Search

CN-122000074-A - Functional gastrointestinal disease needling scheme recommendation method and system based on big data

CN122000074ACN 122000074 ACN122000074 ACN 122000074ACN-122000074-A

Abstract

The application provides a functional gastrointestinal disease needling scheme recommendation method and system based on big data, the method comprises the steps of obtaining patient symptom information and patient physique information, inputting the patient symptom information into a gastrointestinal disease judgment model to obtain a gastrointestinal disease judgment result, inputting the patient physique information and the gastrointestinal disease judgment result into a needling scheme prediction model to obtain the use effect of the needling scheme prediction result, screening the needling scheme prediction result according to the use effect ranking of the needling scheme prediction result, inputting the screened needling scheme into a needling scheme verification model to obtain physique verification information and a gastrointestinal disease verification result, and determining the needling scheme recommendation result according to the comparison result of the physique verification information and the patient physique information and the comparison result of the gastrointestinal disease verification result and the gastrointestinal disease judgment result; the application can determine the recommended result of the needling scheme, and is beneficial to avoiding determining an unsuitable needling scheme for a patient due to insufficient personal experience of doctors.

Inventors

  • XU MINGMIN
  • HAO YAN
  • LIN SHUJUN
  • TANG XIAORONG

Assignees

  • 广州中医药大学(广州中医药研究院)

Dates

Publication Date
20260508
Application Date
20260127

Claims (10)

  1. 1. A functional gastrointestinal disorder needling protocol recommendation method based on big data, the method comprising: Acquiring patient symptom information and patient constitution information; Inputting the symptom information of the patient into a gastrointestinal disease judging model to obtain a gastrointestinal disease judging result; Inputting the physique information of the patient and the gastrointestinal disease judgment result into a needling scheme prediction model to obtain a needling scheme prediction result and a use effect of the needling scheme prediction result, wherein the needling scheme prediction model is obtained by training a pre-established prediction model through first sample data; screening the needling scheme prediction results according to the ranking of the using effects of the needling scheme prediction results to obtain a screening needling scheme; Inputting the screening needling scheme into a needling scheme verification model to obtain constitution verification information and gastrointestinal disease verification results, wherein the needling scheme verification model is obtained by training a pre-established verification model through second sample data; And determining a recommended needling scheme result according to the comparison result of the constitution verification information and the constitution information of the patient and the comparison result of the gastrointestinal disease verification result and the gastrointestinal disease judgment result.
  2. 2. The method of claim 1, wherein training a pre-established predictive model from first sample data to obtain the predictive model for the needling protocol comprises: Inputting the first sample data into the pre-established prediction model to obtain a predicted needling scheme and a use effect of the predicted needling scheme, wherein the first sample data comprises a patient history diagnosis result, patient history physique information, a needling scheme used by a patient, a treatment effect of the patient using the needling scheme, a corresponding relation between a combination of the patient history diagnosis result and the patient history physique information and the needling scheme used by the patient, and a corresponding relation between the needling scheme used by the patient and the treatment effect of the needling scheme used by the patient; And aiming at minimizing the gap between the predicted needling scheme and the needling scheme used by the patient and the gap between the using effect of the predicted needling scheme and the treatment effect of the needling scheme used by the patient, adjusting parameters in the pre-established prediction model to obtain the needling scheme prediction model.
  3. 3. The method of claim 2, wherein the gap between the predicted lancing regimen and the lancing regimen used by the patient and the gap between the effectiveness of the predicted lancing regimen and the effectiveness of the treatment of the patient with the lancing regimen are characterized by a loss function expressed as follows: , Wherein, the The loss function is represented by a function of the loss, Representing the cross-entropy loss weight, Representing the cross-entropy loss, Representing the mean square error loss weight, Indicating the mean square error loss, Indicating the total number of categories of needling protocol, Representing the patient's use The category labels corresponding to the individual needling schemes, Represent the first The predicted probability of the individual needling protocol, Representing the total number of samples, Is shown in the first The therapeutic effect of the patient in the individual samples using the needle punching protocol, Representing warp The use effect of the predicted needling scheme obtained by predicting each sample is that Therapeutic effects of patient in individual samples using a needle stick regimen The use effect of the predicted needling scheme obtained by predicting each sample is characterized by a coded form or a numerical form, 。
  4. 4. The method according to claim 2, wherein if the pre-established prediction model includes a first input layer, a first multi-layer perceptron layer, a first attention mechanism layer, a first feature fusion layer, a first full connection layer, and a first output layer, inputting the first sample data into the pre-established prediction model to obtain a predicted needling scheme and an effect of the predicted needling scheme comprises: The first input layer is used for carrying out normalization operation and encoding operation on the historical diagnosis result of the patient, the historical constitution information of the patient, the needling scheme used by the patient and the treatment effect of the needling scheme used by the patient; Inputting the first coding vector into the first multi-layer perceptron layer to obtain first nonlinear characteristic information; the first multi-layer perceptron layer is used for carrying out nonlinear feature extraction and feature dimension transformation on the first coding vector; The first coding vector is input into the first attention mechanism layer to obtain attention weight information, and the first attention mechanism layer is used for calculating the attention weight among the features in the first coding vector; inputting the first nonlinear characteristic information and the attention weight information into the first characteristic fusion layer to obtain first fusion characteristic information, wherein the first characteristic fusion layer is used for carrying out characteristic splicing and characteristic dimension transformation on the first nonlinear characteristic information and the attention weight information; The first fusion characteristic information is input into the first full-connection layer to obtain the predicted needling scheme and the use effect of the predicted needling scheme, and the first full-connection layer is used for mapping the first fusion characteristic information to a first preset space dimension to obtain a first preset space dimension characteristic; Inputting the predicted needling scheme and the using effect of the predicted needling scheme into the first output layer to obtain a gap between the predicted needling scheme and the needling scheme used by the patient and a gap between the using effect of the predicted needling scheme and the treating effect of the needling scheme used by the patient, wherein the first output layer is used for calculating the gap between the predicted needling scheme and the needling scheme used by the patient and the gap between the using effect of the predicted needling scheme and the treating effect of the needling scheme used by the patient.
  5. 5. The method of claim 1, wherein training a pre-established verification model with second sample data to obtain the needling protocol verification model comprises: Inputting the second sample data into the pre-established verification model to obtain constitution prediction information and gastrointestinal disease prediction results, wherein the second sample data comprises a patient historical diagnosis result, patient historical constitution information, a needling scheme used by a patient, and a corresponding relation between the needling scheme used by the patient and a combination of the patient historical diagnosis result and the patient historical constitution information; And aiming at minimizing the difference between the constitution prediction information and the patient history constitution information and the difference between the gastrointestinal disease prediction result and the patient history diagnosis result, adjusting parameters in the pre-established verification model to obtain the needling scheme verification model.
  6. 6. The method according to claim 5, wherein if the pre-established verification model includes a second input layer, a second multi-layer perceptron layer, a second feature fusion layer, a second full-connection layer, and a second output layer, the step of inputting the second sample data into the pre-established verification model to obtain the constitution prediction information and the gastrointestinal disease prediction result comprises: inputting the second sample data into the second input layer to obtain a second coding vector; the second input layer is used for carrying out normalization operation and encoding operation on the second sample data; The second coding vector is input into the second multi-layer perceptron layer to obtain second nonlinear characteristic information, and the second multi-layer perceptron layer is used for carrying out nonlinear characteristic extraction and characteristic dimension transformation on the second coding vector; Inputting the second nonlinear characteristic information into the second characteristic fusion layer to obtain second fusion characteristic information, wherein the second characteristic fusion layer is used for carrying out characteristic splicing on the second nonlinear characteristic information; The second fusion characteristic information is input into the second full-connection layer to obtain constitution prediction information and a gastrointestinal disease prediction result, and the second full-connection layer is used for mapping the second fusion characteristic information to a second preset space dimension to obtain a second preset space dimension characteristic; and inputting the constitution prediction information and the gastrointestinal disease prediction result into the second output layer to obtain a difference between the constitution prediction information and the patient history constitution information and a difference between the gastrointestinal disease prediction result and the patient history diagnosis result, wherein the second output layer is used for calculating the difference between the constitution prediction information and the patient history constitution information and the difference between the gastrointestinal disease prediction result and the patient history diagnosis result.
  7. 7. The method of claim 1-6, wherein the step of screening the predicted results of the needling plan according to the ranking of the use effects of the predicted results of the needling plan to obtain a screened needling plan comprises: if the number of the needling schemes in the needling scheme prediction results is smaller than a preset number threshold, the needling scheme prediction results are used as the screening needling schemes; If the number of the needling schemes in the needling scheme prediction results is larger than or equal to the preset number threshold, ranking the using effects of the needling scheme prediction results according to ascending order to obtain an ascending order sequence, using the preset number of needling schemes ranked and backed in the ascending order sequence as the screening needling scheme, or ranking the using effects of the needling scheme prediction results according to descending order to obtain a descending order sequence, and using the needling schemes ranked in the descending order sequence with the preset number as the screening needling scheme.
  8. 8. The method of claim 1-6, wherein determining the recommended needling plan based on the comparison of the constitution information and the patient constitution information and the comparison of the gastrointestinal disease verification result and the gastrointestinal disease judgment result comprises: If the constitution verification information comprises constitution information of the patient and the gastrointestinal disease verification result comprises the gastrointestinal disease judgment result, taking the screening needling scheme as a needling scheme recommendation result; and if the constitution verification information does not comprise the constitution information of the patient and/or the gastrointestinal disease verification result does not comprise the gastrointestinal disease judgment result, taking a non-recommended scheme as the needling scheme recommendation result.
  9. 9. The method of claim 1-7, wherein prior to obtaining patient symptom information and patient constitution information, the method further comprises: inputting the patient symptom information and the patient constitution information through interaction equipment; Or inputting patient history diagnosis information through the interaction equipment, and extracting the patient symptom information and the patient constitution information from the patient history diagnosis information, wherein the patient history diagnosis information comprises patient diagnosis information and patient treatment information.
  10. 10. A functional gastrointestinal disorder needling protocol recommendation system based on big data, comprising: The information acquisition module is used for acquiring patient symptom information and patient constitution information; the disease judging module is used for inputting the symptom information of the patient into a gastrointestinal disease judging model to obtain a gastrointestinal disease judging result; The needling scheme prediction module is used for inputting the constitution information of the patient and the gastrointestinal disease judgment result into a needling scheme prediction model to obtain a needling scheme prediction result and a use effect of the needling scheme prediction result, wherein the needling scheme prediction model is obtained by training a pre-established prediction model through first sample data; The scheme screening module is used for screening the needling scheme prediction results according to the ranking of the using effects of the needling scheme prediction results to obtain a screening needling scheme; The needling scheme verification module is used for inputting the screening needling scheme into a needling scheme verification model to obtain constitution verification information and gastrointestinal disease verification results, wherein the needling scheme verification model is obtained by training a pre-established verification model through second sample data; and the information comparison module is used for determining a needling scheme recommendation result according to the comparison result of the constitution verification information and the constitution information of the patient and the comparison result of the gastrointestinal disease verification result and the gastrointestinal disease judgment result.

