compiled and imported into a map database JanusGraph through Python scripts after pre-research and screening.
Through the six steps, the tumor knowledge graph is established and obtained, and the knowledge graph is embedded into a clinical assistant decision support system, so that the user can be helped to quickly inquire and accurately judge. Another embodiment of the present invention will be described in detail with respect to the construction and use of a tumor knowledge map.
Constructing a tumor knowledge map:
establishing a connection relation among a tumor index body model, a tumor treatment scheme model, a tumor medicament and adverse reaction model and a tumor treatment rule model; wherein: (1) establishing a relation between the tumor treatment rule model and the tumor index ontology model through the tumor indexes; (2) establishing a relation between the tumor treatment rule model and the tumor treatment scheme model through reasoning conclusions; (3) the tumor treatment scheme model establishes a connection with the tumor medicament and the adverse reaction model through the medicament and the medicament dosage contained in the treatment scheme.
Use of tumor knowledge map:
developing and deploying a query interface for each model for a user to query; the method comprises the following steps: step 1, receiving patient indexes and index values, medication situations and drug allergy histories input by a user;
step 2, matching the tumor index and the index value in the tumor treatment rule model, and judging whether a corresponding 'treatment condition' exists; if yes, entering step 3, and if not, entering step 4;
step 3, carrying out secondary screening on the screened treatment conditions according to the input medication condition and the drug allergy history, and judging whether corresponding treatment conditions exist; if yes, sorting the 'treatment conditions' meeting the patient indexes and index values, the medication condition and the drug allergy history according to the rule recommendation levels corresponding to the treatment rules and returning the sorted treatment conditions to the user; if not, entering step 4;
step 4, matching in the tumor index ontology model according to the patient indexes input by the user to obtain all indexes and index structures containing the index names; according to the medication condition and the drug allergy history input by the user, matching treatment conditions in the tumor treatment scheme model to obtain treatment conditions corresponding to the medication condition of the patient, and returning the treatment conditions to the user according to the rule recommendation level sequence corresponding to the treatment rules.
The mechanism for perfecting and updating the tumor knowledge map comprises the following steps: after step 4, the "treatment condition" confirmed by the user and the patient index and index value are added as a new "treatment condition" to the tumor treatment rule model according to the confirmation result of the user. In addition, the tumor knowledge map also comprises a self-updating mechanism, and according to a preset time interval, the tumor knowledge map triggers the updating of a tumor index body model, a tumor treatment scheme model, a tumor medicament and adverse reaction model and a tumor treatment rule model according to historical clinical medical records and multi-guideline data.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (7)
1. A knowledge graph construction method of a multi-disease multi-guideline clinical assistant decision support system is characterized by comprising the following steps:
s01, designing a tumor knowledge map framework, which comprises a tumor index body model, a tumor treatment scheme model, a tumor drug and adverse reaction model and a tumor treatment rule model;
s02, collecting relevant knowledge of different tumor diseases in multiple guidelines, establishing a tumor knowledge graph according to a tumor knowledge graph framework designed in S01 after fusion and duplication removal operations, and storing the tumor knowledge graph by using a graph database;
s03, carrying out accuracy verification on the tumor knowledge map established in the S02, wherein the accuracy verification comprises entity verification, attribute verification, relation verification and spelling verification;
and S04, developing and deploying a query interface of the tumor knowledge graph.
2. The method of claim 1, wherein the tumor index ontology model is established by the method comprising the following steps:
step 1, extracting indexes of different tumor disease types and logic relations among the indexes from a guide;
step 2, establishing a derivation rule among tumor indexes;
step 3, establishing a multi-level tree-shaped index structure and index attributes which are universal for tumor diseases;
step 4, carrying out standardization processing on indexes collected from the multi-guide, and mapping the indexes to standard terms;
and 5, establishing a connection relation between the index and the index structure.
3. The method of claim 1, wherein the establishment of the tumor treatment protocol model comprises the steps of:
step 1, extracting a structured file of a tumor treatment scheme in multiple guidelines;
step 2, extracting drug entities in the treatment scheme, and establishing a triple relation of treatment scheme-drug dosage;
step 3, establishing a 'treatment condition' entity corresponding to records of each treatment scheme appearing in different guidelines, including drug treatment scheme names, source guidelines, all applicable treatment stages and scheme recommendation levels;
and 4, establishing a non-drug treatment scheme structure comprising a treatment mode, a treatment part and a treatment technology.
