WO2021114624A1

WO2021114624A1 - Artificial intelligence-based medication recommendation method, apparatus, device, and storage medium

Info

Publication number: WO2021114624A1
Application number: PCT/CN2020/099255
Authority: WO
Inventors: 赵惟; 徐卓扬; 左磊; 孙行智; 刘卓; 赵婷婷
Original assignee: 平安科技（深圳）有限公司
Priority date: 2020-05-29
Filing date: 2020-06-30
Publication date: 2021-06-17
Also published as: CN111667890A

Abstract

Provided are an artificial intelligence-based medication recommendation method, apparatus, device, and storage medium. The method comprises: obtaining a medication recommendation request, the medication recommendation request comprising disease type, current disease data, and user profile data (S201); according to the target analysis results of the original drug recommendation model corresponding to the type of disease, obtaining a target medication recommendation model (S202); using the target medication recommendation model to analyze and process current disease data to obtain an original recommended medication (S203); performing feature analysis on the original recommended medication to obtain a feature analysis value corresponding to a feature to be analyzed (S204); determining the type of recommendation tendency on the basis of user profile data, and according to the recommended tendency type, obtaining a feature weight corresponding to the feature to be analyzed (S205); on the basis of the feature analysis value and the feature weight corresponding to the feature to be analyzed, obtaining a recommended evaluation value of the original recommended medication, and determining the original recommended medication having the largest recommended evaluation value to be the target recommended medication (S206). The method can effectively improve the effectiveness and pertinence of smart drug recommendation.

Description

Artificial intelligence-based medication recommendation method, device, equipment and storage medium

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on May 29, 2020, the application number is 202010475961.3, and the invention title is "artificial intelligence-based drug recommendation methods, devices, equipment, and storage media", and its entire content Incorporated in this application by reference.

Technical field

This application relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based medication recommendation method, device, equipment, and storage medium.

Background technique

Clinical Decision Support System (Clinical Decision Support System, hereinafter referred to as CDSS system), generally refers to any computer system that can provide support for clinical decision-making. This system makes full use of available and appropriate computer technology to target semi-structured or unstructured It is a system to improve and improve the efficiency of decision-making through human-computer interaction. The CDSS system is an important means to improve the quality of medical care, that is, through the in-depth analysis of medical records through the CDSS system, so as to make the most appropriate diagnosis and treatment decisions. Its fundamental purpose is to evaluate and improve medical quality, reduce medical errors, and control medical expenses. .

The medication recommendation model integrated in the current CDSS system is a model used to recommend medication by analyzing disease data. It is mainly divided into two types: knowledge-driven medication recommendation model and data-driven medication recommendation model. The knowledge-driven medication recommendation model is a medication recommendation model based on various medical guidelines and expert consensus. The data-driven medication recommendation model is a medication recommendation model formed by integrating various artificial intelligence technologies such as machine learning and deep learning. The inventor realizes that the current evaluation of the quality of the medication recommendation model is mainly based on judging whether the recommended medication matches the clinically prescribed medication as the standard. The medication recommendation model determined according to this standard is used for medication recommendation, due to the uneven clinical level Inconsistent, clinically prescribed medications have poor therapeutic efficacy, which will reduce the accuracy of the recommendation model of medication recommendation, that is, the effectiveness of the obtained recommended medication is not high, and affect the promotion of the medication recommendation model.

Summary of the invention

The embodiments of the present application provide an artificial intelligence-based medication recommendation method, device, equipment, and storage medium to solve the problem of low recommendation accuracy when the current medication recommendation model performs medication recommendation.

An artificial intelligence-based medication recommendation method, including:

Obtain a medication recommendation request, the medication recommendation request including disease type, current disease data, and user profile data;

Obtaining at least two target drug recommendation models with good target analysis results according to the target analysis result of the original medication recommendation model corresponding to the disease type;

Using at least two of the target medication recommendation models to analyze and process the current disease data to obtain at least two original recommended medications;

Perform feature analysis on the original recommended medication, and obtain feature analysis values corresponding to at least two features to be analyzed;

Determining a recommendation tendency type based on the user portrait data, and obtaining feature weights corresponding to at least two features to be analyzed according to the recommendation tendency type;

Based on the characteristic analysis values and the characteristic weights corresponding to at least two of the characteristics to be analyzed, the recommended evaluation values of at least two of the original recommended medications are obtained, and the original recommended medication with the largest recommended evaluation value is determined Recommend medication for the target.

An artificial intelligence-based medication recommendation device, including:

A recommendation request acquisition module for acquiring a medication recommendation request, the medication recommendation request including disease type, current disease data, and user profile data;

The recommendation model determination module is configured to obtain at least two target drug recommendation models with good target analysis results according to the target analysis result of the original drug recommendation model corresponding to the disease type;

The original recommended medication acquisition module is used to analyze and process the current disease data by using at least two of the target medication recommendation models to obtain at least two original recommended medications;

The characteristic analysis value acquisition module is configured to perform characteristic analysis on the original recommended medication, and acquire characteristic analysis values corresponding to at least two characteristics to be analyzed;

The feature weight obtaining module is configured to determine a recommendation tendency type based on the user portrait data, and obtain the feature weights corresponding to at least two features to be analyzed according to the recommendation tendency type;

The target recommended medication acquisition module is configured to acquire at least two recommended evaluation values of the original recommended medications based on the feature analysis values corresponding to the at least two features to be analyzed and the feature weights, and compare the recommended evaluation values The largest of the original recommended medication is determined to be the target recommended medication.

A computer device includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and the processor implements the aforementioned artificial intelligence-based medication recommendation method when the processor executes the computer program, For example, implement the following steps:

A computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the above-mentioned artificial intelligence-based medication recommendation method is implemented, for example, the following steps are implemented:

The aforementioned artificial intelligence-based medication recommendation method, device, equipment, and storage medium can improve the effectiveness and pertinence of intelligent medication recommendation.

Description of the drawings

FIG. 1 is a schematic diagram of an application environment of an artificial intelligence-based medication recommendation method in an embodiment of the present application;

FIG. 2 is a flowchart of a medication recommendation method based on artificial intelligence in an embodiment of the present application;

FIG. 3 is another flowchart of a method for recommending medication based on artificial intelligence in an embodiment of the present application;

FIG. 4 is another flowchart of a method for recommending medication based on artificial intelligence in an embodiment of the present application;

FIG. 5 is another flowchart of a method for recommending medication based on artificial intelligence in an embodiment of the present application;

FIG. 6 is another flowchart of a method for recommending medication based on artificial intelligence in an embodiment of the present application;

FIG. 7 is another flowchart of the artificial intelligence-based medication recommendation method in an embodiment of the present application;

FIG. 8 is another flowchart of a method for recommending medications based on artificial intelligence in an embodiment of the present application;

Fig. 9 is a schematic diagram of an artificial intelligence-based medication recommendation device in an embodiment of the present application;

Fig. 10 is a schematic diagram of a computer device in an embodiment of the present application.

