A doctor professional ability training system and method based on AI semantic analysis of medical records

Abstract

The present invention discloses a doctor's professional ability training system based on AI semantic analysis of medical records, including: a medical record storage module for storing sample electronic medical records and newly added electronic medical records; a semantic annotation module for reading sample electronic medical records and annotating the symptom descriptions and diagnosis results therein with entities, attributes, and relationships for neural network training to form a knowledge graph; a neural network module for reading annotated sample electronic medical records for training, for semantic annotation of entities, attributes, and relationships of newly added electronic medical records, and for determining the results of doctor training; a doctor training module for reading the symptom descriptions in newly added electronic medical records, and receiving the diagnosis results and matching results input during doctor training. The present invention also discloses a doctor's professional ability training method based on AI semantic analysis of medical records.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a doctor's professional ability training system and method based on medical record AI semantic analysis. Background Art

[0002] As for the improvement of doctors' professional ability, they currently rely mainly on learning from various books, which are relatively dogmatic and lack the combination with practice. As a result, doctors are too stereotyped in diagnosis and lack the combination with the actual situation. If they learn through existing medical records, they can often only read the medical records. Due to the lack of a special training system, the learning efficiency is low. Summary of the invention

[0003] In order to solve the problems existing in the prior art, the present invention provides a doctor's professional ability training system and method based on medical record AI semantic analysis.

[0004] The technical solution adopted by the present invention is:

[0005] A doctor professional ability training system based on AI semantic analysis of medical records, including:

[0006] A medical record storage module, comprising a training sample medical record storage unit and a newly added medical record storage unit, wherein the training sample medical record storage unit is used to store sample electronic medical records input for neural network training, and the newly added medical record storage unit is used to store newly added electronic medical records input;

[0007] The semantic annotation module is used to read sample electronic medical records and annotate the entity, attribute, and relationship of the disease description and diagnosis results therein for neural network training to form a knowledge graph;

[0008] A neural network module is used to read annotated sample electronic medical records for training, to read newly added electronic medical records and automatically perform semantic annotation of entities, attributes and relationships on the symptom descriptions and diagnosis results in the newly added electronic medical records in combination with the knowledge graph, and to automatically perform semantic annotation of entities, attributes and relationships on the diagnosis results input during doctor training in combination with the knowledge graph, and to compare and match the annotations of the input diagnosis results with the original annotations of the diagnosis results in the newly added electronic medical records;

[0009] The doctor training module is used to read the disease description in the newly added electronic medical record and receive the diagnosis results and matching results input during the doctor training;

[0010] The display module is used to display the disease description in the newly added electronic medical record read by the doctor training module, the diagnosis results input during the doctor training, and the matching results output by the neural network module.

[0011] Preferably, the semantic annotation module is further used to annotate entities with the same or similar meanings in the symptom descriptions in the sample electronic medical records with synonyms or near-synonyms, and annotate entities with a hyponym-hypernym relationship.

[0012] Preferably, the neural network module is also used to determine that the diagnosis is successful if the annotation of the input diagnosis result is the same as the original annotation of the diagnosis result of the newly added electronic medical record or belongs to a synonym, near-synonym or hyponymous entity relationship, otherwise it is determined as a diagnosis failure.

[0013] Preferably, the display module displays the diagnosis result of the newly added electronic medical record while displaying the matching result.

[0014] A method for training doctors' professional competence based on AI semantic analysis of medical records, comprising the following steps:

[0015] Read sample electronic medical records;

[0016] Semantically annotate the entities, attributes, and relationships of the disease descriptions and diagnosis results in the sample electronic medical records and construct a knowledge graph;

[0017] Use knowledge graphs to train neural networks;

[0018] Read newly added electronic medical records;

[0019] Use the trained neural network combined with the knowledge graph to automatically perform semantic annotation of entities, attributes, and relationships on the disease descriptions and diagnosis results in the newly added electronic medical records;

[0020] Read the disease description in the newly added electronic medical record for display for doctor training, receive and display the diagnosis results input during training, use the trained neural network combined with the knowledge graph to automatically semantically annotate the input diagnosis results with entities, attributes and relationships, compare and match the annotations of the input diagnosis results with the original annotations of the diagnosis results in the newly added electronic medical record, and display the matching results.

[0021] Preferably, the method further comprises the following steps:

[0022] Entities with the same or similar meanings in the symptom descriptions in the sample electronic medical records are marked with synonyms or near-synonyms, and entities with a hyponym-hypernym relationship are marked with a hyponym-hypernym entity relationship.

[0023] Preferably, the method further comprises the following steps:

[0024] If the annotation of the input diagnosis result is the same as the original annotation of the diagnosis result of the newly added electronic medical record or is a synonym, near-synonym or a hyponymous entity relationship, the diagnosis is displayed as correct, otherwise it is displayed as incorrect.

