Medical prescription recommendation method, device, electronic device and storage medium

Abstract

The present invention provides a medical prescription recommendation method, device, electronic device and storage medium. The method includes acquiring current condition information, dividing the current condition information to obtain first current condition information and second current condition information, scoring the medical features in the first current condition information with different preset medical prescriptions respectively, determining a first medical prescription to be recommended based on the scoring results, calculating the interaction between different medical features in the second current condition information, determining a second medical prescription to be recommended based on the calculation results, and determining a final recommended medical prescription based on the first medical prescription to be recommended and the second medical prescription to be recommended; by performing different methods of medical prescription recommendation for the first current condition information and the second current condition information respectively, the advantages and disadvantages of the two recommendation methods can be effectively balanced, and recommendation errors caused by the inaccuracy of a single recommendation method can be reduced, thereby maintaining high robustness and accuracy.

Description

Technical Field

[0001] The present application relates to the field of medical service technology, and specifically to a medical prescription recommendation method, device, electronic device and storage medium. Background Art

[0002] In the current medical field, doctors usually prescribe appropriate medical prescriptions for patients by judging the patient's condition and combining their own medical knowledge and experience. With the continuous updating of medical knowledge and the complexity of patients' conditions, this traditional way of prescribing medical prescriptions may have certain limitations. On the one hand, doctors' prescriptions are subjective, and their accuracy varies with the doctor's experience and professional knowledge. Especially for the treatment of some rare diseases or special populations, doctors may lack sufficient experience and knowledge, which affects the accuracy and effectiveness of the prescriptions they prescribe. On the other hand, the efficiency of manual prescriptions is not high. Therefore, it is difficult to stably and accurately recommend appropriate medical prescriptions to patients based solely on the doctor's subjective judgment and different levels of experience and professional knowledge.

[0003] With the development of information technology, some intelligent medical prescription recommendation systems have begun to emerge. These systems can recommend medical prescriptions to patients by analyzing their conditions. However, the recommendation results provided by these systems are not accurate and are prone to recommendation errors. Summary of the invention

[0004] In view of the above-mentioned shortcomings of the prior art, the present application provides a medical prescription recommendation method, device, electronic device and storage medium to solve the technical problem that the recommendation results provided by the above-mentioned medical prescription recommendation system are not accurate and recommendation errors are prone to occur.

[0005] The present application provides a medical prescription recommendation method, which is applied to a computer device. The medical prescription recommendation method includes: obtaining current condition information, wherein the current condition information includes medical features of different categories; dividing the current condition information to obtain first current condition information and second current condition information, wherein the categories of medical features in the first current condition information are different from the categories of medical features in the second current condition information; scoring the medical features in the first current condition information and different preset medical prescriptions respectively, determining a first medical prescription to be recommended based on the scoring results, and calculating the interaction between different medical features in the second current condition information, and determining a second medical prescription to be recommended based on the calculation results; and determining a final recommended medical prescription based on the first medical prescription to be recommended and the second medical prescription to be recommended.

[0006] In one embodiment of the present application, the medical features in the first current condition information are scored respectively with different preset medical prescriptions, and the first medical prescription to be recommended is determined based on the scoring results, including: calculating the score of each medical prescription for the first current condition information based on the medical features in the first current condition information and the prescription features in each medical prescription, the medical prescription including prescription features of different categories; sorting the scores of each medical prescription for the first current condition information, and selecting the medical prescription with the highest score as the first medical prescription to be recommended.

[0007] In one embodiment of the present application, based on the medical features in the first current condition information and the prescription features in each medical prescription, the score of the first current condition information for each medical prescription is calculated, including: based on the medical features in each first condition information and the prescription features in each medical prescription, an initial first condition information-medical prescription scoring matrix is ​​established, and the initial first condition information-medical prescription scoring matrix is ​​matrix decomposed to obtain a first condition information feature matrix and a medical prescription feature matrix, wherein each medical prescription has corresponding first condition information, the first condition information includes medical features of different categories, and the category of medical features in the first condition information is the same as the category of medical features in the first current condition information; the first condition information feature matrix and the medical prescription feature matrix are alternately optimized until the convergence condition is reached to obtain a converged first condition information-medical prescription scoring matrix; according to the medical features in the first current condition information, the score of the first current condition information for each medical prescription is determined from the converged first condition information-medical prescription scoring matrix.

[0008] In one embodiment of the present application, based on the medical features in the first current condition information, the score of the first current condition information for each medical prescription is determined from the converged first condition information-medical prescription scoring matrix, including: calculating the similarity between the first current condition information and each first condition information in the converged first condition information-medical prescription scoring matrix based on all medical features in the first current condition information and all medical features in the first condition information in the converged first condition information-medical prescription scoring matrix, and determining the first target condition information according to the similarity; reading the score of the first target condition information for each medical prescription from the converged first condition information-medical prescription scoring matrix as the score of the first current condition information for each medical prescription.

[0009] In one embodiment of the present application, based on the medical features in the first current condition information and the prescription features in each medical prescription, the score of the first current condition information for each medical prescription is calculated, including: performing vector conversion on all medical features in the first current condition information and all prescription features in each medical prescription, respectively, to obtain all medical feature vectors of the first current condition information and all prescription feature vectors of each medical prescription; based on all medical feature vectors of the first current condition information and all prescription feature vectors of a medical prescription, calculating the similarity between the first current condition information and the medical prescription as the score of the first current condition information for the medical prescription, so as to obtain the score of the first current condition information for each medical prescription.

[0010] In one embodiment of the present application, the interaction between different medical features in the second current condition information is calculated, and the second medical prescription to be recommended is determined based on the calculation result, including: obtaining multiple second condition information and medical prescriptions corresponding to each second condition information, wherein the second condition information includes medical features of different categories, and the categories of medical features in the second condition information are the same as the categories of medical features in the second current condition information; based on multiple second condition information and the medical prescriptions corresponding to each second condition information, a preset model is trained to determine the second medical prescription to be recommended through the trained model.