4. The method of claim 1, wherein the tumor drug and adverse reaction model building process comprises the following steps:
step 1, extracting tumor treatment medicine data from multiple tumor treatment guidelines and medicine websites;
step 2, establishing a multi-level tree-shaped medicine structure and medicine attributes which are universal for tumor diseases;
step 3, standardizing the extracted medicine names and mapping the medicine names to standard terms;
step 4, extracting adverse reactions and prevention and treatment schemes of different medicines;
and 5, establishing a relation among the medicines, the medicine types, the adverse reactions and the prevention and treatment schemes.
5. The method of claim 1, wherein the process of establishing the tumor treatment rule model comprises the steps of:
step 1, obtaining tumor treatment rules, wherein each treatment rule comprises a combination of a tumor index, an index value and an inference conclusion;
step 2, establishing a treatment condition entity corresponding to records of each treatment scheme appearing in different guidelines, including treatment rule names, source guidelines, treatment stages and rule recommendation levels.
6. The method of claim 1, wherein the tumor knowledge map creation process further comprises the steps of:
step 1, establishing a connection relation among a tumor index body model, a tumor treatment scheme model, a tumor medicament and adverse reaction model and a tumor treatment rule model; wherein:
establishing a relation between the tumor treatment rule model and the tumor index ontology model through the tumor indexes;
establishing a relation between the tumor treatment rule model and the tumor treatment scheme model through reasoning conclusions;
the tumor treatment scheme model establishes a relation with the tumor medicament and the adverse reaction model through the medicament and the medicament dosage contained in the treatment scheme;
step 2, developing and deploying a query interface for each model for a user to query; wherein:
step 2.1, receiving patient indexes and index values, medication situations and drug allergy histories input by a user;
step 2.2, matching the tumor index and the index value in the tumor treatment rule model, and judging whether a corresponding 'treatment condition' exists; if yes, entering step 2.3, and if not, entering step 2.4;
step 2.3, carrying out secondary screening on the screened treatment conditions according to the input medication condition and the drug allergy history, and judging whether corresponding treatment conditions exist; if yes, sorting the 'treatment conditions' meeting the patient indexes and index values, the medication condition and the drug allergy history according to the rule recommendation levels corresponding to the treatment rules and returning the sorted treatment conditions to the user; if not, entering step 2.4;
step 2.4, matching in the tumor index ontology model according to the patient indexes input by the user to obtain all indexes and index structures containing the index names; according to the medication condition and the drug allergy history input by the user, matching treatment conditions in the tumor treatment scheme model to obtain treatment conditions corresponding to the medication condition of the patient, and returning the treatment conditions to the user according to the rule recommendation level sequence corresponding to the treatment rules.
7. The method of claim 6, further comprising the steps of tumor knowledge-map self-refinement and knowledge update:
the self-improvement step comprises: after step 2.4 is finished, according to the confirmation result of the user, adding the 'treatment condition' confirmed by the user, the patient index and the index value as a new 'treatment condition' into the tumor treatment rule model;
the knowledge updating step comprises the following steps: according to a preset time interval, according to historical clinical medical records and multi-guideline data, the updating of a tumor index body model, a tumor treatment scheme model, a tumor drug and adverse reaction model and a tumor treatment rule model is triggered.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210237471.9A CN114722213A (en) | 2022-03-11 | 2022-03-11 | Knowledge graph construction and application method of multi-disease multi-guideline clinical assistant decision support system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210237471.9A CN114722213A (en) | 2022-03-11 | 2022-03-11 | Knowledge graph construction and application method of multi-disease multi-guideline clinical assistant decision support system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114722213A true CN114722213A (en) | 2022-07-08 |
Family
ID=82236792