Detailed ways

The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

The technical solution of this application can be applied to the field of artificial intelligence and/or blockchain technology, and the data involved, such as disease type, disease data, user profile data, etc., can be stored in blockchain nodes, such as distributed storage through blockchain , Etc., this application is not limited.

According to the artificial intelligence-based medication recommendation method provided by the embodiments of the present application, the artificial intelligence-based medication recommendation method can be applied to the application environment as shown in FIG. 1. Specifically, the artificial intelligence-based medication recommendation method is applied in a CDSS system, which includes a client and a server as shown in FIG. 1, and the client and the server communicate through a network to implement medication triggered by the client Recommendation request, intelligent analysis of the current disease data and user profile data, so that the obtained target recommends the medication to ensure the accuracy, efficiency and pertinence of the recommended medication. Among them, the client is also called the client, which refers to the program that corresponds to the server and provides local services to the client. The client can be installed on, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices. The server can be implemented as an independent server or a server cluster composed of multiple servers.

In one embodiment, as shown in FIG. 2, a method for recommending medication based on artificial intelligence is provided. Taking the method applied to the server in FIG. 1 as an example for description, the method includes the following steps:

S201: Obtain a medication recommendation request. The medication recommendation request includes disease type, current disease data, and user profile data.

Among them, the medication recommendation request is a request for triggering the execution of the medication recommendation service. As an example, the server of the CDSS system may receive a medication recommendation request triggered by the user through the client, and obtain the disease type, current illness data, and user profile data in the medication recommendation request.

Among them, the disease type is used to reflect the corresponding type of the patient's disease. As an example, the types of conditions include, but are not limited to, diabetes, hypertension, and chronic kidney disease. Understandably, the determination of the type of disease can help determine a more accurate medication recommendation model for medication recommendation, and improve the accuracy of medication recommendation.

Among them, the current disease data is data corresponding to the current physical state of the patient. The current disease data refers to the data corresponding to the detection index corresponding to the disease type detected in real time by the patient in the medical facility. For example, for diabetes, it is necessary to collect real-time measured data corresponding to indicators such as blood glucose, glycosylated hemoglobin, glycosylated serum protein, and antibodies.

Among them, the user portrait data is the portrait data used to reflect patient-related information. These user portrait data may include data that affects the diagnosis of the disease, such as age, gender, and other data, and may also include data that affects medication recommendations, for example, in patients When income is low, it may be more inclined to use low-cost medication.

S202: According to the target analysis result of the original medication recommendation model corresponding to the disease type, obtain at least two target medication recommendation models with better target analysis results.

Among them, the original medication recommendation model is a medication recommendation model corresponding to the disease type pre-stored in the CDSS system, and is a model that can output medication recommendations based on the disease data. That is, for each disease type, the CDSS system is connected to at least one original medication recommendation model corresponding to the disease type, so that the CDSS system can perform the current disease data corresponding to the disease type based on the pre-accessed original medication recommendation model. Intelligent analysis improves the pertinence and effectiveness of the recommended medications obtained from the analysis. Understandably, different original drug recommendation models corresponding to the same disease type may be data obtained by model training on clinical medical data (including clinical disease data and corresponding clinical drug data) using different neural network models, so that each original The medication recommendation model is based on clinical medicine guidelines to avoid clinical mistakes in the recommended medications. It also needs to use medical big data to achieve personalized medication recommendations and learn medication plans for similar patients to ensure the effectiveness of medication recommendations.

Since the neural network models used by different original medication recommendation models are different, and their training processes are different, when the original medication recommendation model obtained by its training performs intelligent medication recommendation on the same current disease data, there may be different medication recommendation results. Therefore, the CDSS system needs to comprehensively evaluate each original drug recommendation model in advance, that is, comprehensively evaluate the original drug recommendation model based on clinical medical data of similar patients to obtain the target analysis result, and store the target analysis result on the CDSS system , So that the CDSS system can quickly query the target analysis results of the original drug recommendation model according to the type of disease during the drug recommendation process, and sort all the original drug recommendation models in descending order based on the target analysis results, and select at least two of the first target analysis results The original medication recommendation model determines the target medication recommendation model adopted for this medication recommendation. Understandably, the number of target medication recommendation models can be set independently according to actual needs, and at least two original medication recommendation models in the descending order of the target analysis results are determined as target medication recommendation models, so that intelligence is based on the target medication recommendation model. When the medication is recommended, it helps to ensure the validity of the results of the medication recommendation.

S203: Analyze and process current disease data using at least two target medication recommendation models to obtain at least two original recommended medications.

Among them, the original recommended medication is the output result of the target medication recommendation model after analyzing and processing the current disease data. As an example, after acquiring at least two target medication recommendation models, the CDSS system uses at least two target medication recommendation models to analyze and process current disease data in parallel to obtain the original recommended medication output from the at least two target medication recommendation models.

S204: Perform feature analysis on the original recommended medication, and obtain feature analysis values corresponding to at least two features to be analyzed.

Among them, the feature to be analyzed is a pre-configured feature that needs to be analyzed and processed, and the feature to be analyzed can be understood as a feature used to evaluate the quality of a medication recommendation. In this example, the features to be analyzed include but are not limited to features such as medication cost, curative effect time, and side effects.

As an example, after acquiring at least two original recommended medications, the CDSS system uses a preset feature analysis strategy to perform feature analysis on the original recommended medications to obtain feature analysis values corresponding to the at least two features to be analyzed. In this example, the CDSS system adopts a preset feature analysis strategy to perform feature analysis on the original recommended medication. This may include: First, query the system database or related database based on the original recommended medication, and obtain the original feature of the original recommended medication that corresponds to the feature to be analyzed. Data, for example, to obtain the actual cost corresponding to the feature to be analyzed, the cost of medication, that is, the cost of medication required for a course of treatment. Then, normalize the feature raw data corresponding to all the features to be analyzed to convert the feature raw data corresponding to all the features to be analyzed into non-dimensional quantities to ensure the feasibility of the subsequent calculation of recommended evaluation values.

S205: Determine a recommendation tendency type based on the user portrait data, and obtain feature weights corresponding to at least two features to be analyzed according to the recommendation tendency type.