[0011] In one embodiment of the present application, a preset model is trained based on multiple second condition information and the medical prescriptions corresponding to each second condition information to determine the second medical prescription to be recommended through the trained model, including: taking all medical features in each second condition information as input variables of the model, and taking the medical prescription corresponding to each second condition information as the target variable of the model, training the model to learn the relationship between medical features and medical prescriptions, and obtaining the trained model; inputting all medical features in the second current condition information into the trained model, calculating the interaction between different medical features in the second current condition information according to the learned relationship between medical features and medical prescriptions through the trained model, and determining the second medical prescription to be recommended based on the calculation results.

[0012] In one embodiment of the present application, a medical prescription recommendation device is also provided, and the medical prescription recommendation device includes: a data acquisition module, which is used to obtain current condition information, and the current condition information includes different categories of medical features; an information processing module, which is used to divide the current condition information to obtain first current condition information and second current condition information, and the categories of medical features in the first current condition information are different from the categories of medical features in the second current condition information; according to the first medical prescription to be recommended and the second medical prescription to be recommended, a final recommended medical prescription is determined; a first recommendation module, which is used to score the medical features in the first current condition information with different preset medical prescriptions, and determine the first medical prescription to be recommended based on the scoring results; a second recommendation module, which is used to calculate the interaction between different medical features in the second current condition information, and determine the second medical prescription to be recommended based on the calculation result.

[0013] In one embodiment of the present application, an electronic device is also provided, comprising: one or more processors; a storage device for storing one or more programs, and when the one or more programs are executed by the one or more processors, the electronic device implements the medical prescription recommendation method as described above.

[0014] In one embodiment of the present application, a computer-readable storage medium is further provided, on which a computer program is stored. When the computer program is executed by a processor of a computer, the computer executes the medical prescription recommendation method as described above.

[0015] Beneficial effects of the present invention: The present invention provides a medical prescription recommendation method, device, electronic device and storage medium. The method divides the current condition information into the first current condition information and the second current condition information containing different categories of medical features, performs different recommendation processing for different categories of medical features, and determines the final recommended medical prescription by combining the results of the two methods. This dual recommendation method can effectively balance the advantages and disadvantages of the two recommendation methods, reduce recommendation errors caused by the inaccuracy of a single recommendation method, and thus maintain high robustness and accuracy. In addition, the method can also avoid errors in medical prescription recommendations due to subjective judgment or lack of experience and professional knowledge, and the recommendation efficiency is higher, which helps to improve the efficiency of doctors' consultations.

[0016] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings herein are incorporated into the specification and constitute a part of the specification, showing embodiments consistent with the present application, and together with the specification, are used to explain the principles of the present application. Obviously, the drawings described below are only some embodiments of the present application, and for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0018] Figure 1 is a schematic diagram of an implementation environment of a medical prescription recommendation method shown in an exemplary embodiment of the present application;

[0019] Figure 2 is a flow chart of a medical prescription recommendation method shown in an exemplary embodiment of the present application;

[0020] Figure 3 is a block diagram of a medical prescription recommendation device shown in an exemplary embodiment of the present application;

[0021] Figure 4 A schematic diagram of the structure of a computer system suitable for implementing an electronic device of an embodiment of the present application is shown. DETAILED DESCRIPTION

[0022] The following will describe the implementation methods of the present application with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be understood that the preferred embodiments are only for illustrating the present application, not for limiting the scope of protection of the present application.

[0023] It should be noted that in this application, "first", "second", etc. are only used to distinguish similar objects, and are not used to limit the order or precedence of similar objects. The variations of "including", "having", etc. described above indicate that the scope covered by the subject of the word is not exclusive except for the examples shown by the word.

[0024] It is understood that the various numbers, step numbers, and other reference numerals recorded in this application are distinguished for the convenience of description and are not intended to limit the scope of this application. The size of the reference numerals in this application does not mean the order of execution. The execution order of each process should be determined by its function and internal logic.

[0025] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application. However, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present application difficult to understand.

[0026] The embodiments of the present application respectively propose a medical prescription recommendation method, a medical prescription recommendation device, an electronic device, and a computer-readable storage medium, and these embodiments will be described in detail below.

[0027] See also Figure 1 , Figure 1 It is a schematic diagram of an implementation environment of a medical prescription recommendation method shown as an exemplary embodiment of the present application.

[0028] like Figure 1 As shown, the implementation environment may include a medical terminal 110 and a computer device 120, wherein the medical terminal 110 refers to a terminal device used in the medical field, which is used to collect, transmit, process and display medical-related data and provide medical services, including at least one of a medical monitoring terminal, a medical imaging terminal, an electronic medical record terminal, a telemedicine terminal, an intelligent medical device terminal, etc., which are not limited here. The computer device 120 may be at least one of a microcomputer, an embedded computer, a neural network computer, etc. The computer device 120 may be configured in the medical terminal 110, or the computer device 120 may be a computer device independent of the medical terminal 110, which are not limited here. The medical terminal 110 may be used to collect current condition information and provide it to the computer device 120 for medical prescription recommendation.

[0029] Exemplarily, the computer device 120 obtains the current condition information through the medical terminal 110, and the current condition information includes different categories of medical features; divides the current condition information to obtain the first current condition information and the second current condition information, and the categories of the medical features in the first current condition information are different from the categories of the medical features in the second current condition information; scores the medical features in the first current condition information and the preset different medical prescriptions, and determines the first medical prescription to be recommended based on the scoring results, and calculates the interaction between the different medical features in the second current condition information, and determines the second medical prescription to be recommended based on the calculation results; and determines the final recommended medical prescription according to the first medical prescription to be recommended and the second medical prescription to be recommended. It can be seen that the technical solution of the embodiment of the present application divides the current condition information into the first current condition information and the second current condition information containing different categories of medical features, so as to perform different recommendation processing for different categories of medical features, and determine the final recommended medical prescription by combining the results of the two methods. This dual recommendation method can effectively balance the advantages and disadvantages of the two recommendation methods, reduce the recommendation errors caused by the inaccuracy of a single recommendation method, and thus maintain high robustness and accuracy. In addition, this technical solution can effectively solve the problem that the accuracy and effectiveness of prescriptions may be affected due to the doctor's lack of sufficient experience and skills, avoid errors in medical prescription recommendations due to subjective judgment or lack of experience and professional knowledge, and have higher recommendation efficiency, which helps to improve the efficiency of doctors' consultations.