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210237471.9A Pending CN114722213A (en) | 2022-03-11 | 2022-03-11 | Knowledge graph construction and application method of multi-disease multi-guideline clinical assistant decision support system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114722213A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116737945A (en) * | 2023-05-10 | 2023-09-12 | 百洋智能科技集团股份有限公司 | Mapping method for EMR knowledge map of patient |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130096946A1 (en) * | 2011-10-13 | 2013-04-18 | The Board of Trustees of the Leland Stanford, Junior, University | Method and System for Ontology Based Analytics |
| US20200242148A1 (en) * | 2017-09-29 | 2020-07-30 | Koninklijke Philips N.V. | Natural language processing using ontology mapping |
| CN112466463A (en) * | 2020-12-10 | 2021-03-09 | 求臻医学科技（北京）有限公司 | Intelligent answering system based on tumor accurate diagnosis and treatment knowledge graph |
| CN113221541A (en) * | 2021-07-09 | 2021-08-06 | 清华大学 | Data extraction method and device |
| CN113628734A (en) * | 2021-08-02 | 2021-11-09 | 浙江海心智惠科技有限公司 | Design method of oncology electronic medical advice system with clinical decision intelligent recommendation function |
-
2022
- 2022-03-11 CN CN202210237471.9A patent/CN114722213A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130096946A1 (en) * | 2011-10-13 | 2013-04-18 | The Board of Trustees of the Leland Stanford, Junior, University | Method and System for Ontology Based Analytics |
| US20200242148A1 (en) * | 2017-09-29 | 2020-07-30 | Koninklijke Philips N.V. | Natural language processing using ontology mapping |
| CN112466463A (en) * | 2020-12-10 | 2021-03-09 | 求臻医学科技（北京）有限公司 | Intelligent answering system based on tumor accurate diagnosis and treatment knowledge graph |
| CN113221541A (en) * | 2021-07-09 | 2021-08-06 | 清华大学 | Data extraction method and device |
| CN113628734A (en) * | 2021-08-02 | 2021-11-09 | 浙江海心智惠科技有限公司 | Design method of oncology electronic medical advice system with clinical decision intelligent recommendation function |
Non-Patent Citations (1)
| Title |
|---|
| 崔洁;陈德华;乐嘉锦;: "基于EMR的乳腺肿瘤知识图谱构建研究", 计算机应用与软件, no. 12, 15 December 2017 (2017-12-15) * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116737945A (en) * | 2023-05-10 | 2023-09-12 | 百洋智能科技集团股份有限公司 | Mapping method for EMR knowledge map of patient |
| CN116737945B (en) * | 2023-05-10 | 2024-05-07 | 百洋智能科技集团股份有限公司 | Mapping method for EMR knowledge map of patient |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2020243732A1 (en) | Systems and methods of clinical trial evaluation | |
| JP6588584B2 (en) | Anonymized medical information retrieval support system using structured medical data integrated management database | |
| Konopka | Biomedical ontologies—a review | |
| CN1839404A (en) | Method for computerising and standardizing medical information | |
| CN114255884A (en) | Hypertension drug treatment knowledge graph construction method and device | |
| WO2021238436A1 (en) | Multi-drug sharing query method, mobile terminal and storage medium | |
| Basu et al. | Call for data standardization: lessons learned and recommendations in an imaging study | |
| Zhang et al. | Computable eligibility criteria through ontology-driven data access: a case study of hepatitis C virus trials | |
| Jonnagaddala et al. | Integration and analysis of heterogeneous colorectal cancer data for translational research | |
| CN114091676A (en) | Causal relationship determination method | |
| CN114722213A (en) | Knowledge graph construction and application method of multi-disease multi-guideline clinical assistant decision support system | |
| CN115938565A (en) | Tumor assistant decision support system based on intelligent reasoning and evidence correlation | |
| US20190073403A1 (en) | Querying data using master terminology data model | |
| Scapicchio et al. | DICOM-MIABIS integration model for biobanks: a use case of the EU PRIMAGE project | |
| Schulz et al. | Representing oncology in datasets: standard or custom biomedical terminology? | |
| Sander et al. | Integrating terminologies into standard SQL: a new approach for research on routine data | |
| Liu et al. | PatientEG dataset: bringing event graph model with temporal relations to electronic medical records | |
| Humphrey | A knowledge-based expert system for computer-assisted indexing | |
| Fung et al. | The use of inter-terminology maps for the creation and maintenance of value sets | |
| Harman et al. | Standardized mapping of nursing assessments across 59 US military treatment facilities | |
| Bousquet et al. | Ontological and non-ontological resources for associating medical dictionary for regulatory activities terms to SNOMED clinical terms with semantic properties | |
| Sun et al. | Using NLP in openEHR archetypes retrieval to promote interoperability: a feasibility study in China | |
| KR20100086404A (en) | Clinical contents structure and the clinical contents modeling method | |
| Yan et al. | Terminology and ontology development for semantic annotation: A use case on sepsis and adverse events | |
| Berlanga et al. | Medical data integration and the semantic annotation of medical protocols |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |