Intelligent diagnosis and treatment system based on artificial intelligence

Abstract

The invention discloses an intelligent diagnosis and treatment method based on artificial intelligence, and the method comprises the steps: obtaining the registration information inputted by a patient and the currently inputted diagnosis and treatment record, obtaining the historical diagnosis and treatment record in a case library, obtaining the current diagnosis and treatment information inputted by a doctor and expert experience knowledge, and carrying out the case analysis of the historical diagnosis and treatment record. Providing a diagnosis and treatment scheme reference according to an analysis result, adding a body feature tag to the user, predicting a current disease of the patient based on department information selected by the patient, performing preliminary correlation sorting on historical diagnosis and treatment records of the patient, and performing secondary sorting on the historical diagnosis and treatment records of the patient based on the currently input diagnosis and treatment records. And receiving whether a reference scheme needs to be made, updating diagnosis and treatment information of a patient in real time, acquiring diagnosis and treatment actions of a doctor, analyzing diagnosis and treatment habits of the doctor, and recommending a personalized intelligent analysis result to the doctor. The problems that historical diagnosis and treatment resources are wasted and the retrieval result is not intelligent are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent diagnosis and treatment, and particularly to an intelligent diagnosis and treatment system based on artificial intelligence. Background Art

[0002] In contemporary society, the application of artificial intelligence technology in the field of intelligent medical diagnosis has made breakthrough progress, which not only significantly improves the quality and efficiency of medical services, but also brings a more accurate diagnosis and treatment experience to patients. However, despite the increasing accumulation of medical data, the storage and application of these data in the database have not been fully utilized. Patients often need to describe their conditions to doctors again during each visit. Due to the lack of professional medical knowledge, patients often have difficulty providing clear and accurate descriptions of their conditions. This situation not only increases the workload of doctors during the diagnosis process, but also may lead to diagnostic errors and affect the diagnosis and treatment effect of patients. In addition, a large amount of information stored in the medical database is idle and fails to effectively assist doctors in making more accurate diagnostic decisions. This not only wastes diagnostic resources, but also fails to fully utilize the potential of data in improving the conditions of patients. Therefore, it is necessary to design an intelligent diagnosis and treatment system based on artificial intelligence that can effectively utilize diagnostic resources and improve the work efficiency of doctors. Summary of the Invention

[0003] The purpose of the present invention is to provide an intelligent diagnosis and treatment system based on artificial intelligence to solve the problems raised in the above background art.

[0004] To solve the above technical problems, the present invention provides the following technical solution: An intelligent diagnosis and treatment system based on artificial intelligence, which executes an intelligent diagnosis and treatment method based on artificial intelligence. The running steps of the method include:

[0005] Step S1: Obtain the registration information input by the patient and the current input diagnosis and treatment record, obtain the historical diagnosis and treatment records in the case library, and obtain the current diagnosis and treatment information input by the doctor and the expert experience knowledge;

[0006] Step S2: Conduct case analysis on the historical diagnosis and treatment records, provide a reference for the diagnosis and treatment plan according to the analysis results, and add physical feature tags to the user;

[0007] Step S3: Based on the department information selected by the patient, predict the current disease of the patient, and conduct a preliminary correlation ranking on the historical diagnosis and treatment records of the patient according to the prediction results;

[0008] Step S4: Conduct a secondary ranking on the historical diagnosis and treatment records of the patient based on the current input diagnosis and treatment record, and receive whether the historical diagnosis and treatment record is needed as a reference;

[0009] Step S5: Update the patient's diagnosis and treatment information in real time, obtain the doctor's diagnosis and treatment actions, analyze the doctor's diagnosis and treatment habits, and recommend personalized intelligent analysis results to the doctor.

[0010] According to the above technical solution, step S1 further includes the following steps:

[0011] Step S11: The registration information includes: ID card information, selected department information. The historical diagnosis and treatment records include: the patient's historical diagnosis and treatment records, other patients' diagnosis and treatment records, with corresponding inspection reports attached. The current diagnosis and treatment information refers to the description of the patient's current medical condition;

[0012] Step S12: Clean the historical diagnosis and treatment records, identify and delete duplicate diagnosis and treatment records, use regular expressions to remove irrelevant characters and spaces in the text, and standardize the diagnosis and treatment record format using DICOM;

[0013] Step S13: When obtaining the historical diagnosis and treatment records, protect the patient's privacy, process the historical diagnosis and treatment records using data desensitization and data encryption technologies, and display the diagnosis and treatment information of other patients anonymously with the ID card number partially hidden.

[0014] According to the above technical solution, step S2 further includes the following steps:

[0015] Step S21: Based on the historical diagnosis and treatment records, adopt the model technology based on Transformer to intelligently segment the key information of the historical diagnosis and treatment record text, perform intelligent text word segmentation on the historical diagnosis and treatment record text, convert the key information of the disease description and treatment plan into a word vector sequence through the pre-trained vector model GloVe, and use a bidirectional LSTM neural network and an attention mechanism to integrate the word vectors to extract the diagnosis and treatment feature values of the diagnosis and treatment records, and construct a diagnosis and treatment feature vector H = [h1, h2,..., h n , where h i represents the i-th word vector in the historical diagnosis and treatment record description sentence, and n represents the number of word vectors in the historical diagnosis and treatment record description sentence;

[0016] Step S22: The intelligent deep learning terminal analyzes the historical diagnosis and treatment records, classifies the diseases, stores the diagnosis and treatment records of the same type of diseases of the same patient in one document. The intelligent deep learning terminal tracks the development of the disease and the treatment plan, and learns the optimal treatment plan according to different treatment methods for the same type of diseases;

[0017] Step S23: Based on the processing of the historical diagnosis and treatment records, extract the patient's physical characteristics and add them to the patient's basic account information. The patient's physical characteristics refer to those that are reference for the diagnosis and treatment of the patient's condition. For example, whether the patient is allergic to drugs, whether there is a smoking history, whether the smoking history is long, the alcohol and tobacco history, whether there is a surgical history that affects the diagnosis and treatment of other diseases, etc. And for which specific diseases the patient's physical characteristics have an impact on the diagnosis and treatment plan, add labels to the patient's physical characteristics.

[0018] According to the above technical solution, step S21 further includes the following steps:

[0019] Step S211: The intelligent deep learning terminal processes the expert experience knowledge through step S21 to obtain an expert knowledge feature vector. By extracting a large number of historical diagnosis and treatment records, classify the diseases. For the same disease of the same patient twice or more times, judge the disease type. When the disease is directly caused by external factors and there is no recurrence due to the current disease, mark it as having no possibility of recurrence;

[0020] Step S212: Otherwise, divide the historical diagnosis and treatment records into diagnosis and treatment record cases, extract the feature vector of the diagnosis and treatment record cases as the case feature vector. Analyze the situation where the patient has the same disease as in the diagnosis and treatment record cases only once and when the patient has multiple times, analyze the latest diagnosis and treatment record, and extract the feature vector as the prediction feature vector. Through the expert knowledge feature vector and the case feature vector, perform an association calculation on the prediction feature vector. The calculation formula for the correlation degree of the predicted disease recurrence situation of the diagnosis and treatment record is:

[0021]

[0022]

[0023] F = α·F(P,C) + β·F(P,Q)

[0024] In the formula, F(P,C) represents the correlation degree between the prediction feature vector and the expert knowledge feature vector, P represents the previous patient prediction feature vector, C represents the expert knowledge feature vector, F(P,Q) represents the correlation degree between the prediction feature vector and the case feature vector, Q represents the case feature vector, F represents the correlation degree of the predicted disease recurrence situation, and α and β respectively represent the weights of the prediction feature vector for the expert knowledge feature vector and the case feature vector;

[0025] Step S213: Through the intelligent learning terminal, predict the recurrence situation of the disease based on the predicted disease recurrence value, and add the result of the disease prediction to the patient's basic diagnosis and treatment information in the form of a label.

[0026] According to the above technical solution, step S3 further includes the following steps:

[0027] Step S31: Based on the department information selected by the patient, match the disease types responsible for the department selected by the patient with the diseases in the patient's historical diagnosis and treatment records. When the match is inconsistent, terminate the initial prediction.

[0028] Step S32: When the match is consistent, pre-display the matching diagnosis and treatment records for the doctor, and hide the historical diagnosis and treatment records unrelated to the department. When the match is a recurrence disease, obtain all the historical diagnosis and treatment records related to the patient and the recurrence disease, sort the correlation degrees of the predicted disease recurrence situations in descending order, compare the correlation degree with the system threshold, obtain relevant diagnosis and treatment cases when it is greater than the threshold, otherwise do not take any action, and combine the patient's historical diagnosis and treatment records with the diagnosis and treatment cases with high correlation degrees to form a prototype of a diagnosis and treatment plan.

[0029] Step S33: Match the labels of the patient's physical characteristics with the disease types responsible for the department. When a corresponding type is matched, display the corresponding physical characteristics in the patient's basic information, and sort the patient's physical characteristics in descending order according to the strength of the correlation.

[0030] According to the above technical solution, step S4 further includes: Based on the currently input diagnosis and treatment records, the specific disease of the patient can be judged, and the historical diagnosis and treatment records for pre-display are sorted again, and the matching historical diagnosis and treatment records are ranked in front of all the patient's historical diagnosis and treatment records. During the diagnosis and treatment process, when the system receives a button for the doctor to refer to relevant cases, match the expert experience knowledge and the correlation degree of the historical diagnosis and treatment records according to the currently input diagnosis and treatment records, and sort them in descending order according to the strength of the correlation degree and display them to the doctor.

[0031] According to the above technical solution, step S5 further includes the following steps:

[0032] Step S51: Statistically analyze the first part viewed by the doctor on the diagnosis and treatment interface using a time decay model, and the part with the highest statistics for the first part viewed will be jumped to first when the diagnosis and treatment interface is opened.

[0033] Step S52: Statistically analyze the sorting between the doctor's diagnosis and treatment plan and the recommended relevant reference diagnosis and treatment plans, and adjust the weights of the recommended diagnosis and treatment plans according to the doctor's diagnosis and treatment habits.

[0034] Step S53: In the recommended diagnosis and treatment plans, provide a multi-dimensional display method, including display in multiple dimensions such as time, disease severity, age of the case patients, treatment effect, etc., statistically analyze which dimension the doctor prefers to use, and the system automatically recommends a display that conforms to the doctor's preference dimension.

[0035] According to the above technical solution, the system includes a diagnosis and treatment information acquisition module, a historical diagnosis and treatment analysis module, an intelligent recommendation module, and a personalized recommendation module:

[0036] The diagnosis and treatment information acquisition module is used to obtain the registration information input by the patient and the current diagnosis and treatment record input, obtain the historical diagnosis and treatment records in the case library, and obtain the current diagnosis and treatment information input by the doctor and the expert experience knowledge;

[0037] The historical diagnosis and treatment analysis module is used to perform case analysis on the historical diagnosis and treatment records, provide a reference for the diagnosis and treatment plan according to the analysis results, and add physical feature tags to the user;

[0038] The intelligent recommendation module is used to predict the current disease of the patient based on the department information selected by the patient, perform a preliminary correlation ranking on the historical diagnosis and treatment records of the patient according to the prediction result, perform a secondary ranking on the historical diagnosis and treatment records of the patient based on the current input diagnosis and treatment record, and receive whether the historical diagnosis and treatment record is needed as a reference;

[0039] The personalized recommendation module is used to update the diagnosis and treatment information of the patient in real time and obtain the diagnosis and treatment actions of the doctor, analyze the diagnosis and treatment habits of the doctor, and recommend personalized intelligent analysis results to the doctor.

[0040] According to the above technical solution, the historical diagnosis and treatment analysis module includes an intelligent semantic segmentation module, an intelligent deep learning module, and a physical feature extraction module:

[0041] The intelligent semantic segmentation module is used to intelligently segment key information from the historical diagnosis and treatment record text and the expert experience knowledge, perform intelligent text word segmentation on the historical diagnosis and treatment record text, convert the key information of the disease description and the diagnosis and treatment plan into a word vector sequence through the pre-trained vector model GloVe, and use a bidirectional LSTM neural network and an attention mechanism to integrate the word vectors, extract the diagnosis and treatment feature values of the diagnosis and treatment record, and construct a diagnosis and treatment feature vector as H = [h1, h2,..., h n , where h i represents the i-th word vector in the historical diagnosis and treatment record description sentence, and n represents the number of word vectors in the historical diagnosis and treatment record description sentence;

[0042] The intelligent deep learning module is used for the intelligent deep learning terminal to process the expert experience knowledge to obtain an expert knowledge feature vector, classify diseases by extracting a large number of historical diagnosis and treatment records, judge the disease type for the same patient with the same disease twice or more, predict the recurrence value of the predicted disease for the recurrence of the disease condition, and add the result of the disease condition prediction to the patient's basic diagnosis and treatment information in the form of a label;

[0043] The body feature extraction module is used to process the historical diagnosis and treatment records, extract the patient's body features and add them to the basic account information of the patient. The patient's body features refer to those that are reference for the diagnosis and treatment of the patient's condition. For example, whether the patient is allergic to drugs, whether there is a smoking history, whether the smoking history is long, the history of alcohol and tobacco, whether there is a surgical history that affects the diagnosis and treatment of other diseases, etc. And add tags to the patient's body features to indicate which specific disease diagnosis and treatment plans they affect.

[0044] According to the above technical solution, the personalized recommendation module includes a diagnosis and treatment action analysis module, a diagnosis and treatment habit analysis module, and a multi-dimensional display module:

[0045] The diagnosis and treatment action analysis module is used to statistically analyze the first viewed part of the diagnosis and treatment interface by the doctor using a time decay model. The part with the highest statistics of the first viewed part will be jumped to first when the diagnosis and treatment interface is opened;

[0046] The diagnosis and treatment habit analysis module is used to statistically analyze the sorting of the doctor's diagnosis and treatment plan and the recommended relevant reference diagnosis and treatment plans, and adjust the weight of the recommended diagnosis and treatment plan according to the doctor's diagnosis and treatment habits;

[0047] The multi-dimensional display module is used to provide a multi-dimensional display method in the recommended diagnosis and treatment plan, including display in multiple dimensions such as time, disease severity, age of the case patient, treatment effect, etc. Statistically analyze which dimension the doctor prefers to use, and the system automatically recommends the dimension display that meets the doctor's preference.

[0048] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: Through the historical diagnosis and treatment analysis module, the present invention is used to perform case analysis on the historical diagnosis and treatment records, provide reference for diagnosis and treatment plans according to the analysis results, add body feature tags to users, classify the diseases in the historical diagnosis and treatment records, predict the diseases that may recur, and provide the definite development direction of the diseases and treatment plans in advance. Through the intelligent recommendation module, the diseases of the patient are predicted, providing effective historical diagnosis and treatment records of the patient for the doctor, reducing the doctor's screening and viewing of the patient's historical diagnosis and treatment, and providing better diagnosis and treatment decisions for the doctor. The system can identify the doctor's diagnosis and treatment actions in real time. Through the personalized recommendation module, analyze the doctor's diagnosis and treatment habits, and provide personalized intelligent analysis results for the doctor, thereby effectively utilizing the diagnosis and treatment resources and improving the doctor's work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention, but do not constitute a limitation to the present invention. In the drawings:

[0050] Figure 1 Flowchart schematic diagram of a method for an artificial intelligence-based intelligent diagnosis and treatment system to assist doctors in viewing diagnosis and treatment records provided in Embodiment 1 of the present invention

[0051] Figure 2 Schematic diagram of the module composition of an artificial intelligence-based intelligent diagnosis and treatment system provided in Embodiment 2 of the present invention Detailed implementation manners

[0052] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention

[0053] Embodiment 1: This embodiment can be applied to the scenario of assisting doctors in viewing diagnosis and treatment records in intelligent diagnosis and treatment. This method can be executed by an artificial intelligence-based intelligent diagnosis and treatment system provided in this embodiment Figure 1 Flowchart schematic diagram of a method for an artificial intelligence-based intelligent diagnosis and treatment system to assist doctors in viewing diagnosis and treatment records provided in Embodiment 1 of the present invention. The method specifically includes the following steps

[0054] Step S1: Obtain the registration information input by the patient and the currently input diagnosis and treatment records, obtain the historical diagnosis and treatment records in the case database, and obtain the currently input diagnosis and treatment information and expert experience knowledge of the doctor

[0055] Step S2: Perform case analysis on the historical diagnosis and treatment records, provide a reference for the diagnosis and treatment plan according to the analysis results, and add body feature labels to the user

[0056] Step S3: Based on the department information selected by the patient, predict the current disease of the patient, and perform a preliminary correlation ranking on the historical diagnosis and treatment records of the patient according to the prediction results

[0057] Step S4: Perform a secondary ranking on the historical diagnosis and treatment records of the patient based on the currently input diagnosis and treatment records, and receive whether the historical diagnosis and treatment records are needed as a reference

[0058] Step S5: Real-time update the diagnosis and treatment information of the patient and obtain the diagnosis and treatment actions of the doctor, analyze the diagnosis and treatment habits of the doctor, and recommend personalized intelligent analysis results to the doctor

[0059] In the embodiment of the present invention, step S1 further includes the following steps

[0060] Step S11: The registration information includes: ID card information and selected department information. The historical diagnosis and treatment records include: the patient's historical diagnosis and treatment records and other patients' diagnosis and treatment records, along with corresponding inspection reports. The current diagnosis and treatment information refers to the description of the patient's current medical condition during the visit.

[0061] Step S12: Clean the historical diagnosis and treatment records, identify and delete duplicate diagnosis and treatment records, use regular expressions to remove irrelevant characters and spaces in the text, and standardize the diagnosis and treatment record format using DICOM.

[0062] Step S13: When obtaining the historical diagnosis and treatment records, protect the privacy of the patient. Use data desensitization and data encryption technologies to process the historical diagnosis and treatment records, and display the diagnosis and treatment information of other patients anonymously with the ID number partially hidden.

[0063] In the embodiment of the present invention, step S2 further includes the following steps:

[0064] Step S21: Based on the historical diagnosis and treatment records, adopt the model technology based on Transformer to intelligently segment key information from the historical diagnosis and treatment record text, perform intelligent text word segmentation on the historical diagnosis and treatment record text, convert the key information of the disease description and treatment plan into a word vector sequence through the pre-trained vector model GloVe, and use a bidirectional LSTM neural network and an attention mechanism to integrate the word vectors, extract the diagnosis and treatment feature values of the diagnosis and treatment records, and construct a diagnosis and treatment feature vector H = [h1, h2,..., h n , where h i represents the i-th word vector in the historical diagnosis and treatment record description sentence, and n represents the number of word vectors in the historical diagnosis and treatment record description sentence.

[0065] Step S22: The intelligent deep learning terminal analyzes the historical diagnosis and treatment records, classifies the diseases, stores the diagnosis and treatment records of the same type of diseases of the same patient in one document. The intelligent deep learning terminal tracks the development of the disease and the treatment plan, and learns the optimal treatment plan according to different treatment methods for the same type of diseases.

[0066] Step S23: Based on the processing of the historical diagnosis and treatment records, extract the patient's physical characteristics and add them to the patient's account basic information. The patient's physical characteristics refer to those that are reference for the diagnosis and treatment of the patient's condition. For example, whether the patient is allergic to drugs, whether there is a smoking history, whether the smoking history is long, the history of alcohol and tobacco use, whether there is a surgical history that affects the diagnosis and treatment of other diseases, etc., and add labels to which specific disease diagnosis and treatment plans the patient's physical characteristics affect.

[0067] In the embodiment of the present invention, step S21 further includes the following steps:

[0068] Step S211: The intelligent deep learning terminal processes the expert experience knowledge through Step S21 to obtain an expert knowledge feature vector. By extracting a large number of historical diagnosis and treatment records, disease classification is carried out. For the same patient with the same disease twice or more, the disease type is judged. When the disease is directly caused by external factors and there is no possibility of recurrence due to the current disease, it is marked as having no recurrence possibility;

[0069] Step S212: Conversely, divide the historical diagnosis and treatment records into diagnosis and treatment record cases, extract the feature vector of the diagnosis and treatment record cases as the case feature vector. Analyze that the patient has the same disease as the diagnosis and treatment record case only once, and when the patient has multiple times, analyze the latest diagnosis and treatment record, and extract the feature vector as the prediction feature vector. Through the expert knowledge feature vector and the case feature vector, perform an association calculation on the prediction feature vector. The calculation formula for the association degree of the predicted disease recurrence situation of the diagnosis and treatment record is:

[0070]

[0071] F = α·F(P, C)+β·F(P, Q)

[0072] In the formula, F(P, C) represents the association degree between the prediction feature vector and the expert knowledge feature vector, P represents the previous patient prediction feature vector, C represents the expert knowledge feature vector, F(P, Q) represents the association degree between the prediction feature vector and the case feature vector, Q represents the case feature vector, F represents the association degree of the predicted disease recurrence situation, and α and β respectively represent the weights of the prediction feature vector for the expert knowledge feature vector and the case feature vector;

[0073] Step S213: The intelligent learning terminal predicts the disease recurrence situation based on the predicted disease recurrence value, and adds the result of the disease prediction to the patient's basic diagnosis and treatment information in the form of a label.

[0074] In the embodiment of the present invention, Step S3 further includes the following steps:

[0075] Step S31: Based on the department information selected by the patient, match the disease types responsible for the department selected by the patient with the diseases in the patient's historical diagnosis and treatment records. When the match is inconsistent, terminate the initial prediction;

[0076] Step S32: When the match is successful, the matched medical records are pre-displayed for the doctor, and the historical medical records unrelated to the department are hidden. When the match is a recurrence of a disease, all historical medical records related to the recurrence of the disease for the patient are obtained, the correlation degrees of the predicted disease recurrence situations are sorted in descending order, the correlation degrees are compared with the system threshold. When it is greater than the threshold, relevant medical treatment cases are obtained, otherwise no action is taken. A prototype of a medical treatment plan is formulated by combining the patient's historical medical records with the medical treatment cases with high correlation degrees;

[0077] Step S33: The labels of the patient's physical characteristics are matched with the types of diseases responsible for the department. When the corresponding type is matched, the corresponding physical characteristics are displayed in the patient's basic information, and the patient's physical characteristics are sorted in descending order according to the strength of the correlation.

[0078] In an embodiment of the present invention, step S4 further includes: Based on the currently input medical record, the specific disease of the patient can be judged, and the pre-displayed historical medical records are sorted again, and the matched historical medical records are arranged in front of all the patient's historical medical records. During the medical treatment process, when the system receives a button for the doctor to refer to relevant cases, the expert experience knowledge and the correlation degree of the historical medical records are matched according to the currently input medical record, and are sorted in descending order according to the strength of the correlation degree and displayed to the doctor.

[0079] In an embodiment of the present invention, step S5 further includes the following steps:

[0080] Step S51: According to the first viewed part of the doctor on the medical treatment interface, a time decay model is used for statistics. For the part with the highest statistics in the first viewed part, when the medical treatment interface is opened, it first jumps to this part;

[0081] Step S52: The sorting in the doctor's medical treatment plan and the recommended relevant reference medical treatment plans for the doctor are counted, and the weights of the recommended medical treatment plans are adjusted according to the doctor's medical treatment habits;

[0082] Step S53: In the recommended medical treatment plans, a multi-dimensional display method is provided, and it is displayed in multiple dimensions such as time, disease severity, age of the case patient, medical treatment effect, etc. The dimension that the doctor prefers to use is counted, and the system automatically recommends a display that conforms to the doctor's preference dimension.

[0083] Embodiment 2: Embodiment 2 of the present invention provides the first viewed part, Figure 2 which is a schematic diagram of the module composition of an artificial intelligence-based intelligent medical treatment system provided by Embodiment 2 of the present invention. As Figure 2 shown, the system includes a medical treatment information collection module, a historical medical treatment analysis module, an intelligent recommendation module, and a personalized recommendation module:

[0084] The medical information collection module is used to obtain the registration information input by the patient and the current medical record input, obtain the historical medical records in the case database, and obtain the current medical information input by the doctor and the expert experience knowledge;

[0085] The historical medical record analysis module is used to perform case analysis on the historical medical records, provide a reference for the medical treatment plan according to the analysis results, and add physical feature labels to the user;

[0086] The intelligent recommendation module is used to predict the current disease of the patient based on the department information selected by the patient, perform a preliminary correlation ranking on the historical medical records of the patient according to the prediction results, perform a secondary ranking on the historical medical records of the patient based on the current input medical record, and receive whether the historical medical records are needed as a reference;

[0087] The personalized recommendation module is used to update the medical information of the patient in real time, obtain the medical treatment actions of the doctor, analyze the medical treatment habits of the doctor, and recommend personalized intelligent analysis results to the doctor.

[0088] In some embodiments of the present invention, the historical medical record analysis module includes an intelligent semantic segmentation module, an intelligent deep learning module, and a physical feature extraction module:

[0089] The intelligent semantic segmentation module is used to intelligently segment the key information of the historical medical record text and the expert experience knowledge, perform intelligent text word segmentation on the historical medical record text, convert the key information of the disease description and the medical treatment plan into a word vector sequence through the pre-trained vector model GloVe, use a bidirectional LSTM neural network and an attention mechanism to integrate the word vectors, extract the medical treatment feature values of the medical record, and construct a medical treatment feature vector as H = [h1, h2,..., h n , where h i represents the i-th word vector in the historical medical record description sentence, and n represents the number of word vectors in the historical medical record description sentence;

[0090] The intelligent deep learning module is used to process the expert experience knowledge by the intelligent deep learning terminal to obtain the expert knowledge feature vector, classify the diseases by extracting a large number of historical medical records, judge the disease type for the same patient with the same disease twice or more times, predict the recurrence value of the predicted disease for the recurrence of the disease condition, and add the result of the disease condition prediction to the basic medical information of the patient in the form of a label;

[0091] The body feature extraction module is used to process the historical diagnosis and treatment records, extract the patient's body features and add them to the patient's basic account information. The patient's body features refer to those that are reference for the diagnosis and treatment of the patient's condition. For example, whether the patient is allergic to drugs, whether there is a smoking history, whether the smoking history is long, the history of alcohol and tobacco, whether there is a surgical history that affects the diagnosis and treatment of other diseases, etc. And for which specific diseases the patient's body features have an impact on the diagnosis and treatment plan, add labels to the patient's body features.

[0092] In some embodiments of the present invention, the personalized recommendation module includes a diagnosis and treatment action analysis module, a diagnosis and treatment habit analysis module, and a multi-dimensional display module:

[0093] The diagnosis and treatment action analysis module is used to statistically analyze the first viewed part of the diagnosis and treatment interface by the doctor using a time decay model. The part with the highest statistics of the first viewed part will be jumped to first when the diagnosis and treatment interface is opened;

[0094] The diagnosis and treatment habit analysis module is used to statistically analyze the sorting of the doctor's diagnosis and treatment plan and the recommended relevant reference diagnosis and treatment plans, and adjust the weight of the recommended diagnosis and treatment plan according to the doctor's diagnosis and treatment habits;

[0095] The multi-dimensional display module is used to provide a multi-dimensional display method in the recommended diagnosis and treatment plan, and display in multiple dimensions such as time, disease severity, age of the case patient, treatment effect, etc. Statistically analyze which dimension the doctor prefers to use, and the system automatically recommends to display in the dimension that the doctor prefers.

[0096] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0097] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An artificial intelligence-based intelligent diagnosis and treatment method, characterized in that: The running steps of the method include: Step S1: Obtain the registration information input by the patient and the current input diagnosis and treatment records, obtain the historical diagnosis and treatment records in the case database, and obtain the current diagnosis and treatment information input by the doctor and the expert experience knowledge; Step S2: Conduct a case analysis on the historical diagnosis and treatment records, provide a reference for the diagnosis and treatment plan based on the analysis results, and add physical feature tags to the user; Step S3: Based on the department information selected by the patient, predict the current condition of the patient, and perform a preliminary correlation ranking on the historical diagnosis and treatment records of the patient according to the prediction results; Step S4: Perform a secondary ranking on the historical diagnosis and treatment records of the patient based on the current input diagnosis and treatment records, and receive whether the historical diagnosis and treatment records are needed as a reference; Step S5: Real-time update the diagnosis and treatment information of the patient and obtain the doctor's diagnosis and treatment actions, analyze the doctor's diagnosis and treatment habits, and recommend personalized intelligent analysis results to the doctor.

2. The intelligent diagnosis and treatment method based on artificial intelligence according to claim 1, wherein: The step S1 further includes the following steps: Step S11: The registration information includes: ID card information, selected department information, and the historical diagnosis and treatment records include: the patient's historical diagnosis and treatment records, other patients' diagnosis and treatment records, with corresponding inspection reports attached, and the current diagnosis and treatment information refers to the description of the patient's current medical condition; Step S12: Clean the historical diagnosis and treatment records, identify and delete duplicate diagnosis and treatment records, use regular expressions to remove irrelevant characters and spaces in the text, and standardize the diagnosis and treatment record format using DICOM; Step S13: When obtaining the historical diagnosis and treatment records, protect the privacy of the patient, process the historical diagnosis and treatment records using data desensitization and data encryption technologies, and display the diagnosis and treatment information of other patients anonymously with the ID card number partially hidden.