[0030] It should be noted that the medical prescription recommendation method provided in the embodiment of the present application can be specifically executed by the computer device 120, and the corresponding medical prescription recommendation device can be set in the computer device 120.

[0031] See also Figure 2 , Figure 2 is a flowchart of a medical prescription recommendation method shown in an exemplary embodiment of the present application. The method can be applied to Figure 1 The implementation environment shown is specifically executed by the computer device 120 in the implementation environment. It should be understood that the method can also be applied to other exemplary implementation environments and be specifically executed by devices in other implementation environments. This embodiment does not limit the implementation environment to which the method is applicable.

[0032] like Figure 2 As shown, in an exemplary embodiment, the medical prescription recommendation method includes at least steps S210 to S240, which are described in detail as follows:

[0033] Step S210, obtaining current condition information.

[0034] In one embodiment of the present application, after diagnosing a patient, the doctor can input the patient's current condition information into a medical terminal such as a computer, mobile phone, or tablet, so that the medical terminal transmits the current condition information to the computer device, wherein the current condition information includes different categories of medical features, for example: the current condition information includes medical features 1-N, wherein medical feature 1 is symptoms, medical feature 2 is diagnosis results, medical feature 3 is disease history, medical feature 4 is risk factors, medical feature 4 is lifestyle, medical feature 6 is family history...

[0035] Step S220, dividing the current condition information to obtain first current condition information and second current condition information.

[0036] In one embodiment of the present application, the current condition information can be classified according to the similarity between each medical feature, so as to classify medical features of similar categories into the same set, and determine two sets of condition information, namely, the first current condition information and the second current condition information. The first current condition information includes multiple medical features, and the second current condition information includes multiple medical features. The categories of the medical features in the first current condition information are different from the categories of the medical features in the second current condition information. For example, the first current condition information includes medical feature 1, medical feature 2, and medical feature 3, and the second current condition information includes medical feature 4, medical feature 5, and medical feature 6.

[0037] Illustratively, a clustering algorithm may be used to determine the similarity between the medical features, wherein the clustering algorithm may be at least one of a K-Means algorithm, a hierarchical clustering algorithm, a DBSCAN algorithm, and the like.

[0038] In one embodiment of the present application, the current condition information is divided to obtain the first current condition information and the second current condition information, including: setting the number of clusters to 2, and randomly selecting two medical features from the current condition information as the initial cluster centers; calculating the distance from each medical feature in the current condition information to the cluster center, and assigning each medical feature to the nearest cluster center according to the distance to form two initial clusters; calculating the mean of all medical features in each cluster as the new cluster center to update the position of the cluster center; repeatedly calculating the distance from each medical feature to the cluster center, and assigning each medical feature to the nearest cluster center according to the distance to form two initial clusters, and calculating the mean of all medical features in each cluster as the new cluster center to update the position of the cluster center until the cluster center no longer changes or meets the stop condition, and obtaining the final two clusters, which are respectively used as the first current condition information and the second current condition information. Among them, the stop condition may include that the number of cluster center iterations reaches a preset number or the change amount of the cluster center is less than the preset change amount, etc.

[0039] Step S230, scoring the medical features in the first current condition information respectively with different preset medical prescriptions, determining the first recommended medical prescription based on the scoring results, and calculating the interaction between different medical features in the second current condition information, and determining the second recommended medical prescription based on the calculation results.

[0040] In one embodiment of the present application, the first medical prescription to be recommended and the second medical prescription to be recommended are determined respectively in two different ways. For the first way, the medical features in the first current condition information are scored respectively with different preset medical prescriptions, and the first medical prescription to be recommended based on the scoring results is determined, including: calculating the score of each medical prescription for the first current condition information based on the medical features in the first current condition information and the prescription features in each medical prescription, and the medical prescription includes prescription features of different categories; sorting the scores of each medical prescription for the first current condition information, and taking the medical prescription with the highest score as the first medical prescription to be recommended. Schematically, historical medical data can be collected in advance, and the historical medical data can be cleaned and prescription features extracted to obtain different preset medical prescriptions, each of which includes prescription features of at least one of the drug name, dosage, usage (frequency of medication), precautions, etc.

[0041] For the second method, that is, when determining the second medical prescription to be recommended, the interaction between different medical features in the second current condition information can be calculated through a pre-trained model, wherein the interaction between different medical features refers to the relationship of interaction and mutual influence between different medical features, and this interaction can be captured and represented by the inner product of latent vectors. Each medical feature corresponds to a latent vector, and the interaction strength between medical features is determined by the inner product of these latent vectors. The interaction strength between different medical features is comprehensively calculated by the trained model to obtain a predicted value, and the specific prescription information is parsed according to the predicted value as the second medical prescription to be recommended.

[0042] Step S240: determining a final recommended medical prescription according to the first medical prescription to be recommended and the second medical prescription to be recommended.

[0043] In one embodiment of the present application, at least one may be selected from the first medical prescription to be recommended and the second medical prescription to be recommended as the final recommended medical prescription. The first medical prescription to be recommended and the second medical prescription to be recommended may also be vector-converted respectively, and the vector corresponding to the first medical prescription to be recommended and the vector corresponding to the second medical prescription to be recommended obtained by the conversion are added, and then the added vector is text-converted to obtain the final recommended medical prescription. It is not difficult to understand that the content of the first medical prescription to be recommended may be partially the same as or different from the content of the second medical prescription to be recommended, and this application does not make specific limitations on this, depending on the actual application scenario.

[0044] In one embodiment of the present application, based on the medical features in the first current condition information and the prescription features in each medical prescription, the score of the first current condition information for each medical prescription is calculated, including: performing vector conversion on all medical features in the first current condition information and all prescription features in each medical prescription, respectively, to obtain all medical feature vectors of the first current condition information and all prescription feature vectors of each medical prescription; based on all medical feature vectors of the first current condition information and all prescription feature vectors of a medical prescription, calculating the similarity between the first current condition information and the medical prescription as the score of the first current condition information for the medical prescription, so as to obtain the score of the first current condition information for each medical prescription.

[0045] In this embodiment, all medical features in the first current condition information and all prescription features of each medical prescription can be converted into numerical vectors, respectively, to obtain all medical feature vectors of the first current condition information and all prescription feature vectors of each medical prescription. Then, taking the calculation of the score of a medical prescription by the first current condition information as an example, the similarity between the first current condition information and the medical prescription is calculated based on all medical feature vectors of the first current condition information and all prescription feature vectors of the medical prescription, and the similarity is used as the score of the first current condition information for the medical prescription, and so on, to obtain the scores of the first current condition information for different medical prescriptions.

[0046] In some embodiments, the similarity between the first current condition information and the medical prescription is calculated based on all medical feature vectors of the first current condition information and all prescription feature vectors of the medical prescription, including calculating the cosine similarity, Euclidean distance or Manhattan distance between the first current condition information and the medical prescription based on all medical feature vectors of the first current condition information and all prescription feature vectors of the medical prescription as the similarity between the first current condition information and the medical prescription.

[0047] Taking the calculation of the cosine similarity between the first current condition information and the medical prescription as an example, a high-dimensional vector of the first current condition information is formed based on all medical feature vectors of the first current condition information, and a high-dimensional vector of the medical prescription is formed based on all prescription feature vectors of the medical prescription. A dot product operation is performed on the high-dimensional vector of the first current condition information and the high-dimensional vector of the medical prescription, and the norm of the high-dimensional vector of the first current condition information and the norm of the high-dimensional vector of the medical prescription are calculated respectively. The dot product of the two is divided by the product of the two norms to obtain the cosine similarity between the first current condition information and the medical prescription.

[0048] In another embodiment of the present application, based on the medical features in the first current condition information and the prescription features in each medical prescription, the score of the first current condition information for each medical prescription is calculated, including: based on the medical features in each first condition information and the prescription features in each medical prescription, an initial first condition information-medical prescription scoring matrix is ​​established, and the initial first condition information-medical prescription scoring matrix is ​​matrix decomposed to obtain a first condition information feature matrix and a medical prescription feature matrix, wherein each medical prescription has corresponding first condition information, the first condition information includes medical features of different categories, and the categories of medical features in the first condition information are the same as the categories of medical features in the first current condition information; the first condition information feature matrix and the medical prescription feature matrix are alternately optimized until the convergence condition is reached to obtain a converged first condition information-medical prescription scoring matrix; according to the medical features in the first current condition information, the score of the first current condition information for each medical prescription is determined from the converged first condition information-medical prescription scoring matrix.

[0049] In this embodiment, an ALS (Alternating Least Squares) collaborative recommendation algorithm may be used to determine the first medical prescription to be recommended corresponding to the first current condition information. The following describes in detail the content of using the ALS collaborative recommendation algorithm to determine the first medical prescription to be recommended.

[0050] Specifically, in some embodiments, historical medical data of different patients can be collected through multiple channels or data sources such as the hospital's electronic medical record system, drug monitoring agency, and patient feedback platform to ensure data diversity and integrity, wherein historical medical data includes historical medical information and medical prescriptions issued for historical medical information, and the historical medical information includes different categories of medical features, and the categories and quantity of medical features in the historical medical information are the same as the number and categories of medical features in the current medical information. The historical medical information is divided to obtain first medical information and second medical information, wherein the categories and quantity of medical features in the first medical information are the same as the categories and quantity of medical features in the first current medical information.

[0051] Then, a correspondence between the first condition information and the medical prescription issued for the historical condition information to which the first condition information belongs is established, so that the first condition information corresponds one-to-one with the medical prescription, and a first condition information and its corresponding medical prescription are used as a first training sample, thereby obtaining multiple first training samples. Use multiple first training samples to train the ALS collaborative recommendation algorithm, including, based on each first training sample, calculating the similarity between the first condition information and the corresponding medical prescription as a known score or actual score to construct an initial first condition information-medical prescription scoring matrix. Since the initial first condition information-medical prescription scoring matrix is ​​a sparse matrix, it is missing some of the scores of the first condition information for the medical prescription. Therefore, it is necessary to optimize the initial first condition information-medical prescription scoring matrix to fill in the missing values ​​in the first condition information-medical prescription scoring matrix to obtain a complete and accurate first condition information-medical prescription scoring matrix. The specific optimization process is as follows:

[0052] 1. Randomly initialize the initial first condition information-medical prescription score matrix, that is, initialize the unscored elements in the matrix to 0 or other values. The initial value has a certain impact on the final result, but with the iteration of the algorithm, it will gradually converge to a more accurate result;

[0053] 2. Decomposing the first condition information-medical prescription score matrix into the product of two low-rank matrices, namely, the first condition information feature matrix and the medical prescription feature matrix, by using a matrix decomposition algorithm. The dimensions of the two low-rank matrices are the same and can be determined based on experience or experiments.

[0054] 3. Fix the first condition information feature matrix, solve the medical prescription feature matrix, and find the medical prescription feature matrix that minimizes the difference between the predicted score and the actual score by minimizing the loss function. The loss function can use the mean square error plus a regularization term to prevent overfitting.

[0055] 4. Fix the medical prescription feature matrix, solve the first condition information feature matrix, and also find the first condition information feature matrix that minimizes the difference between the predicted score and the actual score by minimizing the loss function;

[0056] 5. Alternate steps 3 and 4, continuously iteratively update the first condition information feature matrix and the medical prescription feature matrix until the convergence condition is reached, and obtain the converged first condition information-medical prescription scoring matrix, wherein the convergence condition includes that the loss function reaches a preset threshold or the number of alternating iterations reaches a predetermined number of alternating iterations.

[0057] In a specific embodiment of the present application, the ALS collaborative algorithm decomposes the sparse matrix, i.e., the initial first condition information-medical prescription score matrix R, into the product of two low-rank matrices (the first condition information feature matrix U and the medical prescription feature matrix V) through matrix decomposition, which is expressed as follows:

[0058] R=UV T Formula (1)

[0059] Wherein, R is the first condition information-medical prescription score matrix, U is the first condition information feature matrix, each row of which corresponds to a feature vector of all medical features in the first condition information, and the feature vector of the medical feature is obtained by performing vector transformation on the medical feature, V is the medical prescription feature matrix, each row of which corresponds to a feature vector of all prescription features in a medical prescription, and the feature vector of the prescription feature is obtained by performing vector transformation on the prescription feature, V T Represents the transposed matrix of V.

[0060] The ALS algorithm solves these two matrices by alternately minimizing the loss function about U and V. Specifically, first fix U and solve the least squares problem about V, then fix V and solve the least squares problem about U, and this process is repeated until convergence. The loss function can be MSE (Mean Squared Error), which is expressed as follows:

[0061]

[0062] Where L is the mean square error, Ω is the set of first condition information-medical prescription pairs with known scores, and R ij is the actual score of the first condition information i on the medical prescription j, U i and V j are the feature vectors of the medical features in the first condition information i and the feature vectors of the prescription features in the medical prescription j, respectively. j T Indicates V j The transposed matrix of , λ is the regularization parameter used to prevent overfitting, U is the first disease information feature matrix, V is the medical prescription feature matrix, and are the Frobenius norms of matrices U and V respectively.

[0063] In one embodiment of the present application, an initial first condition information-medical prescription scoring matrix is ​​established based on the medical features in each first condition information and the prescription features in each medical prescription, including: performing vector conversion on all medical features in each first condition information and all prescription features in each medical prescription, respectively, to obtain all medical feature vectors of each first condition information and all prescription feature vectors of each medical prescription; calculating the similarity between the first condition information and its corresponding medical prescription based on all medical feature vectors of a first condition information and all prescription feature vectors of the medical prescription corresponding to the first condition information, as the score of the first condition information for its corresponding medical prescription, to obtain the score of each first condition information for its corresponding medical prescription; establishing an initial first condition information-medical prescription scoring matrix based on each first condition information, each medical prescription and the score of each first condition information for its corresponding medical prescription.

[0064] In this embodiment, first, the various collected data are sorted and encoded so that they can be input into the ALS algorithm in digital form. This includes data cleaning of each first condition information and its corresponding medical prescription, and standardization or normalization of continuous data such as laboratory test values ​​so that they have similar numerical ranges. The medical features in each first condition information are encoded, and each medical feature corresponds to a vector, that is, a medical feature vector or a feature vector of medical features. Different categories of prescription features such as drug name, dosage, frequency of use, precautions, etc. in each medical prescription are also encoded, and each prescription feature corresponds to a vector, that is, a prescription feature vector or a feature vector of prescription features.

[0065] Then, based on all medical feature vectors of a first condition information and all prescription feature vectors of the medical prescription corresponding to the first condition information, the cosine similarity, Euclidean distance or Manhattan distance between the first condition information and its corresponding medical prescription can be calculated as the similarity between the first condition information and its corresponding medical prescription. The specific calculation method can refer to the method for calculating the similarity between the first current condition information and the medical prescription in the aforementioned embodiment, which will not be repeated here. By analogy, the similarity between each first condition information and its corresponding medical prescription is obtained.

[0066] Finally, based on each first condition information, each medical prescription, and the similarity between each first condition information and its corresponding medical prescription, an initial first condition information-medical prescription scoring matrix is ​​established. For example, a matrix is ​​constructed with each first condition information as a column, each medical prescription as a row, and the score as an element of the matrix, and the similarity between each first condition information and its corresponding medical prescription is added to the matrix as the known score or actual score of the first condition information to its corresponding medical prescription, and its position corresponds to both the first condition information and the medical prescription, thereby forming an initial first condition information-medical prescription scoring matrix.

[0067] In one embodiment of the present application, based on the medical features in the first current condition information, the score of the first current condition information for each medical prescription is determined from the converged first condition information-medical prescription scoring matrix, including: calculating the similarity between the first current condition information and each first condition information in the converged first condition information-medical prescription scoring matrix based on all medical features in the first current condition information and all medical features in each first condition information in the converged first condition information-medical prescription scoring matrix, and determining the first target condition information according to the similarity; reading the score of the first target condition information for each medical prescription from the converged first condition information-medical prescription scoring matrix as the score of the first current condition information for each medical prescription.

[0068] In this embodiment, the converged first condition information-medical prescription scoring matrix includes a complete and accurate score of each medical prescription for each first condition information. Therefore, medical prescriptions can be intelligently recommended based on the converged first condition information-medical prescription scoring matrix. For new patients, based on their first current condition information, the most suitable medical prescription is predicted and recommended by calculating the scores between the medical feature vector and the feature vectors of different prescriptions.

[0069] Indicatively, each first condition information is read from the converged first condition information-medical prescription scoring matrix, and the current first condition information is matched with each first condition information by calculating the similarity between the current first condition information and each first condition information. Specifically, the cosine similarity, Euclidean distance or Manhattan distance between the current first condition information and the first condition information can be calculated based on the vector of each medical feature in the current first condition information and the vector of each medical feature in a first condition information, as the similarity between the previous first condition information and the first condition information. The specific calculation method can refer to the method of calculating the similarity between the first current condition information and the medical prescription in the aforementioned embodiment, which will not be repeated here. By analogy, the similarity between the current first condition information and each first condition information is obtained.

[0070] The first condition information with the highest similarity is taken as the first target condition information, and the medical prescriptions corresponding to the first target condition information and the scores of these medical prescriptions given by the first target condition information are read from the converged first condition information-medical prescription scoring matrix, and the medical prescription with the highest score among the medical prescriptions corresponding to the first target condition information is taken as the first medical prescription to be recommended.

[0071] In one embodiment of the present application, the interaction between different medical features in the second current condition information is calculated, and the second medical prescription to be recommended is determined based on the calculation result, including: obtaining multiple second condition information and medical prescriptions corresponding to each second condition information, wherein the second condition information includes medical features of different categories, and the categories of medical features in the second condition information are the same as the categories of medical features in the second current condition information; based on the multiple second condition information and the medical prescriptions corresponding to each second condition information, a preset model is trained to determine the second medical prescription to be recommended through the trained model.

[0072] In this embodiment, historical medical data of different historical patients can be collected through multiple channels or data sources such as the hospital's electronic medical record system, drug monitoring agency, and patient feedback platform to ensure the diversity and integrity of the data, wherein the historical medical data includes historical medical information and medical prescriptions issued for the historical medical information, and the historical medical information includes different categories of medical features, and the categories and quantities of medical features in the historical medical information are the same as the categories and quantities of medical features in the current medical information. The historical medical information is divided to obtain the first medical information and the second medical information, wherein the categories and quantities of medical features in the second medical information are the same as the categories and quantities of medical features in the second current medical information. A corresponding relationship between the second medical information and the medical prescription issued for the historical medical information to which the second medical information belongs is established, so that the second medical information corresponds to the medical prescription one by one, and a second medical information and its corresponding medical prescription are used as a second training sample, thereby obtaining multiple second training samples. The preset model is trained using multiple second training samples until convergence, so that the interaction between different medical features in the second current medical information is calculated by the trained model, and the second medical prescription to be recommended is determined based on the interaction calculation result.

[0073] In one embodiment of the present application, a preset model is trained based on multiple second condition information and the medical prescriptions corresponding to each second condition information to determine the second medical prescription to be recommended through the trained model, including: taking all medical features in each second condition information as input variables of the model, and taking the medical prescription corresponding to each second condition information as the target variable of the model, training the model to learn the relationship between the medical features and the medical prescription to obtain the trained model; inputting all medical features in the second current condition information into the trained model, calculating the interaction between different medical features in the second current condition information according to the learned relationship between the medical features and the medical prescription through the trained model, and determining the second medical prescription to be recommended based on the calculation results.

[0074] In this embodiment, a FM (Factor Machine) algorithm can be used to determine the second medical prescription to be recommended corresponding to the second current condition information. Schematically, parameters of the FM model are set, including a global bias, a feature weight vector, and a feature latent vector matrix, and the FM model is trained using multiple second training samples. The medical features in the second condition information in the same second training sample are used as input variables of the FM model, and the medical prescriptions in the same second training sample are used as target variables of the FM model, so that the FM model learns the relationship between the medical features in the second condition information and the medical prescriptions. The trained FM model can automatically capture the interaction between different medical features, so as to better predict the recommended medical prescriptions. The medical features in the second current condition information are input into the trained FM model, and the FM model can predict the second medical prescription to be recommended corresponding to the medical features in the second current condition information.

[0075] In one embodiment of the present application, the interaction between different medical features in the second current condition information is calculated based on the relationship between the learned medical features and the medical prescriptions through the trained model, and the second medical prescription to be recommended is determined based on the calculation results, including: combining different medical features in the second current condition information to obtain different feature pairs; calculating the inner product of the latent vectors of the medical features in each feature pair to characterize the interaction between different medical features in the second current condition information; multiplying the inner product of the latent vector corresponding to each feature pair by the eigenvalue of the medical feature in the feature pair, and summing the multiplication results of each feature pair to obtain an interaction term, wherein the eigenvalue of the medical feature is obtained based on encoding the medical feature; obtaining a predicted value based on the interaction term, and parsing specific prescription information according to the predicted value as the second medical prescription to be recommended.

[0076] In addition, the linear relationship of each medical feature in the second current condition information can also be obtained, including calculating the linear weighted sum of each medical feature in the second current condition information, and summing the global offset with the linear weighted sum to obtain a linear term to characterize the linear relationship of each medical feature in the second current condition information. Then, the linear term and the interaction term are summed to obtain a predicted value, so as to parse out specific prescription information according to the predicted value as the second medical prescription to be recommended.

[0077] In one embodiment of the present application, step S240 includes: fusing the first medical prescription to be recommended and the second medical prescription to be recommended to obtain a comprehensive recommended medical prescription; sending the comprehensive recommended medical prescription to the medical terminal to receive feedback on the comprehensive recommended medical prescription returned by the medical terminal; and revising the comprehensive recommended medical prescription based on the feedback to obtain a final recommended medical prescription.

[0078] In this embodiment, the fusion processing can be performed first and then the review and feedback can be performed. If the first medical prescription to be recommended and the second medical prescription to be recommended are the same, any one of the first medical prescription to be recommended and the second medical prescription to be recommended is selected as the comprehensive recommended medical prescription. If the first medical prescription to be recommended and the second medical prescription to be recommended are different, then based on the drugs in the first medical prescription to be recommended and the second medical prescription to be recommended, evaluate whether the functional utility of the first medical prescription to be recommended and the second medical prescription to be recommended are the same. When the functional utility is the same, any one of the first medical prescription to be recommended and the second medical prescription to be recommended is selected as the comprehensive recommended medical prescription. When the functional utility is different, the first medical prescription to be recommended and the second medical prescription to be recommended are added to obtain a comprehensive recommended medical prescription. The above-mentioned fusion processing can remove duplicate medical prescriptions or medical prescriptions with the same functional utility.

[0079] In addition, if the first medical prescription to be recommended and the second medical prescription to be recommended are different, and the difference is only in the dosage, a new dosage is determined according to the dosage of the first medical prescription to be recommended and the second medical prescription to be recommended, and a comprehensive recommended medical prescription is generated based on the new dosage and the drugs and the frequency of use of the first medical prescription / the second medical prescription to be recommended. If the first medical prescription to be recommended and the second medical prescription to be recommended are different, and the difference is only in the frequency of use, a new frequency of use is determined according to the frequency of use of the first medical prescription and the second medical prescription to be recommended, and a comprehensive recommended medical prescription is generated based on the new frequency of use and the drugs and dosage of the first medical prescription / the second medical prescription to be recommended. If the first medical prescription to be recommended and the second medical prescription to be recommended are different, and the difference is in the dosage and the frequency of use, a new dosage is determined according to the dosage of the first medical prescription and the second medical prescription to be recommended, and a new frequency of use is determined according to the frequency of use of the first medical prescription and the second medical prescription to be recommended, and a comprehensive recommended medical prescription is generated based on the new dosage, the new frequency of use and the drugs of the first medical prescription / the second medical prescription to be recommended.

[0080] The comprehensive recommended medical prescription is output through the medical terminal for medical experts to review and provide feedback on the comprehensive recommended medical prescription. The feedback may include whether the comprehensive recommended medical prescription is correct, and when the comprehensive recommended medical prescription is wrong, it also includes the correction information of the comprehensive recommended medical prescription by the medical experts. Finally, if the feedback is correct, the comprehensive recommended medical prescription is directly used as the final recommended medical prescription. If the feedback is wrong, the comprehensive recommended medical prescription is corrected according to the feedback to obtain the final recommended medical prescription.