Among them, the recommendation tendency type is a medication recommendation type that is analyzed according to the patient's user profile data and used to reflect the patient's preference. As an example, the medication recommendation type includes, but is not limited to, a low-cost recommendation type with a large cost weight, a high-efficiency recommendation type with a large curative effect time weight, and a low-risk recommendation type with a large side effect weight. In this example, at least two feature weights corresponding to the features to be analyzed are pre-configured for each medication recommendation type, and the corresponding feature weights are stored in the system database.

As an example, the CDSS system uses a preset recommendation tendency analysis strategy to intelligently analyze the user portrait data to determine the recommendation tendency type corresponding to the user portrait data; then according to the recommendation tendency type, obtain the recommendation tendency type and query the system database to obtain Pre-configured feature weight corresponding to each feature to be analyzed.

S206: Obtain recommended evaluation values of at least two original recommended medications based on the feature analysis values and feature weights corresponding to the at least two features to be analyzed, and determine the original recommended medication with the largest recommended evaluation value as the target recommended medication.

Among them, the target recommended medication is a recommended medication plan that is screened from at least two original recommended medications that best matches the user profile data.

In this embodiment, after the CDSS system is based on the feature analysis value and feature weight corresponding to the at least two features to be analyzed, it uses a weighting algorithm to perform a weighting operation on the feature analysis value and feature weight corresponding to the at least two features to be analyzed to obtain at least The recommended evaluation value of two original recommended medications; then at least two recommended evaluation values are compared to determine the maximum recommended evaluation value, and the original recommended medication corresponding to the maximum recommended evaluation value is determined as the target recommended medication. The recommended evaluation value is a value obtained by weighting the feature analysis values and feature weights corresponding to at least two features to be analyzed, which can reflect the degree of fit with the patient’s user profile data. The higher the recommended evaluation value, the original The recommended medication is more suitable for the actual needs of patients, and the pertinence and effectiveness of medication recommendations will be improved.

As an example, the CDSS system may use P=∑S _i W _i to perform a weighted operation on the feature analysis values and feature weights corresponding to at least two features to be analyzed to obtain the recommended evaluation values of at least two original recommended medications. The recommended evaluation values the acquisition process is simple and convenient calculation; where, P is the recommendation evaluation value, S _i is the i-th feature analysis value corresponding to the feature to be analyzed, W _i is the i-th feature to be analyzed corresponding feature weights.

As another example, after obtaining the target analysis result of each original medication recommendation model in step S202, a model weight conversion strategy may be used to process the target analysis result to obtain the model weight corresponding to each target analysis result; then the CDSS system P=Q∑S _i W _i can be used to perform a weighted calculation on the characteristic analysis values and characteristic weights corresponding to at least two features to be analyzed to obtain the recommended evaluation values of at least two original recommended medications, where Q is the corresponding model of the original medication recommendation model weight, P is the recommended evaluation value, S _i is the i-th feature analysis value corresponding to the feature to be analyzed, W _i is the i-th feature to be analyzed corresponding feature weights. Understandably, the target analysis result of the original medication recommendation model is converted into model weights, and then the model weights, feature analysis values and feature weights are used to determine the recommended evaluation value, so that the recommended evaluation value can effectively reflect the effectiveness of the medication recommendation and The fit of user profile data improves the effectiveness and pertinence of intelligent drug recommendation.

In the artificial intelligence-based medication recommendation method provided in this embodiment, according to the target analysis result of the original medication recommendation model corresponding to the disease type, at least two target medication recommendation models with better target analysis results are obtained, and then the target medication recommendation model is used Analyze and process the current disease data to obtain the original recommended medication, which can ensure the effectiveness and pertinence of the original recommended medication; then perform feature analysis on the original recommended medication to determine the feature analysis value corresponding to the feature to be analyzed, and proceed according to the user profile data Analyze the determined recommendation tendency type, determine the feature weight corresponding to the feature to be analyzed, and perform a weighting operation based on the feature analysis value corresponding to the feature to be analyzed and the corresponding feature weight to determine the recommended evaluation value, so that the recommended evaluation value can effectively reflect the medication recommendation The effectiveness and the degree of fit with user profile data improve the effectiveness and pertinence of intelligent drug recommendation.

In one embodiment, as shown in FIG. 3, before obtaining the medication recommendation request, the artificial intelligence-based medication recommendation method further includes the following steps:

S301: Obtain retrospective data corresponding to the disease type, and the retrospective data includes historical disease data and actual medication data.

Among them, retrospective data refers to the data used to evaluate the quality of the original drug recommendation model to determine the target analysis result. As an example, the retrospective data can be obtained from the hospital information system or the system database of the CDSS system, or can be some public medical data sets, which are specific structured data.

In this example, the retrospective data is a visit record, that is, the data determined by the patient at the clinic one time, specifically including historical disease data and actual medication data. The historical disease data refers to the data related to the disease recorded when the patient visits a doctor, including symptoms and test results information. The actual medication data refers to the medication data given by the doctor when visiting a doctor, which specifically includes data such as disease diagnosis results, prescription medication and treatment time. Understandably, since the comprehensive evaluation process of the original medication recommendation model requires the analysis of the treatment effect of the disease symptoms corresponding to the historical disease data in combination with the actual medication data, the retrospective data also includes short-term effect data and long-term effect data . Among them, the short-term effect data refers to the data on the compliance of the inspection index in a short period of time after the current visit. For example, the data on whether the glycosylated hemoglobin meets the standard about 3 months after treatment can be determined as the short-term effect data. Long-term effect data refers to data on whether there are complications or adverse events for a long period of time after this visit. The proportion can be determined based on the follow-up outpatient or inpatient diagnosis results. Understandably, the time limit of short-term effect data and long-term effect data can be determined based on historical experience.

As an example, the CDSS system can query the hospital information system or the system database of the CDSS system or crawl the hospital data set according to the type of disease, and obtain all the historical medical records corresponding to the disease type; then check all the historical medical records to determine Whether the historical medical records include necessary data such as historical disease data, actual medication data, short-term effect data, and long-term effect data; if the historical medical records contain all necessary data, the historical medical records are determined to be used to evaluate the original drug recommendation model Retrospective data to ensure the feasibility of the follow-up evaluation of the original drug recommendation model.

S302: Use the original medication recommendation model corresponding to the type of illness to analyze and process historical illness data, obtain historical recommended medication data, match historical recommended medication data with actual medication data, and obtain matching results, which will be retrospectively based on the matching results The data is divided into a first data set that meets the model recommendation and a second data set that does not meet the model recommendation.