Description

Functional gastrointestinal disease needling scheme recommendation method and system based on big data Technical Field The application relates to the technical field of big data, in particular to a functional gastrointestinal disease needling scheme recommendation method and system based on big data. Background Functional gastrointestinal diseases are a group of diseases which are mainly represented by gastrointestinal dysfunction and have no clear organic lesions, and mainly comprise functional dyspepsia, irritable bowel syndrome, functional constipation, functional abdominal distention, functional diarrhea, functional abdominal pain syndrome and the like. Functional gastrointestinal disorders can be symptomatically alleviated or treated by a needle punching regimen. At present, in the process of determining a needling scheme for a functional gastrointestinal disease, due to certain subjectivity and limitation of personal experience and technical level of doctors, improper needling schemes are likely to be formulated and implemented for patients, so that the treatment effect on different patients is not ideal, and therefore, how to provide a reasonable needling scheme for the functional gastrointestinal disease patients to assist the doctors in formulating the needling scheme for the patients is important. Disclosure of Invention In view of the above-mentioned shortcomings of the prior art, the present application provides a method and system for functional gastrointestinal disease acupuncture scheme recommendation based on big data to solve the above-mentioned technical problems. According to one aspect of the embodiment of the application, a functional gastrointestinal disease needling scheme recommendation method based on big data is provided, and the method comprises the steps of obtaining patient symptom information and patient physique information, inputting the patient symptom information into a gastrointestinal disease judgment model to obtain a gastrointestinal disease judgment result, inputting the patient physique information and the gastrointestinal disease judgment result into a needling scheme prediction model to obtain a needling scheme prediction result and a using effect of the needling scheme prediction result, training the pre-established prediction model through first sample data by the needling scheme prediction model, screening the needling scheme prediction result according to the using effect ranking of the needling scheme prediction result to obtain a screening needling scheme, inputting the screening needling scheme into a needling scheme verification model to obtain physique verification information and a gastrointestinal disease verification result, training the pre-established verification model by the needling scheme verification model through second sample data, comparing the physique verification information and the patient information to obtain a using effect of the needling scheme prediction result, and determining a needling scheme recommendation result of the gastrointestinal disease. In one embodiment of the application, the process of training a pre-established prediction model through first sample data to obtain the needling scheme prediction model comprises the steps of inputting the first sample data into the pre-established prediction model to obtain a predicted needling scheme and the use effect of the predicted needling scheme, wherein the first sample data comprises the corresponding relation between a patient historical diagnosis result, patient historical physique information, a patient used needling scheme, a combination of the patient historical diagnosis result and the patient historical physique information and the needling scheme used by the patient, and the corresponding relation between the needling scheme used by the patient and the treatment effect of the needling scheme used by the patient is adjusted by taking the difference between the use effect of the predicted needling scheme and the treatment effect of the needling scheme used by the patient as a target, and the parameters in the pre-established prediction model are adjusted to obtain the needling scheme. In one embodiment of the present application, the difference between the predicted lancing regimen and the lancing regimen used by the patient, and the difference between the effectiveness of the predicted lancing regimen and the effectiveness of the treatment of the patient using the lancing regimen, is characterized by a loss function, the expression of which is as follows: , wherein, The loss function is represented by a function of the loss,Representing the cross-entropy loss weight,Representing the cross-entropy loss,Representing the mean square error loss weight,Indicating the mean square error loss,Indicating the total number of categories of needling protocol,Representing the patient's useThe category labels corresponding to the individual needling schemes,Represent the firstThe predic