[0081] In another embodiment of the present application, step S240 includes: sending the first medical prescription to be recommended and the second medical prescription to be recommended to the medical terminal, to receive a first feedback on the first medical prescription to be recommended and a second feedback on the second medical prescription to be recommended returned by the medical terminal; correcting the first recommended medical prescription according to the first feedback to obtain a corrected first recommended medical prescription, and correcting the second recommended medical prescription according to the second feedback to obtain a corrected second recommended medical prescription; and fusing the corrected first medical prescription to be recommended and the corrected second medical prescription to be recommended to obtain a final recommended medical prescription.

[0082] In this embodiment, the review and feedback can be conducted first and then integrated. The first medical prescription to be recommended and the second medical prescription to be recommended are output through the medical terminal, so that the medical expert can review the first medical prescription to be recommended and the second medical prescription to be recommended, and provide feedback, including the first feedback and the second feedback, wherein the first feedback (or the second feedback) can include whether the first recommended medical prescription (or the second recommended medical prescription) is correct, and also include the medical expert's correction information on the first recommended medical prescription (or the second recommended medical prescription) when the first recommended medical prescription (or the second recommended medical prescription) is wrong.

[0083] If both the first feedback and the second feedback are correct, the first medical prescription to be recommended and the second medical prescription to be recommended are directly fused to obtain the final recommended medical prescription. If the first feedback is correct and the second feedback is wrong, the second recommended medical prescription is first corrected according to the second feedback, and then the first medical prescription to be recommended and the corrected second medical prescription to be recommended are fused to obtain the final recommended medical prescription. If the first feedback is wrong and the second feedback is correct, the first recommended medical prescription is first corrected according to the first feedback, and then the second medical prescription to be recommended and the corrected first medical prescription to be recommended are fused to obtain the final recommended medical prescription. If both the first feedback and the second feedback are wrong, the first recommended medical prescription is first corrected according to the first feedback, and the second recommended medical prescription is corrected according to the second feedback, and then the corrected first medical prescription to be recommended and the corrected second medical prescription to be recommended are fused to obtain the final recommended medical prescription. The specific fusion processing method can refer to the contents recorded in the above embodiment, which will not be repeated here.

[0084] In addition, the first feedback and the second feedback can also be used for secondary training of the ALS collaborative recommendation algorithm or the FM algorithm, so that the ALS collaborative recommendation algorithm or the FM algorithm can be continuously updated and optimized to improve the accuracy and practicality of the recommendation.

[0085] See also Figure 3 , Figure 3 is a block diagram of a medical prescription recommendation device shown in an exemplary embodiment of the present application. The device can be applied to Figure 1 The implementation environment shown in the figure is specifically configured in the computer device 120. The device can also be applied to other exemplary implementation environments and specifically configured in other devices. This embodiment does not limit the implementation environment to which the device is applied.

[0086] like Figure 3 As shown, the exemplary medical prescription recommendation device includes: a data acquisition module 310, which is used to obtain current condition information, and the current condition information includes different categories of medical features; an information processing module 320, which is used to divide the current condition information to obtain a first current condition information and a second current condition information, and the categories of medical features in the first current condition information are different from the categories of medical features in the second current condition information; according to the first medical prescription to be recommended and the second medical prescription to be recommended, a final recommended medical prescription is determined; a first recommendation module 330, which is used to score the medical features in the first current condition information with different preset medical prescriptions, and determine the first medical prescription to be recommended based on the scoring results; a second recommendation module 340, which is used to calculate the interaction between different medical features in the second current condition information, and determine the second medical prescription to be recommended based on the calculation result.

[0087] In this embodiment, the data acquisition module 310 can be a medical terminal such as a computer, a tablet, or a mobile phone, such as at least one of a medical monitoring terminal, a medical imaging terminal, an electronic medical record terminal, a telemedicine terminal, and an intelligent medical device terminal. The data acquisition module 310 can also be a hardware device for collecting data from various medical terminals, such as a data transceiver. The information processing module 320 can be a computer device or a server. The first recommendation module 330 and the second recommendation module 340 can be processing chips, computer devices, or servers. For example, the first recommendation module 330 and the second recommendation module 340 can be respectively configured as processing chips in the information processing module 320, or can be computer devices or servers that are independent of the information processing module 320. The data acquisition module 310, the information processing module 320, the first recommendation module 330, and the second recommendation module 340 can be configured in the computer device 120, or can be independent devices. There is no restriction here.

[0088] It should be noted that the medical prescription recommendation device provided in the above embodiment and the medical prescription recommendation method provided in the above embodiment belong to the same concept, wherein the specific manner in which each module and unit performs the operation has been described in detail in the method embodiment and will not be repeated here. In actual application, the medical prescription recommendation device provided in the above embodiment can allocate the above functions to different functional modules as needed, that is, divide the internal structure of the device into different functional modules to complete all or part of the functions described above, and this is not limited here.

[0089] An embodiment of the present application also provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the electronic device implements the medical prescription recommendation method provided in the above-mentioned embodiments.

[0090] See also Figure 4 , Figure 4 The structure diagram of the computer system suitable for implementing the electronic device of the embodiment of the present application is shown. It should be noted that: Figure 4 The computer system 400 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0091] like Figure 4 As shown, the computer system 400 includes a central processing unit 401, which can perform various appropriate actions and processes according to the program stored in the read-only memory 402 or the program loaded from the storage part 408 into the random access memory 403, such as executing the method in the above embodiment. In the random access memory 403, various programs and data required for system operation are also stored. The central processing unit 401, the read-only memory 402 and the random access memory 403 are connected to each other through a bus 404. The input / output interface 405 is also connected to the bus 404.

[0092] The following components are connected to the input / output interface 405: an input section 406 including a keyboard, a mouse, etc.; an output section 407 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the input / output interface 405 as needed. A removable medium 411, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 410 as needed so that a computer program read therefrom is installed into the storage section 408 as needed.

[0093] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 409, and / or installed from a removable medium 411. When the computer program is executed by the central processing unit 401, various functions defined in the system of the present application are executed.