Since the original medication recommendation model is a model that can output medication recommendations based on disease data, the CDSS system can use the original medication recommendation model to analyze and process historical disease data to obtain historical recommended medication data. That is, the historical recommended medication data refers to the medication data output by analyzing and processing historical disease data using the original medication recommendation model.

After obtaining the historical recommended medication data, the CDSS system can match the historical recommended medication data with the actual medication data to determine whether the historical recommended medication data matches the actual medication data, so as to divide the retrospective data into conformance according to the matching results. The first data set recommended by the model and the second data set that do not meet the model recommendation, that is, if the historical recommended medication data matches the actual medication data, the retrospective data is classified into the first data set that meets the model recommendation; If the recommended medication data does not match the actual medication data, the retrospective data is classified into the second data set that does not meet the model recommendation.

S303: Based on the retrospective data in the first data set and the second data set, respectively calculate the short-term effectiveness index compliance rate and the long-term complication rate, and perform a significance check to obtain an effect analysis result.

Among them, the short-term effectiveness index compliance rate is used to evaluate the impact of the short-term effectiveness indicators on whether the short-term effectiveness indicators meet the standards based on retrospective data. The long-term complication rate is used to evaluate whether the model recommendation from the visit level or the patient level is consistent with the impact on the occurrence of long-term complications based on retrospective data. The visit level refers to the record of a patient's visit, which is mainly used to assess whether the short-term effectiveness index meets the standard, such as whether the short-term effectiveness index meets the standard within a short time after each visit. This patient level refers to all medical records of the same patient, and is mainly used to assess whether long-term complications have occurred. As an example, according to the same patient’s multiple visit records, the recommendation compliance rate between the actual medication data and the historical recommended medication data can be determined, and then the patient-level recommendation compliance rate and the long-term complication rate can be further analyzed based on the recommendation compliance rate. relationship. For example, if there are 10 visits for the same patient, there are 10 visit records, among which 6 actual medication data are consistent with historical recommended medication data, that is, the recommendation compliance rate is 60%, and then the recommendation rate is based on the compliance model. Significance check, so as to obtain the results of the effect analysis reflecting the relationship between the model recommendation rate and the long-term complication rate. Significance test is to make a hypothesis on the parameters of the population (random variable) or the distribution form of the population in advance, and then use the sample information to judge whether the hypothesis (alternative hypothesis) is reasonable, that is, to judge the true situation and the original situation of the population. Assume whether there is a significant difference.

As an example, based on the retrospective data in the first data set and the second data set, the CDSS system separately counts the corresponding short-term effectiveness index compliance rate and long-term complication rate, so as to base the statistics on the short-term effectiveness indicator compliance rate Perform a significant check with the incidence of long-term complications to obtain the results of the effect analysis. The results of the effect analysis can effectively reflect the relationship between the short-term effectiveness index compliance rate and the long-term complication rate and the model compliance rate for subsequent synthesis Evaluation.

S304: If the result of the effect analysis meets the preset analysis condition, perform a difference analysis on the retrospective data in the first data set and the second data set to obtain the target short-term confounding factors and the target long-term confounding factors.

Among them, the preset analysis conditions are pre-set conditions that require subsequent difference analysis. As an example, when the effect analysis result is the P-value obtained by the significance check, the preset analysis result can be set to be less than the preset value, such as less than 0.05.

Because the retrospective data in the first data set and the second data set have differences in the short-term effectiveness index compliance rate and the long-term complication rate, it may be caused by therapeutic factors, or it may be caused by non-therapeutic factors. Factors are related to medication, and non-therapeutic factors are not related to medication. In order to avoid non-therapeutic factors from affecting the effectiveness of the model’s recommended medications, it is necessary to perform a differential analysis on the retrospective data in the first data set and the second data set to obtain Target short-term confounding factors and target long-term confounding factors. Confounding factors can be understood as non-therapeutic factors; target short-term confounding factors refer to the top non-therapeutic factors that have the greatest or greater impact on the short-term effectiveness index compliance rate; target long-term confounding factors refer to the impact on the incidence of long-term complications The first few non-therapeutic factors that have the greatest or greater impact.

For example, the retrospective data includes N confounding factors such as F1, F2, F3...Fn. When the difference analysis is performed on the retrospective data in the first data set and the second data set, when analyzing any confounding factor, Calculate the corresponding short-term effectiveness index compliance rate and long-term complication rate respectively to obtain the effect analysis results corresponding to the confounding factors; then according to the effect analysis results, the short-term effectiveness indicator compliance rate will have the greatest or greater impact The top confounding factors are determined as the target short-term confounding factors, and the top confounding factors that have the greatest or greater impact on the incidence of long-term complications are determined as the target short-term confounding factors.

S305: Based on the target short-term confounding factor and the target short-term confounding factor, perform a tendency analysis on the retrospective data in the first data set and the second data set, and obtain and store the target analysis result of the original medication recommendation model.

In this example, based on the target short-term confounding factors and target long-term confounding factors determined by analysis, a tendency analysis is performed on the retrospective data in the first data set and the second data set to weaken the confounding factors’ impact on the short-term effectiveness index compliance rate and The impact of the long-term complication rate makes the target analysis results obtained reasonable. In this example, after the CDSS system obtains the target analysis result of the original medication recommendation model, it needs to store the target analysis result in the system database so that the CDSS system can call the target analysis result of the original medication recommendation model after obtaining the medication recommendation request. Quickly determine at least two target medication recommendation models with good target analysis results, so as to ensure that the effectiveness of the target medication recommendation model that has been clinically verified and learned from similar patient medication regimens can be selected.

In one embodiment, as shown in FIG. 4, matching the historical recommended medication data with the actual medication data in step S302 to obtain the matching result specifically includes the following steps:

S401: Acquire at least one recommended medication category based on historical recommended medication data, and acquire at least one actual medication category based on actual medication data.

S402: If all recommended medication categories match all actual medication categories, the matching result is in compliance with the model recommendation.

S403: If there is at least one recommended medication category that does not match the actual medication category, the matching result is that it does not meet the model recommendation.

In this example, you can determine the recommended medication category based on the historical recommended medication data, and determine the actual medication category based on the actual medication data, and compare whether all recommended medication categories match all actual medication categories; if all recommended medication categories match all actual medication categories Uniform matching, that is, the combination of recommended medication categories matches the combination of actual medication categories, the matching result is in line with the model recommendation; if there is at least one recommended medication category that matches the actual medication category, that is, the combination of recommended medication categories matches the actual medication category. If the combination of medication categories does not match, the matching result is not in compliance with the model recommendation.

For example, diabetes medications can be divided into medication categories such as biguanides, sulfonylureas, etc. If the actual medication used by the patient contains metformin, it is considered to use biguanide medications. Understandably, the actual medication data can be defined as a combination of multiple medication categories, such as biguanides + sulfonylureas, that is, drugs of a single medication category can be used or drugs of multiple medication categories can be used at the same time. The match between the historical recommended medication data and the doctor's actual medication data means that they are completely consistent in the combination of medication categories. If there is a difference in the number of medication categories or the name of the medication category, it is considered a mismatch. Understandably, because different drugs have different effects under the same medication category, and their dosages are inconsistent, there is no need to consider whether the dosages of the medications match. Therefore, the historical recommended medication data can be determined based on whether the medication categories match. Whether the actual medication data matches or not can help simplify the calculation amount of the matching process and improve the processing efficiency.

In one embodiment, the retrospective data further includes short-term effect data and long-term effect data. As shown in Figure 5, the retrospective data in the first data set and the second data set in step S303 are respectively counted on the short-term effectiveness index compliance rate and the long-term complication rate, and the significance check is performed to obtain the effect analysis As a result, it specifically includes the following steps:

S501: Based on the retrospective data in the first data set, obtain the first short-term effectiveness index compliance rate and the first long-term complication rate.

S502: Obtain the second short-term effectiveness index compliance rate and the second long-term complication rate based on the retrospective data in the second data set.

S503: Perform a significant check on the first short-term effectiveness index compliance rate, the first long-term complication rate, the second short-term effectiveness indicator compliance rate, and the second long-term complication rate to obtain the effect analysis results.

Among them, the short-term effectiveness index compliance rate refers to the ratio of the number of samples that meet the short-term effectiveness indicators to the total number of all samples. The incidence of long-term complications refers to the ratio of the number of samples with long-term complications to the total number of all samples.

As an example, suppose the number of retrospective data corresponding to the disease type acquired by the CDSS system is A, the CDSS system can divide the first data set and the second data set according to the matching result of the historical recommended medication data and the actual medication data, Suppose the number of retrospective data in the first data set is A1, and the number of retrospective data in the second data set is A2.

Correspondingly, the CDSS system will judge whether it has reached the standard range corresponding to the short-term effectiveness index based on the short-term effect data in the retrospective data. If the short-term effect data reaches the standard range corresponding to the short-term effectiveness index, the short-term effectiveness index will be determined Up to the standard, update the number of samples that meet the short-term effectiveness index, even if the number of samples that meet the short-term effectiveness index is increased by 1, until all retrospective data are analyzed, the number of samples that obtain the short-term effectiveness index up to the standard is B, of which, the first data The number of samples in the centralized short-term effectiveness index that meets the standard is B1, and the number of samples in the second data set that meets the short-term effectiveness index is B2.

Correspondingly, the CDSS system will determine whether the long-term complication standard is met based on the long-term effect data in the retrospective data. If the long-term effect data meets the standard for long-term complication, then it will determine the occurrence of long-term complications and update the occurrence of long-term complications. Even if the number of samples with long-term complications is increased by 1, until all retrospective data is analyzed, the number of samples with long-term complications is C, and the number of samples with long-term complications in the first data set is C1 , The number of samples with long-term complications in the second data set is C2.

In this example, for the retrospective data in the first data set and the second data set, the short-term effectiveness index compliance rate and the long-term complication rate are respectively counted, and the significance check is performed, including: (1) The CDSS system is based on The retrospective data in the first data set is used to obtain the first short-term effectiveness index compliance rate and the first long-term complication rate. Set the first short-term effectiveness indicator compliance rate to Q1, then Q1=B1/A1; set the first long-term The complication rate is P1, then P1=C1/A1. (2) The CDSS system obtains the second short-term effectiveness index compliance rate and the second long-term complication rate based on the retrospective data in the second data set. Assuming the second short-term effectiveness indicator compliance rate is Q2, then Q2=B2/ A2; assuming the second long-term complication rate is P2, then P2=C2/A2. (3) The CDSS system uses the chi-square test method to make significant results for the first short-term effectiveness index compliance rate Q1, the first long-term complication rate P1, the second short-term effectiveness indicator compliance rate Q2, and the second long-term complication rate P2 Check the performance and obtain the results of the effect analysis.

As shown in Table 1 below, for all the retrospective data that meet the short-term effectiveness indicators, if they meet the model recommendation, they will be marked as “Yes”; if they do not meet the model recommendations, they will be marked as “No”. The statistics are in the first data set and the second data set. The number of samples B1 and B2 in the data set are calculated based on B1 and B2, respectively, to calculate the first short-term effectiveness index compliance rate Q1 and the second short-term effectiveness indicator compliance rate Q2, and then according to the difference between Q1 and Q2, and pass the significance calibration Empirically calculate P-value1. Correspondingly, for retrospective data on the occurrence of long-term complications, if it meets the model recommendation, it is marked as "Yes"; if it does not meet the model recommendation, it is marked as "No", and the number of samples in the first data set and the second data set is counted C1 and C2, and then calculate the first long-term complication rate P1 and the second long-term complication rate P2 based on C1 and C2, and then calculate the P-value2 through the significance check based on the difference between P1 and P2. In this example, whether compliance with the model recommendation is used as the exposure variable, whether the short-term effectiveness index meets the standard and whether long-term complications occur as the outcome variable, the P-value between the exposure variable and the outcome variable is calculated by the chi-square check method, if P-value<0.05 indicates that there is significant between the exposure variable and the outcome variable, and the effect analysis results are obtained to achieve the purpose of analyzing whether the model recommendation is related to the short-term effectiveness index compliance rate and the long-term complication rate.

In one embodiment, as shown in FIG. 6, step S304, which is to perform difference analysis on the retrospective data in the first data set and the second data set, to obtain the target short-term confounding factor and the target long-term confounding factor, specifically includes the following steps:

S601: Determine the confounding factors to be analyzed.

Among them, the confounding factors to be analyzed refer to the confounding factors that need to be analyzed and processed this time. For example, if the retrospective data includes N confounding factors such as F1, F2, F3...Fn, one confounding factor can be randomly selected as the confounding factor to be analyzed each time. The initial glycosylated hemoglobin is determined to be a confounding factor to be analyzed.

S602: Based on the confounding factors to be analyzed, the first analysis subset and the second analysis subset corresponding to the confounding factors to be analyzed are selected from the retrospective data in the first data set and the second data set.

In this example, the CDSS system can divide the confounding factors to be analyzed into multiple first analysis subsets corresponding to the classification criteria according to the preset classification criteria, and divide the second data set into the classification criteria. The multiple second analysis subsets corresponding to the standard, and then according to the specific values of the retrospective data corresponding to the confounding factors that meet the corresponding classification criteria, the retrospective data is divided into the corresponding first analysis subset and the second analysis Subset.

S603: Based on the retrospective data in the first analysis subset and the second analysis subset, respectively calculate the short-term effectiveness index compliance rate and the long-term complication rate, and perform a significance check, and determine the obtained effect analysis result as pending Analyze the results of the confounding analysis corresponding to the confounding factors.

It is understandable that the actual processes of step S603 and S303 are basically the same. For detailed implementation schemes, please refer to steps S501-S503. The difference is that the data sets are different. To avoid repetition, we will not repeat them here.

S604: Obtain the target short-term confounding factor and the target long-term confounding factor according to the confounding analysis result corresponding to the confounding factor to be analyzed.

In this example, after the short-term effectiveness index compliance rate and long-term complication rate statistics and significance check are performed on the retrospective data in the first analysis subset and the second analysis subset, the corresponding effect analysis results are determined For the confounding analysis results corresponding to the confounding factors to be analyzed, based on the results of the confounding analysis, the top confounding factors that have the greatest or greater impact on the short-term effectiveness index compliance rate are identified as the target short-term confounding factors, which will affect long-term complications. The top confounding factors that have the greatest or greater impact on the incidence rate are determined as the target short-term confounding factors. For example, for the glycation compliance rate, the initial glycosylated hemoglobin is the biggest confounding factor, therefore, it is determined as the target short-term confounding factor; for the complication rate, the initial concomitant risk is the biggest confounding factor, therefore, Identify it as a target long-term confounding factor.

In one embodiment, as shown in FIG. 7, step S305, based on the target short-term confounding factor and the target short-term confounding factor, perform a tendency analysis on the retrospective data in the first data set and the second data set, and obtain and store the original The target analysis result of the medication recommendation model includes the following steps:

S701: Based on the target short-term confounding factor and the target short-term confounding factor, stratify the first data set and the second data set to obtain a stratified data set.

S702: Perform a tendency analysis on the retrospective data in the hierarchical data set, and obtain a tendency result corresponding to each hierarchical data set.

S703: Based on the tendency result corresponding to the hierarchical data set, obtain and store the target analysis result corresponding to the original medication recommendation model.

Among them, the hierarchical data set is a hierarchical data set formed by dividing the first data set and the second data set according to the target short-term confounding factors and the target long-term confounding factors. As an example, when the initial glycosylated hemoglobin is the target short-term confounding factor, the initial glycosylated hemoglobin can be divided into three hierarchical data sets: initial compliance, initial excess, and severe excess, and each hierarchical data set meets the model recommendation and non-compliant The model recommends that the short-term effectiveness index compliance rate of the two situations is compared, that is, the treatment group (treat group) and the control group (ie control group) are compared, and other confounding factors are controlled by the propensity score matching method, and each Propensity results in hierarchical data sets. As another example, when the initial concurrency risk is the target long-term confounding factor, the initial concurrency risk can be divided into three hierarchical data sets, low-risk, medium-risk, and high-risk. Comparing the long-term complication rate of the two cases recommended by the model, that is, as the treatment group (treat group) and the control group (ie control group) for comparison, other confounding factors are controlled by propensity score matching method, and each score is obtained Tendency results in layered data sets.

In this example, after the CDSS system obtains the tendency result in each hierarchical data set, it compares the tendency result with the pre-set eligibility evaluation threshold for judging whether the result is qualified, thereby determining the corresponding hierarchical data set Whether it is qualified; then, the ratio of the qualified quantity corresponding to the stratified data set to the total quantity is counted, and the target analysis result is obtained to realize the curative effect of the original drug recommendation model based on the objectively existing short-term effect data and long-term effect data in the retrospective data Effectiveness conducts objective analysis to ensure the objectivity of target analysis results.

In one embodiment, as shown in FIG. 8, step S205, that is, determining the recommendation tendency type based on user portrait data, specifically includes the following steps:

S801: Judge whether the user portrait data contains existing tendency types.

S802: If the user portrait data includes an existing tendency type, determine the existing tendency type as the recommended tendency type.

S803: If the user portrait data does not include the existing tendency type, determine similar groups based on the user portrait data, and determine the common tendency type corresponding to the similar groups as the recommended tendency type; or, obtain related webpage data based on the user portrait data, and compare the related webpages The feature tags of the data are counted, and the recommended tendency type is obtained.

Among them, the existing tendency type refers to the tendency type explicitly indicated in the user portrait data. For example, the input interface of the CDSS system is configured with existing trend types for users to choose, so that the user can configure the corresponding trend types through the input interface.

As an example, the CDSS system uses character matching algorithms or other algorithms to analyze user portrait data to determine whether the user portrait data includes an existing tendency type. If the user portrait data contains an existing tendency type, the existing tendency type is directly added Determined as the recommended tendency type, based on the recommended tendency type to determine the feature weight corresponding to each feature to be analyzed in the original recommended medication, so that the target recommended medication determined based on the final recommended evaluation value is more matched with the user profile data, which helps to improve the target The pertinence and effectiveness of the recommended medication.

As an example, the CDSS system uses character matching algorithms or other algorithms to analyze user portrait data to determine whether the user portrait data includes existing tendency types. If the user portrait data does not contain existing tendency types, it is determined based on the user portrait data For similar groups, the common tendency type corresponding to the similar group is determined as the recommended tendency type. Among them, the similar group is the group closest to or most similar to the user profile data of the patient, and the similar group specifically refers to the group that is the closest or most similar to the user profile data corresponding to the disease corresponding to the disease type. The similar group includes a plurality of similar users who are closest or most similar to the patient's user profile data, and each similar user corresponds to a medication tendency type, and the medication tendency type refers to a certain tendency type of similar users. The common propensity type corresponding to similar groups of people refers to the medication propensity type with the largest proportion determined by the statistics of the medication propensity types of all similar users in all similar groups.

