WO2021179694A1

WO2021179694A1 - Drug recommendation method, apparatus, computer device, and storage medium

Info

Publication number: WO2021179694A1
Application number: PCT/CN2020/132479
Authority: WO
Inventors: 徐卓扬; 赵惟; 左磊; 孙行智; 胡岗
Original assignee: 平安科技（深圳）有限公司
Priority date: 2020-10-09
Filing date: 2020-11-27
Publication date: 2021-09-16
Also published as: CN111933302A; CN111933302B

Abstract

Provided are a drug recommendation method, apparatus, computer device, and storage medium; the method is suitable for use in the field of digital medicine, and comprises: obtaining at least two types of target user attribute information of a target user, and inputting the at least two types of target user attribute information into a medication index prediction model (S101); on the basis of a medication index prediction network in the medication index prediction model, determining the medication index of each target user attribute information of the target user under the action of each drug (S102); on the basis of the medication index prediction model, determining a target medication index corresponding to at least two target user attribute information under the action of any drug so as to obtain a plurality of target medication indicators corresponding to a plurality of drugs (S103); determining a maximum target medication index from among the plurality of target medication indexes, and pushing to the target user the target drug having the largest target medication index (S104). Using the method, it is possible to accurately recommend drugs to a user, improving the accuracy of model drug recommendations, and also increasing the user stickiness of drug recommendations.

Description

Drug recommendation method, device, computer equipment and storage medium

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on October 9, 2020, the application number is 202011070302.8, and the invention title is "method, device, computer equipment and storage medium for drug recommendation", the entire content of which is incorporated by reference In this application.

Technical field

This application relates to the field of computer technology, and in particular to a method, device, computer equipment and storage medium for drug recommendation.

Background technique

Currently, drugs can be recommended to patients through neural network models or statistical models. The inventor realized that when recommending drugs to a patient through a neural network model, the neural network model can predict the drugs that need to be recommended to the patient based on all the characteristics of the patient, but because the neural network shares the characteristics of neurons, it can only The characteristics are explained, but a certain characteristic of the patient cannot be explained. As a result, the patient may not accept the drugs recommended by the above-mentioned model. It can be seen that the above-mentioned neural network model lacks sufficient interpretability and poor applicability. In addition, when recommending drugs to patients through a statistical model, although the statistical model has sufficient interpretability, the accuracy of the model is low.

Summary of the invention

This application provides a drug recommendation method, device, computer device, and storage medium, which can accurately recommend drugs to users, improve the accuracy of model drug recommendation, and also increase the user stickiness of drug recommendation.

In the first aspect, this application provides a drug recommendation method, which includes:

Obtain at least two target user attribute information of the target user, and input the at least two target user attribute information into the medication index prediction model. The medication index prediction model includes at least two medication index prediction networks, and one medication index prediction network is used to output the user’s information Multiple medication indicators of a user attribute information under the action of multiple drugs;

Based on the medication index prediction network in the medication index prediction model, determine the medication index of each target user attribute information of the target user under the action of each drug, and one type of target user attribute information corresponds to a medication index under the action of a drug;

Determine the target drug use indicators corresponding to at least two target user attribute information under the action of any drug based on the drug use index prediction model to obtain multiple target drug use indicators corresponding to multiple drugs;

Determine the maximum target medication index from multiple target medication indicators, and push the target drug with the largest target medication index to the target user.

In the second aspect, the present application provides a drug recommendation device, which includes:

The first acquisition module is used to acquire at least two types of target user attribute information of the target user, and input the at least two types of target user attribute information into the medication index prediction model. The medication index prediction model includes at least two medication index prediction networks, one medication index The prediction network is used to output multiple medication indicators of a user attribute information of the user under the action of multiple drugs;

The first determination module is used to determine the medication index of each target user attribute information of the target user under the action of each drug based on the drug index prediction network in the medication index prediction model, and one type of target user attribute information corresponds to the action of a drug A medication index;

The second determining module is configured to determine the target drug use indicators corresponding to at least two target user attribute information under any drug action based on the drug use indicator prediction model to obtain multiple target drug indicators corresponding to multiple drugs;

The push module is used to determine the maximum target medication indicator from multiple target medication indicators, and push the target drug with the largest target medication indicator to the target user.

In the third aspect, this application provides a computer device, including: a processor, a memory, and a network interface;

The processor is connected to a memory and a network interface, where the network interface is used to provide data communication functions, the memory is used to store a computer program, and the processor is used to call the computer program to execute the following methods:

In a fourth aspect, the present application provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, the computer program includes program instructions, and when executed by a processor, the program instructions execute the following methods:

In this application, each medication index is determined through the medication index prediction network, and the target medication index is further determined according to each medication index, thereby improving the accuracy of the model for predicting the target medication index. Further, the computer device can determine multiple target medication indicators corresponding to multiple drugs based on the medication indicator prediction model, and push the target drug with the largest target medication indicator to the target user, thereby accurately recommending medication to the user and improving The accuracy of model drug recommendation is improved, and the user stickiness of drug recommendation is also improved, and the applicability is strong.