[0094] It should be noted that the computer-readable medium shown in the embodiment of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a computer-readable computer program is carried. This propagated data signal can take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. A computer program contained on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0095] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the apparatus, methods and computer program products according to various embodiments of the present application. Among them, each box in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0096] The units involved in the embodiments described in this application may be implemented by software or hardware, and the units described may also be set in a processor. The names of these units do not, in some cases, constitute limitations on the units themselves.

[0097] Another aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor of a computer, causes the computer to execute the medical prescription recommendation method as described above. The computer-readable storage medium may be included in the electronic device described in the above embodiment, or may exist independently without being assembled into the electronic device.

[0098] The above embodiments are merely illustrative of the principles and effects of the present application, and are not intended to limit the present application. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed in the present application shall still be covered by the claims of the present application.

Claims

1. A recommended method for medical prescription, characterized in that Applied to computer equipment, the recommended method of medical prescription includes: Obtain current condition information, which includes different categories of medical characteristics; The current condition information is divided to obtain first current condition information and second current condition information, wherein the categories of medical features in the first current condition information are different from the categories of medical features in the second current condition information; Scoring the medical features in the first current condition information and different preset medical prescriptions respectively, determining a first medical prescription to be recommended based on the scoring results, and calculating the interaction between different medical features in the second current condition information, and determining a second medical prescription to be recommended based on the calculation results; Based on the first to be recommended medical prescription and the second to be recommended medical prescription, the final recommended medical prescription is determined.

2. The recommended method of medical prescription according to claim 1, characterized in that The medical characteristics in the first current condition information are respectively scored with the preset different medical prescriptions, and the first medical prescription to be recommended is determined based on the scoring results, including: Calculating the score of each medical prescription based on the first current condition information based on the medical features in the first current condition information and the prescription features in each medical prescription, wherein the medical prescription includes prescription features of different categories; The first current condition information is used to sort the scores of each medical prescription, and the highest-rated medical prescription is used as the first to be recommended medical prescription.

3. The recommended method of medical prescription according to claim 2, characterized in that Based on the medical characteristics in the first current condition information and the prescription characteristics in each medical prescription, the score of the first current condition information for each medical prescription is calculated, including: Based on the medical features in each first condition information and the prescription features in each medical prescription, an initial first condition information-medical prescription scoring matrix is ​​established, and the initial first condition information-medical prescription scoring matrix is ​​matrix decomposed to obtain a first condition information feature matrix and a medical prescription feature matrix, wherein each medical prescription has corresponding first condition information, the first condition information includes medical features of different categories, and the categories of the medical features in the first condition information are the same as the categories of the medical features in the first current condition information; The first condition information feature matrix and the medical prescription feature matrix are alternately optimized until the convergence conditions are reached, and the convergence first condition information-medical prescription scoring matrix is ​​obtained; According to the medical characteristics in the first current condition information, the score of each medical prescription by the first current condition information is determined from the converged first condition information-medical prescription scoring matrix.

4. The recommended method of medical prescription according to claim 3, characterized in that According to the medical features in the first current condition information, determining the score of each medical prescription by the first current condition information from the converged first condition information-medical prescription score matrix includes: Calculate the similarity between the first current condition information and each first condition information in the converged first condition information-medical prescription scoring matrix according to all medical features in the first current condition information and all medical features in each first condition information in the converged first condition information-medical prescription scoring matrix, and determine the first target condition information according to the similarity; The scores of the first target condition information for each medical prescription are read from the converged first condition information-medical prescription score matrix as the scores of the first current condition information for each medical prescription.

5. The recommended method of medical prescription according to claim 2, characterized in that Based on the medical characteristics in the first current condition information and the prescription characteristics in each medical prescription, the score of the first current condition information for each medical prescription is calculated, including: Respectively performing vector conversion on all medical features in the first current condition information and all prescription features in each medical prescription to obtain all medical feature vectors of the first current condition information and all prescription feature vectors of each medical prescription; Based on all medical feature vectors of the first current condition information and all prescription feature vectors of a medical prescription, the similarity between the first current condition information and the medical prescription is calculated as the score of the first current condition information for the medical prescription, so as to obtain the score of the first current condition information for each medical prescription.

6. The recommended method of medical prescription according to claim 1, characterized in that The interaction between different medical characteristics in the second current condition information is calculated, and the second medical prescription to be recommended is determined based on the calculation results, including: Acquire multiple second condition information and medical prescriptions corresponding to each second condition information, wherein the second condition information includes medical features of different categories, and the categories of the medical features in the second condition information are the same as the categories of the medical features in the second current condition information; Based on the medical prescriptions corresponding to the plurality of second condition information and the second condition information, the preset model is trained to determine the second recommended medical prescription through the trained model.

7. The recommended method of medical prescription according to claim 6, characterized in that Based on the medical prescription corresponding to the plurality of second condition information and the second condition information, the preset model is trained to determine the second recommended medical prescription through the trained model, including: Using all medical features in each second condition information as input variables of the model, and using the medical prescription corresponding to each second condition information as the target variable of the model, training the model to learn the relationship between the medical features and the medical prescription, and obtaining the trained model; All medical features in the second current condition information are input into the trained model, and the trained model is used to calculate the interaction between different medical features in the second current condition information based on the relationship between the learned medical features and the medical prescriptions, and the second medical prescription to be recommended is determined based on the calculation results.

8. A medical prescription recommendation device, characterized in that The medical prescription recommendation device comprises: The data collection module is used to obtain current condition information, which includes different categories of medical characteristics; an information processing module, configured to divide the current condition information to obtain first current condition information and second current condition information, wherein the categories of medical features in the first current condition information are different from the categories of medical features in the second current condition information; and determine a final recommended medical prescription based on the first medical prescription to be recommended and the second medical prescription to be recommended; The first recommendation module is used to score the medical characteristics in the first current condition information with the preset different medical prescriptions, and to determine the first recommended medical prescription based on the scoring results; The second recommendation module is used to calculate the interaction between different medical characteristics in the second current condition information, and determine the second medical prescription to be recommended based on the calculation results.

9. An electronic device, characterized in that: The electronic device comprises: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the electronic device to implement the medical prescription recommendation method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: There is a computer program stored thereon, and when the computer program is executed by a processor of the computer, the computer is caused to perform the recommended method of medical prescription as claimed in any one of claims 1-7.

Images