[0025] Preferably, the method further comprises the following steps:

[0026] The diagnosis results of the newly added electronic medical records are displayed while displaying the matching results.

[0027] The beneficial effects of the present invention are: doctors are trained through a large number of real medical records, and natural language analysis can be realized by using neural networks and knowledge graphs, so that the diagnosis results input during doctor training can be judged more accurately. Compared with traditional answering software, which often requires the input of diagnosis results that are the same as standard answers, the misjudgment rate is greatly reduced. Doctors can completely input natural language to answer questions according to their personal habits, reducing the learning cost of doctors. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a structural schematic diagram of an embodiment of the present invention;

[0029] Figure numerals: 1. medical record storage module, 2. semantic annotation module, 3. neural network module, 4. doctor training module, 5. medical record collection module, 6. display module, 7. sample medical record storage unit, 8. newly added medical record storage unit. DETAILED DESCRIPTION

[0030] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0031] Example

[0032] like Figure 1 As shown, a doctor professional ability training system based on AI semantic analysis of medical records includes:

[0033] A medical record storage module, comprising a training sample medical record storage unit and a newly added medical record storage unit, wherein the training sample medical record storage unit is used to store sample electronic medical records input for neural network training, and the newly added medical record storage unit is used to store newly added electronic medical records input;

[0034] The semantic annotation module is used to read sample electronic medical records and annotate the entity, attribute, and relationship of the disease description and diagnosis results therein for neural network training to form a knowledge graph;

[0035] A neural network module is used to read annotated sample electronic medical records for training, to read newly added electronic medical records and automatically perform semantic annotation of entities, attributes and relationships on the symptom descriptions and diagnosis results in the newly added electronic medical records in combination with the knowledge graph, and to automatically perform semantic annotation of entities, attributes and relationships on the diagnosis results input during doctor training in combination with the knowledge graph, and to compare and match the annotations of the input diagnosis results with the original annotations of the diagnosis results in the newly added electronic medical records;

[0036] The doctor training module is used to read the disease description in the newly added electronic medical record and receive the diagnosis results and matching results input during the doctor training;

[0037] A display module is used to display the disease description in the newly added electronic medical record read by the doctor training module, the diagnosis result input during the doctor training, and the matching result output by the neural network module;

[0038] Medical record collection module, used to input new electronic medical records.

[0039] Specifically, the semantic annotation module is also used to annotate entities with the same or similar meanings in the symptom descriptions in the sample electronic medical records with synonyms or near-synonyms, and annotate entities with upper and lower entity relationships.

[0040] Specifically, the neural network module is also used to determine that the diagnosis is successful if the annotation of the input diagnosis result is the same as the original annotation of the diagnosis result of the newly added electronic medical record or belongs to a synonym, near-synonym or hyponymous entity relationship, otherwise it is determined that the diagnosis is a failure.

[0041] Specifically, the display module displays the diagnosis result of the newly added electronic medical record while displaying the matching result.

[0042] A method for training doctors' professional competence based on AI semantic analysis of medical records, comprising the following steps:

[0043] Read sample electronic medical records;

[0044] Semantically annotate the entities, attributes, and relationships of the disease descriptions and diagnosis results in the sample electronic medical records and construct a knowledge graph;

[0045] Use knowledge graphs to train neural networks;

[0046] Read newly added electronic medical records;

[0047] Use the trained neural network combined with the knowledge graph to automatically perform semantic annotation of entities, attributes, and relationships on the disease descriptions and diagnosis results in the newly added electronic medical records;

[0048] Read the disease description in the newly added electronic medical record for display for doctor training, receive and display the diagnosis results input during training, use the trained neural network combined with the knowledge graph to automatically semantically annotate the input diagnosis results with entities, attributes and relationships, compare and match the annotations of the input diagnosis results with the original annotations of the diagnosis results in the newly added electronic medical record, and display the matching results.

[0049] Specifically, the following steps are also included:

[0050] Entities with the same or similar meanings in the symptom descriptions in the sample electronic medical records are marked with synonyms or near-synonyms, and entities with a hyponym-hypernym relationship are marked with a hyponym-hypernym entity relationship.

[0051] Specifically, the following steps are also included:

[0052] If the annotation of the input diagnosis result is the same as the original annotation of the diagnosis result of the newly added electronic medical record or is a synonym, near-synonym or a hyponymous entity relationship, the diagnosis is displayed as correct, otherwise it is displayed as incorrect.

[0053] Specifically, the following steps are also included:

[0054] The diagnosis results of the newly added electronic medical records are displayed while displaying the matching results.

[0055] By digitizing medical records and directly using real medical records to train doctors' professional abilities, doctors' learning efficiency has been greatly improved.