Description of the drawings

In order to explain the technical solutions in this application more clearly, the following will briefly introduce the drawings needed in the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the application. For those of ordinary skill in the art, without creative work, other drawings can be obtained from these drawings.

Figure 1 is a schematic structural diagram of the network architecture provided by this application;

Figure 2 is a schematic flow chart of the drug recommendation method provided by the present application;

Figure 3 is a schematic diagram of the structure of the medication index prediction model provided by this application;

4 is a schematic diagram of the structure of the drug recommendation device provided by the present application;

Fig. 5 is a schematic diagram of the structure of the computer equipment provided by the present application.

Detailed ways

The technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are only a part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of this application.

The technical solution of this application can be applied to the fields of artificial intelligence, smart city, digital medical, blockchain and/or big data technology to realize intelligent drug recommendation. Optionally, the data involved in this application, such as user attribute information, medication indicators, and/or pushed medication information, can be stored in a database, or can be stored in a blockchain, such as distributed storage through a blockchain, this application Not limited.

Please refer to Fig. 1, which is a schematic structural diagram of the network architecture provided by the present application. As shown in FIG. 1, the network architecture may include a server 10 and a user terminal cluster. The user terminal cluster may include multiple user terminals. As shown in FIG. 1, it may specifically include a user terminal 100a, a user terminal 100b, and a user terminal 100c. ..., the user terminal 100n.

Among them, the server 10 may be an independent physical server, or it may provide cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content distribution networks ( content delivery network (CDN), big data and artificial intelligence platform and other basic cloud computing services cloud server. Each user terminal in the user terminal cluster may include, but is not limited to: smart phones, tablet computers, notebook computers, desktop computers, smart speakers, smart watches, and other smart terminals.

It is understandable that the computer device in the present application may be a physical terminal with a drug recommendation function. The physical terminal may be the server 10 as shown in FIG. 1 or a user terminal, which is not limited here.

As shown in FIG. 1, the user terminal 100a, the user terminal 100b, the user terminal 100c, ..., the user terminal 100n can be connected to the aforementioned server 10 respectively, so that each user terminal can interact with the server 10 through the network connection. . For example, the server 10 may push the target drug to the user interface of the target user's user terminal (or target user terminal for short), and then the target user can view the target drug on the user interface, where the target user terminal may be a user terminal Any user terminal in the cluster (such as the user terminal 100a). The application may refer to the drug recommended to the target user determined based on the drug index prediction model as the target drug, and the application may also call the model with the function of predicting the target drug index of the target user under the action of multiple drugs as the target drug. It is a predictive model of medication index.

The drug recommendation method provided in this application may be applicable to a drug recommendation scenario for any disease, such as a diabetes drug recommendation scenario, a hypertension drug recommendation scenario, or a drug recommendation scenario for other diseases. Assuming that the target user is a doctor, the doctor can input the basic information of the patient into the clinical decision support system (CDSS), and the CDSS contains the above-mentioned medication index prediction model, which can output the recommended target drug based on the basic information of the patient To the user interface of the CDSS, the doctor can view the target drug on the user interface (the target drug here can be used as the preliminary diagnosis result), and then combine with the patient's further diagnosis results to determine the drug suitable for the patient (such as the above target drug). Assuming that the target user is a patient, the patient can input their basic information into self-service terminals (or self-service kiosks for short) provided by hospitals, health stations, or social health institutions. The self-service kiosks contain the above-mentioned medication index prediction model, which can be based on The basic information of the patient outputs the recommended target medicine to the user interface of the self-service machine. The patient can view the target drug in the user interface of the self-service machine, and the subsequent patient can directly purchase the target drug, or allow the doctor to further diagnose and determine the drug suitable for the patient (such as the above target drug).

For the convenience of description, the following will take the diabetic drug recommendation scenario as an example for description, which will not be repeated below. The drug recommendation method, drug recommendation device, and computer equipment of the present application will be described below with reference to FIGS. 2 to 5.

Please refer to FIG. 2, which is a schematic flowchart of the drug recommendation method provided by the present application. As shown in Figure 2, the method may include the following steps S101 to S104:

Step S101: Obtain at least two types of target user attribute information of the target user, and input the at least two types of target user attribute information into the medication index prediction model.

It can be understood that, before performing step S101, the computer device may first train the medication index prediction network in the medication index prediction model (such as neural additive models (NAM)) through the sample data of at least two users. Thus, each medication index prediction network used to output the medication index of each user attribute information of any user under the action of multiple drugs is obtained. Among them, the number of medication index prediction networks in the medication index prediction model is the same as the number of user attribute information of the user. It is assumed that the medication index prediction model corresponds to a neural network. The neural network can contain multiple sub-neural networks. The medication index prediction The network can be a sub-neural network.

In some feasible implementation manners, the computer device can obtain sample data of at least two users. Among them, the sample data of at least two users is used to train the medication index prediction model, one user corresponds to one sample data, and one sample data can include at least two types of user attribute information and the user's target sample medication index under the action of drugs. The target sample medication index may be the actual medication index of the user under the action of the drug. The drugs used by different users for the target disease can be the same or different, and the specific drugs can be determined according to the actual application scenario, which is not limited here. Further, the computer device can input the sample data of at least two users into the medication index prediction model, and perform joint learning on the user attribute information corresponding to each medication index prediction network through each medication index prediction network in the medication index prediction model to obtain the prediction task. The ability of the user's attribute information of each user to use the medication index under the action of each drug.

It can be understood that after obtaining the sample data of at least two users, the computer equipment performs data cleaning and feature screening on the sample data of at least two users, and finally forms a prediction about the user attribute information of any user under the action of each drug. Some characteristic data of medication indicators. The feature data here may include, but is not limited to, the user's age, gender, and health indicators for target diseases (such as diabetes) (such as glycosylated hemoglobin value and creatinine value) and other multiple dimension feature data, which is not limited here. Further, the computer device can input the sample data of at least two users into the medication index prediction model, and use the medication index prediction network in the medication index prediction model to predict the medication of each user attribute information of any user under the action of each drug. The index is a learning task. The user attribute information corresponding to each medication index prediction network is jointly learned to obtain each network parameter corresponding to each medication index prediction network, and based on each network parameter, determine the role of each user attribute information of any user in each drug The ability to follow the medication index. In other words, the computer device can perform data feature learning on the above-mentioned feature data based on each medication index prediction network to obtain the ability to predict the medication index of each user attribute information of any user under the action of each drug.

In the joint learning process of the medication index prediction network in the medication index prediction model, the computer device can calculate the loss value corresponding to each sample data through a loss function (such as a binary cross-entropy loss function). Further, the computer equipment can iteratively update the network parameters corresponding to each medication index prediction network according to the loss value corresponding to each sample data, stop training when the loss value is basically unchanged, and use the iteratively updated network parameters as each medication The index predicts the final network parameters of the network. At this time, it also shows that each medication index prediction network has the ability to predict the medication index of any user's user attribute information under the action of each drug. At the same time, the medication index prediction model has the ability to predict any user (such as the target user) in each drug. The ability to target medication indicators under the action.

For ease of understanding, please refer to Figure 3, which is a schematic structural diagram of the medication index prediction model provided in this application. As shown in Figure 3, the medication index prediction model 1 can include at least two medication index prediction networks (the medication index prediction network 1 to the medication index prediction network n in Figure 3, where n is a positive integer). For ease of description, the following The model training process of the medication index prediction model will be explained by taking at least two medication index prediction networks as medication index prediction network 1 to medication index prediction network n as examples, and will not be repeated here.

Assuming that the above-mentioned target disease is diabetes, the sample data of a user may contain at least two user attribute information (such as user attribute information X ₁ to user attribute information X _n ), and user attribute information X ₁ to user attribute information X _n may include But it is not limited to the user’s age, gender, glycosylated hemoglobin value, creatinine value and other health indicators for diabetes, and the drugs used by the user for diabetes in the above sample data are used as the above-mentioned medication index prediction network 1 to medication index prediction network n. The output dimension, the drugs used by all users for diabetes can include multiple drugs (such as two drugs, drug a (such as biguanide) and drug b (such as sulfonylurea)), and at the same time use the target medication index (such as drug a) The target medication index Y _a and the target medication index Y _{b under the action of the drug b are} used as the output of the medication index prediction model 1 to perform model training on the medication index prediction network 1 to the medication index prediction network n. The target medication index here can be the expected glycation compliance rate of the target user under the action of a drug.

The computer equipment can input the above sample data into the medication index prediction model 1. At this time, each user attribute information (user attribute information X ₁ to user attribute information X _n ) in the sample data can be compared with the medication index prediction network 1 to medication The index prediction network n is matched, and the matched medication index prediction network is used as the medication index prediction network to input user attribute information. For example, user attribute information X ₁ matches the medication index prediction network 1, and user attribute information X ₂ matches the medication index prediction The network 2 matches, ..., the medication index prediction network n matches the user attribute information X _n .

It can be understood that when the computer device inputs the above sample data into the medication index prediction model 1, it will _{input user attribute information X 1} into its matching medication index prediction network 1, and user attribute information X ₂ into its matching medication index prediction network 2...., input user attribute information X _n into its matched medication index prediction network n. _{Further, the computer equipment can perform joint learning of user attribute information X 1} to user attribute information X _n through the medication index prediction network 1 to the medication index prediction network n, so as to obtain the predicted user attribute information of any user (such as the target user’s Target user attribute information X ₁ to target user attribute information X _n ) the medication index under the action of drug a (such as f _1a (X ₁ ), f _2a (X ₂ ),..., f _na (X _n )) and the drug The ability of medication indicators (such as f _1b (X ₁ ), f _2b (X ₂ )..., f _nb (X _{n )) under the action of b.} Among them, f _1a (X ₁ ) is the medication index of any user's user attribute information X ₁ (such as target user attribute information X ₁ ) under the action of drug a, f _1b (X ₁ ) is the user attribute information X ₁ in the drug b Medication index under the action. f _2a (X ₂ ) is the medication index of any user's user attribute information X ₂ (such as target user attribute information X ₂ ) under the action of drug a, f _2b (X ₂ ) is the user attribute information X ₂ acting on the drug b Medication indicators under. By analogy, f _na (X _n ) is the user attribute information X _n of any user (such as target user attribute information X _n ) the medication index under the action of drug a, f _nb (X _n ) is the user attribute information X _n Medication index under the action of drug b. The medication index here may be the expected glycation compliance rate of a user attribute information of any user under the action of a drug.