3. The intelligent diagnosis and treatment method based on artificial intelligence according to claim 2, characterized in that: The step S2 further includes the following steps: Step S21: Based on the historical diagnosis and treatment records, using the Transformer-based model technology, intelligently segment the key information of the historical diagnosis and treatment record text, perform intelligent text word segmentation on the historical diagnosis and treatment record text, convert the key information of the disease description and treatment plan into a word vector sequence through the pre-trained vector model GloVe, and use the bidirectional LSTM neural network and attention mechanism to integrate the word vectors, extract the diagnosis and treatment feature values of the diagnosis and treatment record, and construct a diagnosis and treatment feature vector as H = [h1, h2,..., h n , where h i represents the i-th word vector in the historical diagnosis and treatment record description sentence, and n represents the number of word vectors in the historical diagnosis and treatment record description sentence; Step S22: The intelligent deep learning terminal analyzes the historical diagnosis and treatment records to classify the diseases. The diagnosis and treatment records of the same type of diseases of the same patient are stored in one document. The intelligent deep learning terminal tracks the development of the disease and the diagnosis and treatment plan, and learns the optimal diagnosis and treatment plan according to different diagnosis and treatment methods for the same type of diseases; Step S23: Based on the processing of the historical diagnosis and treatment records, extract the patient's physical features and add them to the basic information of the patient's account. The patient's physical features refer to those that are reference for the diagnosis and treatment of the patient's condition, and which specific disease diagnosis and treatment plans are affected by the patient's physical features, and add tags to the patient's physical features.

4. The intelligent diagnosis and treatment method based on artificial intelligence according to claim 3, characterized in that: The step S21 further includes the following steps: Step S211: The intelligent deep learning terminal processes the expert experience knowledge through step S21 to obtain the expert knowledge feature vector. By extracting a large number of historical diagnosis and treatment records, classify the diseases. For the same patient with the same disease two or more times, judge the disease type. When the disease is directly caused by external factors and there is no possibility of recurrence due to the current disease, mark it as having no possibility of recurrence; Step S212: Conversely, divide the historical diagnosis and treatment records into diagnosis and treatment record cases, extract the feature vectors of the diagnosis and treatment record cases as case feature vectors, analyze that the patient has the same disease condition as that in the diagnosis and treatment record case only once, and when the patient has multiple times, analyze the latest diagnosis and treatment record, extract the feature vectors as prediction feature vectors, and perform correlation calculation on the prediction feature vectors through the expert knowledge feature vectors and the case feature vectors. The calculation formula for the correlation degree of the predicted disease recurrence of the diagnosis and treatment record is: F = α·F(P, C)+β·F(P, Q) In the formula, F(P, C) represents the correlation degree between the prediction feature vector and the expert knowledge feature vector, P represents the previous patient prediction feature vector, C represents the expert knowledge feature vector, F(P, Q) represents the correlation degree between the prediction feature vector and the case feature vector, Q represents the case feature vector, F represents the correlation degree of the predicted disease recurrence, and α and β respectively represent the weights of the prediction feature vector for the expert knowledge feature vector and the case feature vector; Step S213: Use the intelligent learning terminal to predict the disease recurrence situation based on the predicted disease recurrence value, and add the result of the disease prediction to the patient's basic diagnosis and treatment information in the form of a label.

5. The intelligent diagnosis and treatment method based on artificial intelligence according to claim 4, characterized in that: Step S3 further includes the following steps: Step S31: Based on the department information selected by the patient, match the disease types responsible for the department selected by the patient with the diseases in the patient's historical diagnosis and treatment records. When the match is inconsistent, terminate the initial prediction; Step S32: When the match is consistent, pre-display the matched diagnosis and treatment records to the doctor, hide the historical diagnosis and treatment records irrelevant to the department. When the match is a recurrent disease, obtain all the historical diagnosis and treatment records related to the patient's recurrent disease, sort the correlation degrees of the predicted disease recurrence in descending order, compare the correlation degree with the system threshold. When it is greater than the threshold, obtain the relevant diagnosis and treatment cases, otherwise do not take any action, and combine the patient's historical diagnosis and treatment records with the diagnosis and treatment cases with high correlation degrees to form a prototype of a diagnosis and treatment plan; Step S33: Match the label of the patient's physical characteristics with the disease types responsible for the department. When a corresponding type is matched, display the corresponding physical characteristics in the patient's basic information, and sort the patient's physical characteristics in descending order according to the strength of the correlation.

6. The intelligent diagnosis and treatment method based on artificial intelligence according to claim 5, characterized in that: Step S4 further includes: Based on the currently input diagnosis and treatment record, the specific disease of the patient can be judged, and the pre-displayed historical diagnosis and treatment records are sorted again, and the matched historical diagnosis and treatment records are arranged in front of all the patient's historical diagnosis and treatment records. During the diagnosis and treatment process, when the system receives the button for the doctor to refer to relevant cases, match the expert experience knowledge and the correlation degree of the historical diagnosis and treatment records according to the currently input diagnosis and treatment record, and display them to the doctor in descending order according to the strength of the correlation degree.