[0056] The above-mentioned sample annotation process and principle belong to the prior art for neural network training, and there is no need to elaborate. The present invention mainly utilizes the characteristic that the same diagnosis result often corresponds to the same or similar symptoms, while different diagnosis results often correspond to different or dissimilar symptoms, and pre-annotates the symptoms and diagnosis results of the sample electronic medical records to form a knowledge graph to train the neural network, and then uses the trained neural network to quickly annotate the symptoms and diagnosis results of the newly added electronic medical records to further improve the knowledge graph. The newly added electronic medical records can be used as a training question bank after annotation. Of course, the same electronic medical records can also be used as a training question bank. The most important thing about using neural networks combined with knowledge graphs is to support natural language analysis. Compared with traditional training software, which is often just a simple multiple-choice question or a fill-in-the-blank question that requires the input of the same standard answer, the present invention can support natural language, that is, the doctor can completely input text according to personal habits, and then rely on the neural network and the knowledge graph to analyze and judge the text input by the doctor, so that the judgment of the diagnosis results input during the doctor's training is more accurate. Compared with traditional training software, the misjudgment rate and the learning cost of the doctor are greatly reduced.

[0057] The above-mentioned embodiments only express the specific implementation of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A doctor professional ability training system based on AI semantic analysis of medical records, characterized by: include: A medical record storage module, comprising a training sample medical record storage unit and a newly added medical record storage unit, wherein the training sample medical record storage unit is used to store sample electronic medical records input for neural network training, and the newly added medical record storage unit is used to store newly added electronic medical records input; The semantic annotation module is used to read sample electronic medical records and annotate the entity, attribute, and relationship of the symptom descriptions and diagnosis results therein for training the neural network to form a knowledge graph. When annotating entities, the entities with the same or similar meanings in the symptom descriptions in the sample electronic medical records are annotated with synonyms or near-synonyms, and the entities with upper and lower entity relationships are annotated with upper and lower entity relationships; A neural network module is used to read annotated sample electronic medical records for training, to read newly added electronic medical records and automatically perform semantic annotation of entities, attributes and relationships on the symptom descriptions and diagnosis results in the newly added electronic medical records in combination with the knowledge graph, and to automatically perform semantic annotation of entities, attributes and relationships on the diagnosis results input during doctor training in combination with the knowledge graph, and to compare and match the annotation of the input diagnosis result with the original annotation of the diagnosis result of the newly added electronic medical record. If the annotation of the input diagnosis result is the same as the original annotation of the diagnosis result of the newly added electronic medical record or is a synonym, near synonym or a hyponym entity relationship, then it is determined that the diagnosis is successful; otherwise, it is determined that the diagnosis fails; The doctor training module is used to read the disease description in the newly added electronic medical record and receive the diagnosis results and matching results input during the doctor training; The display module is used to display the disease description in the newly added electronic medical record read by the doctor training module, the diagnosis results input during the doctor training, and the matching results output by the neural network module.

2. The doctor professional ability training system based on medical record AI semantic analysis according to claim 1 is characterized in that: The display module displays the diagnosis result of the newly added electronic medical record while displaying the matching result.

3. A method for training doctors' professional competence based on AI semantic analysis of medical records, characterized in that: The steps include: Read sample electronic medical records; Semantically annotate the entities, attributes, and relationships of the disease descriptions and diagnosis results in the sample electronic medical records and construct a knowledge graph; Use knowledge graphs to train neural networks; Read newly added electronic medical records; Use the trained neural network combined with the knowledge graph to automatically semantically annotate the entity, attribute and relationship of the symptom description and diagnosis results in the newly added electronic medical records. When annotating entities, annotate the entity relationship of synonyms or near synonyms for the entities with the same or similar meanings in the symptom descriptions in the sample electronic medical records, and annotate the entity relationship of the upper and lower entities for the entities with the upper and lower relationships; Read the disease description in the newly added electronic medical record for display for doctor training, receive and display the diagnosis results input during training, use the trained neural network combined with the knowledge graph to automatically semantically annotate the entities, attributes and relationships of the input diagnosis results, compare and match the annotations of the input diagnosis results with the original annotations of the diagnosis results in the newly added electronic medical record, and display the matching results. If the annotations of the input diagnosis results are the same as the original annotations of the diagnosis results in the newly added electronic medical record or are synonyms, near-synonyms or hyponymous entity relationships, then the diagnosis is displayed as correct; otherwise, the diagnosis is displayed as wrong.

4. The method for training doctors' professional competence based on AI semantic analysis of medical records according to claim 3 is characterized in that: The following steps are also included: The diagnosis results of the newly added electronic medical records are displayed while displaying the matching results.

Images