At the same time, in the process of joint learning of the medication index prediction network 1 to the medication index prediction network n (the joint learning here may also be called joint training), the computer equipment passes the above f _1a (X ₁ ), f _2a (X ₂ ) _{_{, ..., f na (X n}} ) and a corresponding offset parameter medicament beta] is accumulated to obtain _a Y _a, and through said _{_{f 1b (X 1), f}} 2b (X 2), ..., f nb (X n) and a pharmaceutically offset parameter b corresponding beta] _b are accumulated to obtain Y _b. After obtaining Y _a and Y _b , the computer device can use the binary cross-entropy loss function to pair Y _a (or Y _b ) and the label parameters corresponding to the target sample medication index in the above sample data (for example, the glycosylated hemoglobin value reaches the standard of 1, The glycosylated hemoglobin value is not up to the standard and is 0) to calculate the loss value corresponding to the sample data. At this time, the computer device can iteratively update the bias parameters corresponding to each drug (such as the bias parameter β _a corresponding to drug a and the bias parameter β b corresponding to drug _b ) and the medication index prediction network according to the loss values corresponding to all sample data 1 to the medication index predicts each network parameter corresponding to the network n.

When the aforementioned loss value is basically unchanged, the joint training is stopped, and the computer device can determine the iteratively updated bias parameter β _a and the bias parameter β _b and each network parameter. At this time, the medication index prediction network 1 to the medication index prediction network n can predict the medication index of any user's user attribute information under the action of medication a and medication b based on the iteratively updated network parameters (such as the above f _1a ( X ₁ ), f _2a (X ₂ ), ..., f _na (X _n ), and f _1b (X ₁ ), f _2b (X ₂ ), ..., f _nb (X _n )). At the same time, the medication index prediction model 1 can also predict any user (such as the target user) in each drug (such as the above-mentioned drug a and drug b) _{based on the iteratively updated bias parameter β a} and the bias parameter β _{b and each network parameter.} Each target medication index under the action (such as the above Y _a and Y _b ).

After the medication index prediction model is trained, the computer device can obtain at least two types of target user attribute information of the target user, where at least two types of target user attribute information (such as target user attribute information X ₁ to target user attribute information X _n ) include One or more of the age, gender, and health indicators of the target disease (such as the above-mentioned glycosylated hemoglobin value and creatinine value) of the target user.

After acquiring at least two types of target user attribute information of the target user, the computer device can input at least two types of target user attribute information into the above-mentioned medication index prediction model (such as the above-mentioned medication index prediction model 1), and the attribute information of each target user can be respectively compared with The medication index prediction network 1 matches the medication index prediction network n, and the matched medication index prediction network is used as the medication index prediction network for inputting the attribute information of the target user. For example, the target user attribute information X ₁ matches the medication index prediction network 1, the target user attribute information X ₂ matches the medication index prediction network 2, ..., the target medication index prediction network n matches the user attribute information X _n . It can be understood that when the target user attribute information X ₁ to the target user attribute information X _{n are} input into the medication index prediction model 1, the computer device can _{input the target user attribute information X 1} to the medication index prediction network 1, and the target user attribute information X ₂ Input to the medication index prediction network 2,..., _{input the target user attribute information X n} to the medication index prediction network n, and further determine the target user attribute information of each target user through the medication index prediction network 1 to the medication index prediction network n Medication index under the action of each drug.

Step S102, based on the medication index prediction network in the medication index prediction model, determine the medication index of each target user attribute information of the target user under the action of each drug.

It can be understood that assuming that each medication index prediction network is the above medication index prediction network 1 to medication index prediction network n, the computer equipment can determine the target user attribute information X ₁ medication under the action of drug a through the above medication index prediction network 1 indicator f _1a (X ₁₎ and medication index f _1b (X ₁₎ under b action of drugs, by the above treatment predictor network 2 to determine the target user target user attribute information X ₂ medication index at the drug a function f _2a ( X ₂ _{) and the medication index f 2b} (X ₂ ) under the action of the drug b. So, the computer device determines the target user via the network administration predictor n X _n target user attribute information in the drug action of a drug f index _Na (X _n) and f indicators medication effects of drugs under b _Nb (X _n ), where a kind of user attribute information corresponds to a medication index under the action of a kind of medicine.

Step S103: Determine the target drug use indicators corresponding to at least two target user attribute information under the action of any drug based on the drug use index prediction model to obtain multiple target drug use indicators corresponding to multiple drugs.

In some feasible embodiments, the model parameters corresponding to the medication index prediction model may include the bias parameters corresponding to each drug (such as the above-mentioned drug a and the above-mentioned drug b) (such as the above-mentioned bias parameter β _a and the bias parameter β _b ) , A drug corresponds to a bias parameter (for example, the above-mentioned drug a corresponds to the bias parameter β _a and the drug b corresponds to the bias parameter β _b ). The computer equipment can accumulate the medication index of each target user attribute information under the action of any drug and the bias parameter corresponding to any drug based on the drug index prediction model to obtain at least two target user attribute information corresponding to any drug action Target medication index. _{Assuming that the target medication index is the target medication index Y a} of the target user under the action of the above-mentioned drug a, the following formula (1) can be used to determine Y _a :

Y _a =f _1a (X ₁ )+f _2a (X ₂ )+…+f _na (X _n )+β _a , (1)

Among them, f _1a (X ₁ ) can represent the target user's target user attribute information X ₁ medication index under the action of drug a, f _2a (X ₂ ) can represent the target user's target user attribute information X ₂ under the action of drug a The medication index of ,..., f _na (X _n ) can represent the medication index of the target user attribute information X _n of the target user under the action of the drug a, and β _a can represent the bias parameter corresponding to the drug a.

_{Assuming that the target medication index is the target medication index Y b} of the target user under the action of the above-mentioned medication b, the following formula (2) can be used to determine Y _b :

Y _b =f _1b (X ₁ )+f _2b (X ₂ )+…+f _nb (X _n )+β _b , (2)

Among them, f _1b (X ₁ ) can represent the target user's target user attribute information X ₁ medication index under the action of drug b, f _2b (X ₂ ) can represent the target user's target user attribute information X ₂ under the action of drug b The medication index of,..., f _nb (X _n ) can represent the medication index of the target user attribute information X _n of the target user under the action of the drug _b , and β b can represent the bias parameter corresponding to the drug b.

It can be understood that the computer device can output multiple target medication indicators corresponding to the target user under the action of multiple drugs through the medication indicator prediction model (for example, the target medication index Y _a under the action of drug a or the target medication indicator Y under the action of drug b. _b ), where the target user corresponds to a target medication index under the action of a drug.

Step S104: Determine the maximum target medication index from the multiple target medication indicators, and push the target drug with the largest target medication index to the target user.

In some feasible implementations, the computer device can sort multiple target medication indicators output by the medication indicator prediction model (for example, sort from large to small or from small to large) to obtain a sequence of target medication indicators, and compare the target medication indicators The first or last target medication index in the sequence is used as the maximum target medication index. At this time, the computer device can push the target drug with the largest target medication index to the user interface of the target user terminal (such as the user interface of the above-mentioned CDSS or the user interface of the self-service machine), and the target user can view the target on the user interface. drug.

In some feasible implementations, the multiple drugs include the target drug and at least one other drug other than the target drug, the medication index of the target user attribute information under the action of the target drug is the first medication index, and the target user attribute information is in other drugs. The medication index under the action of the drug is the second medication index. The computer device can determine the optimized medication index based on the first medication index and the second medication index, and display the optimized medication index to the user interface of the target user in a visual form (for example, a graph and/or table), so that the target user can easily access the user interface. View optimized medication indicators on the interface. Wherein, the optimized medication index may be the medication index difference between the first medication index and the second medication index. The optimized medication index is used to instruct the target user to use the target drug with respect to the use of the second drug under a kind of target user attribute information. The advantages. It can be seen that the computer equipment can display the optimized medication index to the target user in a visual form, and the target user can intuitively see the reason for recommending the target drug, thereby improving the interpretability of the model and improving the safety of medication recommendation Be sticky with users.

In this application, the computer device can determine the medication index of each target user attribute information of each target user under the action of each drug through the medication index prediction network in the medication index prediction model, and then determine the effect of any drug based on the medication index prediction model Below are at least two target medication indicators corresponding to the target user attribute information. Here, each medication indicator is determined through the medication indicator prediction network, and the target medication indicator is further determined according to each medication indicator, thereby improving the accuracy of the model predicting the target medication indicator. Further, the computer device can determine multiple target medication indicators corresponding to multiple drugs based on the medication indicator prediction model, and push the target drug with the largest target medication indicator to the target user, thereby accurately recommending medication to the user and improving The accuracy of model drug recommendation is improved, while the user stickiness of drug recommendation is improved, and the applicability is strong.

Further, please refer to FIG. 4, which is a schematic structural diagram of the drug recommendation device provided by the present application. The medication recommendation device may be a computer program (including program code) running in a computer device. For example, the medication recommendation device is an application software; the medication recommendation device may be used to perform corresponding steps in the method provided in this application. As shown in FIG. 4, the medication recommendation device 1 may include: a second acquisition module 10, a third acquisition module 20, a first acquisition module 30, a first determination module 40, a second determination module 50, a push module 60, and a display module 70.

The first obtaining module 30 is configured to obtain at least two types of target user attribute information of the target user, and input the at least two types of target user attribute information into the medication index prediction model. The medication index prediction model includes at least two medication index prediction networks, one medication The index prediction network is used to output multiple medication indicators of a user attribute information of the user under the action of multiple drugs;

The first determining module 40 is used to determine the medication index of each target user attribute information of the target user under the action of each drug based on the drug usage index prediction network in the medication index prediction model, and one type of target user attribute information is under the action of a drug Corresponding to a medication index;

The second determining module 50 is configured to determine the target drug use indicators corresponding to at least two target user attribute information under the action of any drug based on the drug use indicator prediction model to obtain multiple target drug indicators corresponding to multiple drugs;

The pushing module 60 is used to determine the maximum target medication index from a plurality of target medication indicators, and push the target drug with the largest target medication index to the target user.

In some feasible implementation manners, the at least two types of target user attribute information include at least one of the target user's age, gender, and health indicators for the target disease.

In some feasible implementations, the multiple drugs include the target drug and at least one other drug other than the target drug, the medication index of the target user attribute information under the action of the target drug is the first medication index, and the target user attribute information is in other drugs. The medication index under the action of the drug is the second medication index;

The above-mentioned drug recommendation device 1 further includes:

The display module 70 is configured to determine an optimized medication index based on the first medication index and the second medication index, and display the optimized medication index to the target user.

In some feasible implementation manners, the above-mentioned drug recommendation device 1 further includes:

The second acquisition module 10 is configured to acquire sample data of at least two users, and the sample data of one user includes at least two user attribute information of the user and the user's target sample medication index under the action of the sample drug;

The third acquisition module 20 is used to input the sample data of at least two users into the medication index prediction model, and perform joint learning on the user attribute information corresponding to each medication index prediction network through each medication index prediction network in the medication index prediction model to obtain The ability to predict the medication index of each user attribute information of any user under the action of each drug.

In some feasible embodiments, the medication index prediction model includes bias parameters corresponding to each drug, and one drug corresponds to one bias parameter;

The second determining module 50 includes: an accumulating unit 501.

The accumulation unit 501 is used to accumulate the medication index of each target user attribute information under the action of any drug and the bias parameter corresponding to any drug based on the medication index prediction model to obtain at least two target user attributes under the action of any drug The target medication index corresponding to the information.

For the specific implementation of the accumulating unit 501, reference may be made to the description of step S103 in the embodiment corresponding to FIG. 2, which will not be repeated here.

Among them, the specific implementation of the second acquisition module 10, the third acquisition module 20, the first acquisition module 30, the first determination module 40, the second determination module 50, the push module 60, and the display module 70 can be found in the above-mentioned FIG. 2 Corresponding to the description of step S101 to step S104 in the embodiment, the description will not be repeated here. In addition, the description of the beneficial effects of using the same method will not be repeated.

Further, please refer to FIG. 5, which is a schematic diagram of the structure of the computer device provided by the present application. As shown in FIG. 5, the computer device 1000 may be the server 10 in the embodiment corresponding to FIG. 1. The computer device 1000 may include: at least one processor 1001, such as a CPU, at least one network interface 1004, a user interface 1003, and a memory 1005, at least one communication bus 1002. Among them, the communication bus 1002 is used to implement connection and communication between these components. The user interface 1003 may include a display (display) and a keyboard (keyboard), and the network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed random access memory (RAM) memory, or a non-volatile memory (non-volatile memory), such as at least one disk memory. Optionally, the memory 1005 may also be at least one storage device located far away from the aforementioned processor 1001. As shown in FIG. 5, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a device control application program.

In the computer device 1000 shown in FIG. 5, the network interface 1004 is mainly used for network communication with a user terminal; and the user interface 1003 is mainly used to provide an input interface for the user; and the processor 1001 can be used to call the storage in the memory 1005 The device control application to achieve:

It should be understood that the computer device 1000 described in this application can execute the description of the drug recommendation method in the foregoing embodiment corresponding to FIG. 2 and the description of the drug recommendation device 1 in the foregoing embodiment corresponding to FIG. 4, I won't repeat them here. In addition, the description of the beneficial effects of using the same method will not be repeated.

In addition, it should be pointed out here that this application also provides a computer-readable storage medium, and the computer-readable storage medium stores the aforementioned computer program executed by the drug recommendation device 1, and the computer program includes Program instructions. When the processor executes the program instructions, it can execute the description of the drug recommendation method in the embodiment corresponding to FIG. 2 above, and therefore, it will not be repeated here. In addition, the description of the beneficial effects of using the same method will not be repeated. For technical details that are not disclosed in the embodiment of the computer-readable storage medium involved in this application, please refer to the description of the method embodiment of this application. As an example, the program instructions may be deployed to be executed on one computing device, or on multiple computing devices located in one location, or on multiple computing devices that are distributed in multiple locations and interconnected by a communication network Implementation, multiple computing devices distributed in multiple locations and interconnected through a communication network can form a blockchain system.

Optionally, the storage medium involved in this application, such as a computer-readable storage medium, may be non-volatile or volatile.

A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing relevant hardware through a computer program. The above-mentioned program can be stored in a computer readable storage medium. When executed, it may include the procedures of the above-mentioned method embodiments. The foregoing computer-readable storage medium may be the medication recommendation device provided in any of the foregoing embodiments or the internal storage unit of the foregoing device, such as a hard disk or memory of an electronic device. The computer-readable storage medium may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart media card (SMC), or a secure digital (SD) card equipped on the electronic device. Flash card, etc. The above-mentioned computer-readable storage medium may also include a magnetic disk, an optical disk, a read-only memory (read-only memory, ROM), or RAM. Further, the computer-readable storage medium may also include both an internal storage unit of the electronic device and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

The terms "first" and "second" in the claims, specification and drawings of the present application are used to distinguish different objects, rather than to describe a specific sequence. In addition, the terms "including" and "having" and any variations of them are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes unlisted steps or units, or optionally also includes Other steps or units inherent to these processes, methods, products or equipment. The reference to "embodiments" herein means that a specific feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The display of the phrase in various positions in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art clearly and implicitly understand that the embodiments described herein can be combined with other embodiments. The term "and/or" used in the specification of this application and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

A person of ordinary skill in the art may be aware that the units and algorithm steps of the examples described in the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of both, in order to clearly illustrate the hardware and software Interchangeability, in the above description, the composition and steps of each example have been generally described in accordance with the function. Professionals and technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered beyond the scope of this application.

The above-disclosed are only preferred embodiments of this application, and of course the scope of rights of this application cannot be limited by this. Therefore, equivalent changes made in accordance with the claims of this application still fall within the scope of this application.

Claims

A method of drug recommendation, including:

Obtain at least two types of target user attribute information of the target user, and input the at least two types of target user attribute information into the medication index prediction model. The medication index prediction model includes at least two medication index prediction networks, one medication index prediction network uses To output multiple medication indicators of a user attribute information of the user under the action of multiple drugs;

Based on the medication index prediction network in the medication index prediction model, determine the medication index of each target user attribute information of the target user under the action of each drug, and one type of target user attribute information corresponds to a medication index under the action of a drug ；

Determine the target drug use indicators corresponding to the at least two target user attribute information under the action of any drug based on the drug use index prediction model to obtain multiple target drug use indicators corresponding to multiple drugs;

The maximum target drug use index is determined from the multiple target drug use indicators, and the target drug having the maximum target drug use index is pushed to the target user.
The method according to claim 1, wherein the multiple drugs include the target drug and at least one other drug other than the target drug, and the target user attribute information is used under the action of the target drug The index is the first medication index, and the medication index of the target user attribute information under the action of the other drugs is the second medication index;

The method also includes:

An optimized medication index is determined based on the first medication index and the second medication index, and the optimized medication index is displayed to the target user.
The method according to claim 1 or 2, wherein the method further comprises:

Acquiring sample data of at least two users, where the sample data of one user includes at least two user attribute information of the user and the target sample medication index of the user under the action of the sample drug;

The sample data of the at least two users are input into the medication index prediction model, and the user attribute information corresponding to each medication index prediction network is jointly learned through each medication index prediction network in the medication index prediction model to obtain The ability to predict the medication index of each user attribute information of any user under the action of each drug.
The method according to claim 3, wherein the medication index prediction model comprises a bias parameter corresponding to each of the medications, and one medication corresponds to a bias parameter;

The determining the target drug use index corresponding to the at least two target user attribute information under the action of any drug based on the drug use index prediction model includes:

Based on the medication index prediction model, the medication index of each target user attribute information under the action of any drug and the bias parameter corresponding to the any drug are accumulated to obtain the at least two targets under the action of any drug Target medication index corresponding to user attribute information.
The method according to claim 1, wherein the at least two types of target user attribute information include at least one of the target user's age, gender, and health indicators for target diseases.
A drug recommendation device, including:

The first acquisition module is configured to acquire at least two types of target user attribute information of the target user, and input the at least two types of target user attribute information into a medication index prediction model, and the medication index prediction model includes at least two medication index prediction networks , A medication index prediction network is used to output multiple medication indicators of a user attribute information of the user under the action of multiple drugs;

The first determining module is used to determine the medication index of each target user attribute information of the target user under the action of each drug based on the drug use index prediction network in the drug use index prediction model, one type of target user attribute information is in one type Corresponding to a medication index under the action of the drug;

The second determining module is configured to determine the target drug use indicators corresponding to the at least two target user attribute information under the action of any drug based on the drug use indicator prediction model to obtain multiple target drug use indicators corresponding to multiple drugs;

The push module is used to determine the maximum target drug use index from the multiple target drug use indicators, and push the target drug with the maximum target drug use index to the target user.
7. The device according to claim 6, wherein the multiple drugs include the target drug and at least one other drug other than the target drug, and the target user attribute information is used under the action of the target drug The index is the first medication index, and the medication index of the target user attribute information under the action of the other drugs is the second medication index;

The device also includes:

The display module is configured to determine an optimized medication index based on the first medication index and the second medication index, and display the optimized medication index to the target user.
The device according to claim 6 or 7, wherein the device further comprises:

The second acquisition module is configured to acquire sample data of at least two users, and the sample data of one user includes at least two user attribute information of the user and the target sample medication index of the user under the action of the sample drug;

The third acquisition module is configured to input the sample data of the at least two users into the medication index prediction model, and use the medication index prediction network in the medication index prediction model to predict the user corresponding to each medication index network The attribute information is jointly learned to obtain the ability to predict the medication index of each user's attribute information of any user under the action of each drug.
8. The device of claim 8, wherein the medication index prediction model comprises a bias parameter corresponding to each of the medications, and one type of medication corresponds to a bias parameter;

The second determining module is specifically configured to:

Based on the medication index prediction model, the medication index of each target user attribute information under the action of any drug and the bias parameter corresponding to the any drug are accumulated to obtain the at least two targets under the action of any drug Target medication index corresponding to user attribute information.
7. The device according to claim 6, wherein the at least two types of target user attribute information include at least one of the target user's age, gender, and health indicators for target diseases.
A computer device, including: a processor, a memory, and a network interface;

The processor is connected to a memory and a network interface, where the network interface is used to provide data communication functions, the memory is used to store program codes, and the processor is used to call the program codes to execute the following methods:

Obtain at least two types of target user attribute information of the target user, and input the at least two types of target user attribute information into the medication index prediction model. The medication index prediction model includes at least two medication index prediction networks, one medication index prediction network uses To output multiple medication indicators of a user attribute information of the user under the action of multiple drugs;

Based on the medication index prediction network in the medication index prediction model, determine the medication index of each target user attribute information of the target user under the action of each drug, and one type of target user attribute information corresponds to a medication index under the action of a drug ；

Determine the target drug use indicators corresponding to the at least two target user attribute information under the action of any drug based on the drug use index prediction model to obtain multiple target drug use indicators corresponding to multiple drugs;

The maximum target drug use index is determined from the multiple target drug use indicators, and the target drug having the maximum target drug use index is pushed to the target user.
The computer device according to claim 11, wherein the multiple drugs include the target drug and at least one other drug other than the target drug, and the target user attribute information is affected by the target drug The medication index is the first medication index, and the medication index of the target user attribute information under the action of the other medication is the second medication index;

The processor is also used to execute:

An optimized medication index is determined based on the first medication index and the second medication index, and the optimized medication index is displayed to the target user.
The computer device according to claim 11 or 12, wherein the processor is further configured to execute:

Acquiring sample data of at least two users, where the sample data of one user includes at least two user attribute information of the user and the target sample medication index of the user under the action of the sample drug;

The sample data of the at least two users are input into the medication index prediction model, and the user attribute information corresponding to each medication index prediction network is jointly learned through each medication index prediction network in the medication index prediction model to obtain The ability to predict the medication index of each user attribute information of any user under the action of each drug.
11. The computer device according to claim 13, wherein the medication index prediction model comprises a bias parameter corresponding to each of the medications, and one type of medication corresponds to a bias parameter;

When the target medication index corresponding to the at least two types of target user attribute information under the action of any drug is determined based on the medication index prediction model, the following is specifically executed:

Based on the medication index prediction model, the medication index of each target user attribute information under the action of any drug and the bias parameter corresponding to the any drug are accumulated to obtain the at least two targets under the action of any drug Target medication index corresponding to user attribute information.
11. The computer device according to claim 11, wherein the at least two types of target user attribute information include at least one of the target user's age, gender, and health indicators for target diseases.
A computer-readable storage medium, the computer-readable storage medium stores a computer program, the computer program includes program instructions, and when the program instructions are executed by a processor, the following method is executed:

Obtain at least two types of target user attribute information of the target user, and input the at least two types of target user attribute information into the medication index prediction model. The medication index prediction model includes at least two medication index prediction networks, one medication index prediction network uses To output multiple medication indicators of a user attribute information of the user under the action of multiple drugs;

Based on the medication index prediction network in the medication index prediction model, determine the medication index of each target user attribute information of the target user under the action of each drug, and one type of target user attribute information corresponds to a medication index under the action of a drug ；

Determine the target drug use indicators corresponding to the at least two target user attribute information under the action of any drug based on the drug use index prediction model to obtain multiple target drug use indicators corresponding to multiple drugs;

The maximum target drug use index is determined from the multiple target drug use indicators, and the target drug having the maximum target drug use index is pushed to the target user.
The computer-readable storage medium according to claim 16, wherein the multiple drugs include the target drug and at least one other drug other than the target drug, and the target user attribute information is in the target drug The medication index under the action is the first medication index, and the medication index of the target user attribute information under the action of the other medication is the second medication index;

When the program instructions are executed by the processor, they are also used to execute:

An optimized medication index is determined based on the first medication index and the second medication index, and the optimized medication index is displayed to the target user.
The computer-readable storage medium according to claim 16 or 17, wherein, when the program instructions are executed by the processor, they are also used to execute:

Acquiring sample data of at least two users, where the sample data of one user includes at least two user attribute information of the user and the target sample medication index of the user under the action of the sample drug;

The sample data of the at least two users are input into the medication index prediction model, and the user attribute information corresponding to each medication index prediction network is jointly learned through each medication index prediction network in the medication index prediction model to obtain The ability to predict the medication index of each user attribute information of any user under the action of each drug.
18. The computer-readable storage medium of claim 18, wherein the medication index prediction model comprises a bias parameter corresponding to each of the medications, and one type of medication corresponds to a bias parameter;

When the target drug use index corresponding to the at least two target user attribute information under the action of any drug is determined based on the drug use index prediction model, the following is specifically executed:

Based on the medication index prediction model, the medication index of each target user attribute information under the action of any drug and the bias parameter corresponding to the any drug are accumulated to obtain the at least two targets under the action of any drug Target medication index corresponding to user attribute information.
The computer-readable storage medium according to claim 19, wherein the at least two types of target user attribute information include at least one of the target user's age, gender, and health indicators for target diseases.