7. The intelligent diagnosis and treatment method based on artificial intelligence according to claim 6, characterized in that: Step S5 further includes the following steps: Step S51: Statistically analyze the first viewed part of the diagnosis and treatment interface by the doctor using a time decay model. When opening the diagnosis and treatment interface, the part with the highest statistics of the first viewed part will be jumped to first; Step S52: Statistically analyze the sorting between the doctor's diagnosis and treatment plan and the recommended relevant diagnosis and treatment plans, and adjust the weights of the recommended diagnosis and treatment plans according to the doctor's diagnosis and treatment habits; Step S53: Provide a multi-dimensional display method in the recommended diagnosis and treatment plans, including display in multiple dimensions such as time, disease severity, age of the case patients, treatment effects, etc. Statistically analyze which dimension the doctor prefers to use, and the system automatically recommends to display in the dimension preferred by the doctor.

8. An artificial intelligence-based intelligent diagnosis and treatment system, characterized in that: The system includes a diagnosis and treatment information collection module, a historical diagnosis and treatment analysis module, an intelligent recommendation module, and a personalized recommendation module: The diagnosis and treatment information collection module is used to obtain the registration information input by the patient and the current diagnosis and treatment records, obtain the historical diagnosis and treatment records in the case library, and obtain the current diagnosis and treatment information input by the doctor and the expert experience knowledge; The historical diagnosis and treatment analysis module is used to perform case analysis on the historical diagnosis and treatment records, provide a reference for the diagnosis and treatment plan according to the analysis results, and add physical feature labels to the user; The intelligent recommendation module is used to predict the patient's current disease based on the department information selected by the patient, perform a preliminary correlation ranking on the patient's historical diagnosis and treatment records according to the prediction results, perform a secondary ranking on the patient's historical diagnosis and treatment records based on the current input diagnosis and treatment records, and receive whether the historical diagnosis and treatment records are needed as a reference; The personalized recommendation module is used to update the patient's diagnosis and treatment information in real time and obtain the doctor's diagnosis and treatment actions, analyze the doctor's diagnosis and treatment habits, and provide personalized intelligent analysis results for the doctor.

9. An intelligent diagnosis and treatment system based on artificial intelligence according to claim 8, characterized in that: The historical diagnosis and treatment analysis module includes an intelligent semantic segmentation module, an intelligent deep learning module, and a physical feature extraction module: The intelligent semantic segmentation module is used to intelligently segment key information from the historical diagnosis and treatment record text and the expert experience knowledge, perform intelligent text word segmentation on the historical diagnosis and treatment record text, convert the key information of the disease description and the diagnosis and treatment plan into a word vector sequence through the pre-trained vector model GloVe, use a bidirectional LSTM neural network and an attention mechanism to integrate the word vectors, extract the diagnosis and treatment feature values of the diagnosis and treatment record, and construct a diagnosis and treatment feature vector as H = [h1, h2,..., h n , where h i represents the i-th word vector in the historical diagnosis and treatment record description sentence, and n represents the number of word vectors in the historical diagnosis and treatment record description sentence; The intelligent deep learning module is used for the intelligent deep learning terminal to process the expert experience knowledge to obtain expert knowledge feature vectors. By extracting a large number of historical diagnosis and treatment records, perform disease classification. For the same patient with the same disease twice or more, judge the disease type, predict the recurrence value of the predicted disease for the recurrence of the disease condition, and add the result of the disease condition prediction to the patient's basic diagnosis and treatment information in the form of a label; The physical feature extraction module is used to process the historical diagnosis and treatment records, extract the patient's physical features and add them to the patient's account basic information. The patient's physical features refer to those that are reference for the patient's disease diagnosis and treatment, and which specific disease diagnosis and treatment plans are affected by the patient's physical features, and add labels to the patient's physical features.

10. A smart diagnosis and treatment system based on artificial intelligence according to claim 9, characterized in that: The personalized recommendation module includes a diagnosis and treatment action analysis module, a diagnosis and treatment habit analysis module, and a multi-dimensional display module: The diagnosis and treatment action analysis module is used to statistically analyze the first viewed part of the diagnosis and treatment interface by the doctor using a time decay model. When opening the diagnosis and treatment interface, the part with the highest statistics of the first viewed part will be jumped to first; The diagnosis and treatment habit analysis module is used to count the sorting in the diagnosis and treatment plan received by the doctor and the recommended relevant reference diagnosis and treatment plans, and adjust the weight of the recommended diagnosis and treatment plan according to the doctor's diagnosis and treatment habits; The multi-dimensional display module is used to provide a multi-dimensional display method in the recommended diagnosis and treatment plan, which can be displayed in multiple dimensions such as time, disease severity, age of the case patients, treatment effect, etc., count which dimension the doctor prefers to use, and the system automatically recommends the dimension display that conforms to the doctor's preference.

Images