For example, user portrait data includes but is not limited to gender, age, address, address, occupation, consumption habits, and sports data. The CDSS system determines similar groups based on the user profile data, and the process of determining the common tendency type corresponding to the similar group as the recommended tendency type includes: (1) Using the distance algorithm to measure the patient's user profile data and any existing user stored in the system database The distance calculation is performed on the user portrait data of, and the similar distance corresponding to the two user portrait data is obtained. (2) Determine existing users whose similar distances reach a preset distance threshold as similar users, and form similar groups of people based on all similar users. (3) Perform statistical analysis on the medication propensity types of all similar users in the similar population, determine the medication propensity type with the largest proportion as the common propensity type, and determine the common propensity type as the patient's recommended propensity type. The distance algorithm includes but is not limited to the Euclidean distance algorithm. Understandably, since the user portrait data of similar people are the closest or most similar to the user portrait data corresponding to the patient, the tendency types are also relatively similar. Therefore, the common tendency type of the similar groups can be determined as the recommended tendency type of the patient.

As an example, the CDSS system uses character matching algorithms or other algorithms to analyze user portrait data to determine whether the user portrait data includes an existing tendency type. If the user portrait data does not include an existing tendency type, it will be obtained based on the user portrait data Associate webpage data, perform statistics on the feature tags of the associated webpage data, and obtain the type of recommendation tendency. Among them, the associated webpage data refers to the content data in the associated webpage related to disease treatment corresponding to the disease type. The feature tags of the associated webpage data refer to tags that are pre-configured for the content data in each associated webpage and related to the features to be analyzed, for example, tags such as low cost, short curative effect time, and few side effects.

For example, based on the user profile data to obtain the associated web page data, the feature tags of the associated web data are counted, and the process of obtaining the recommendation tendency type includes: (1) Obtain the ID card based on the ID number and/or mobile phone number in the user profile data All historical webpage data visited by the patient corresponding to the phone number and/or mobile phone number. (2) Determine the historical webpage data related to the disease type in the user portrait data among all the historical webpage data as the associated webpage data. (3) Count the number of feature tags in all associated webpage data, and determine the recommendation tendency type according to the number. Understandably, the feature tags in each associated webpage data can be obtained using the Jieba word segmentation tool and the TF-IDF algorithm, that is, the Jieba word segmentation tool can be used to scan, segment, and part-of-speech tagging the text information in the associated webpage data to obtain Word segmentation results; then use the TF-IDF algorithm to extract keywords from the word segmentation results to obtain the feature tags corresponding to the associated webpage data.

It should be understood that the size of the sequence number of each step in the foregoing embodiment does not mean the order of execution, and the execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiment of the present application.

In an embodiment, an artificial intelligence-based medication recommendation device is provided, and the artificial intelligence-based medication recommendation device corresponds to the artificial intelligence-based medication recommendation method in the foregoing embodiment in a one-to-one correspondence. As shown in FIG. 9, the artificial intelligence-based medication recommendation device includes a recommendation request acquisition module 901, a recommendation model determination module 902, an original recommended medication acquisition module 903, a feature analysis value acquisition module 904, a feature weight acquisition module 905, and a target recommended medication Obtaining module 906. The detailed description of each functional module is as follows:

The recommendation request acquisition module 901 is configured to acquire a medication recommendation request, the medication recommendation request including disease type, current disease data, and user profile data.

The recommendation model determination module 902 is configured to obtain at least two target medication recommendation models with good target analysis results according to the target analysis result of the original medication recommendation model corresponding to the disease type.

The original recommended medication acquisition module 903 is configured to analyze and process current disease data using at least two target medication recommendation models to obtain at least two original recommended medications.

The characteristic analysis value acquisition module 904 is configured to perform characteristic analysis on the original recommended medication, and acquire characteristic analysis values corresponding to at least two features to be analyzed.

The feature weight obtaining module 905 is configured to determine the recommendation tendency type based on the user portrait data, and obtain the feature weights corresponding to at least two features to be analyzed according to the recommendation tendency type.

The target recommended medication acquisition module 906 is configured to obtain recommended evaluation values of at least two original recommended medications based on the feature analysis values and feature weights corresponding to the at least two features to be analyzed, and determine the original recommended medication with the largest recommended evaluation value as the target recommendation Medication.

Preferably, the artificial intelligence-based medication recommendation device further includes a retrospective data acquisition module, a data division module, an effect analysis result acquisition module, a confounding factor determination module, and a target analysis result acquisition module.

The retrospective data acquisition module is used to acquire retrospective data corresponding to the disease type. The retrospective data includes historical disease data and actual medication data.

The data division module is used to analyze and process historical disease data using the original medication recommendation model corresponding to the disease type, obtain historical recommended medication data, match historical recommended medication data with actual medication data, and obtain matching results, based on matching The result divides the retrospective data into the first data set that meets the model recommendation and the second data set that does not meet the model recommendation.

The effect analysis result acquisition module is used to calculate the short-term effectiveness index compliance rate and the long-term complication rate of the retrospective data in the first data set and the second data set, and perform a significance check to obtain the effect analysis results.

The confounding factor determination module is used to perform difference analysis on the retrospective data in the first data set and the second data set if the result of the effect analysis meets the preset analysis conditions to obtain the target short-term confounding factors and the target long-term confounding factors.

The target analysis result acquisition module is used to perform tendency analysis on the retrospective data in the first data set and the second data set based on the target short-term confounding factors and the target short-term confounding factors, and obtain and store the target analysis results of the original medication recommendation model.

Preferably, the data division module includes a medication category acquisition unit, a first matching result acquisition unit, and a second matching result acquisition unit.

The medication category obtaining unit is configured to obtain at least one recommended medication category based on historical recommended medication data, and obtain at least one actual medication category based on actual medication data.

The first matching result obtaining unit is used for if all recommended medication categories match all actual medication categories, then the matching result is in compliance with the model recommendation.

The second matching result obtaining unit is configured to, if at least one recommended medication category does not match the actual medication category, the matching result is not in compliance with the model recommendation.

Preferably, the retrospective data also includes short-term effect data and long-term effect data.

The effect analysis result acquisition module includes a first index acquisition unit, a second index acquisition unit, and a significance verification unit.

The first indicator obtaining unit is configured to obtain the first short-term effectiveness indicator compliance rate and the first long-term complication rate based on the retrospective data in the first data set.

The second index acquisition unit is used to acquire the second short-term effectiveness index compliance rate and the second long-term complication rate based on the retrospective data in the second data set.

Significance verification unit, used to perform significant verification on the first short-term effectiveness index compliance rate, the first long-term complication rate, the second short-term effectiveness indicator compliance rate, and the second long-term complication rate to obtain the effect Analyze the results.

Preferably, the confounding factor determination module includes a factor to be analyzed determining unit, an analysis subset dividing unit, a confounding analysis result obtaining unit, and a target confounding factor obtaining unit.

The factor to be analyzed determination unit is used to determine the confounding factor to be analyzed.

The analysis subset dividing unit is used to filter out the first analysis subset and the second analysis sub-set corresponding to the confounding factors to be analyzed from the retrospective data in the first data set and the second data set based on the confounding factors to be analyzed set.

The confounding analysis result acquisition unit is used to calculate the short-term effectiveness index compliance rate and the long-term complication rate of the retrospective data in the first analysis subset and the second analysis subset, and perform a significance check. The result of the effect analysis is determined as the result of the confounding analysis corresponding to the confounding factor to be analyzed.

The target confounding factor obtaining unit is used to obtain the target short-term confounding factor and the target long-term confounding factor according to the confounding analysis result corresponding to the confounding factor to be analyzed.

Preferably, the target analysis result acquisition module includes a hierarchical data set acquisition unit, a tendency result acquisition unit, and a target analysis result acquisition unit.

The hierarchical data set acquisition unit is used to layer the first data set and the second data set based on the target short-term confounding factor and the target short-term confounding factor to obtain a hierarchical data set.

The tendency result obtaining unit is used to perform tendency analysis on the retrospective data in the hierarchical data set, and obtain the tendency result corresponding to each hierarchical data set.

The target analysis result obtaining unit is used to obtain and store the target analysis result corresponding to the original medication recommendation model based on the tendency result corresponding to the hierarchical data set.

Preferably, the feature weight acquisition module 905 includes an existing tendency judgment unit, a first tendency type determination unit, and a second tendency type determination unit.

The existing tendency judging unit is used to judge whether the user portrait data contains the existing tendency type.

The first tendency type determining unit is configured to determine the existing tendency type as the recommended tendency type if the user portrait data includes the existing tendency type.

The second tendency type determining unit is used to determine similar groups based on the user portrait data if the user portrait data does not include the existing tendency types, and determine the common tendency type corresponding to the similar groups as the recommended tendency type; or, obtain based on the user portrait data Associate webpage data, perform statistics on the feature tags of the associated webpage data, and obtain the type of recommendation tendency.

Regarding the specific limitations of the artificial intelligence-based medication recommendation device, please refer to the above limitations on the artificial intelligence-based medication recommendation method, which will not be repeated here. Each module in the above artificial intelligence-based medication recommendation device can be implemented in whole or in part by software, hardware, and a combination thereof. The above-mentioned modules may be embedded in the form of hardware or independent of the processor in the computer equipment, or may be stored in the memory of the computer equipment in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. 10. The computer equipment includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide calculation and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer equipment is used to store the data adopted or generated during the execution of the artificial intelligence-based medication recommendation method. The network interface of the computer device is used to communicate with an external terminal through a network connection. The computer program is executed by the processor to realize an artificial intelligence-based medication recommendation method.

In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, the artificial intelligence-based medication in the above embodiment is implemented. Recommended methods, such as S201-S206 shown in Figure 2, or shown in Figures 3 to 8, are not repeated here to avoid repetition. Or, when the processor executes the computer program, the function of each module/unit in this embodiment of the artificial intelligence-based medication recommendation device is realized, such as the function of each module/unit shown in FIG. 9. To avoid repetition, it will not be repeated here. .

In one embodiment, a computer-readable storage medium is provided, and a computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, the artificial intelligence-based medication recommendation method in the above embodiment is implemented, as shown in FIG. 2 As shown in S201-S206, or shown in Figs. 3 to 8, in order to avoid repetition, details are not repeated here. Or, when the computer program is executed by the processor, the function of each module/unit in the embodiment of the above-mentioned artificial intelligence-based medication recommendation device, such as the function of each module/unit shown in FIG. 9, is implemented. To avoid repetition, here No longer.

Optionally, the foregoing storage medium, 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 computer program can be stored in a non-volatile computer readable storage. In the medium, when the computer program is executed, it may include the processes of the above-mentioned method embodiments. Wherein, any reference to memory, storage, database, or other media used in the embodiments provided in this application may include non-volatile and/or volatile memory. Non-volatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not a limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Channel (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Those skilled in the art can clearly understand that, for the convenience and conciseness of description, only the division of the above functional units and modules is used as an example. In practical applications, the above functions can be allocated to different functional units and modules as needed. Module completion, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

The above-mentioned embodiments are only used to illustrate the technical solutions of the present application, not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still implement the foregoing The technical solutions recorded in the examples are modified, or some of the technical features are equivalently replaced; these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the application, and should be included in Within the scope of protection of this application.

Claims

An artificial intelligence-based medication recommendation method, which includes:

Obtain a medication recommendation request, the medication recommendation request including disease type, current disease data, and user profile data;

Obtaining at least two target drug recommendation models with good target analysis results according to the target analysis result of the original medication recommendation model corresponding to the disease type;

Using at least two of the target medication recommendation models to analyze and process the current disease data to obtain at least two original recommended medications;

Perform feature analysis on the original recommended medication, and obtain feature analysis values corresponding to at least two features to be analyzed;

Determining a recommendation tendency type based on the user portrait data, and obtaining feature weights corresponding to at least two features to be analyzed according to the recommendation tendency type;

Based on the characteristic analysis values and the characteristic weights corresponding to at least two of the characteristics to be analyzed, the recommended evaluation values of at least two of the original recommended medications are obtained, and the original recommended medication with the largest recommended evaluation value is determined Recommend medication for the target.
The artificial intelligence-based medication recommendation method according to claim 1, wherein, before the obtaining the medication recommendation request, the artificial intelligence-based medication recommendation method further comprises:

Obtaining retrospective data corresponding to the disease type, the retrospective data including historical disease data and actual medication data;

The original medication recommendation model corresponding to the disease type is used to analyze and process the historical illness data, obtain historical recommended medication data, match the historical recommended medication data with the actual medication data, and obtain a matching result, Dividing the retrospective data into a first data set that meets the model recommendation and a second data set that does not meet the model recommendation based on the matching result;

For the retrospective data in the first data set and the second data set, respectively calculate the short-term effectiveness index compliance rate and the long-term complication rate, and perform a significance check to obtain the effect analysis result;

If the result of the effect analysis meets the preset analysis condition, perform a difference analysis on the retrospective data in the first data set and the second data set to obtain target short-term confounding factors and target long-term confounding factors;

Based on the target short-term confounding factor and the target short-term confounding factor, perform a tendency analysis on the retrospective data in the first data set and the second data set, and obtain and store the target analysis of the original medication recommendation model result.
The artificial intelligence-based medication recommendation method according to claim 2, wherein said matching the historical recommended medication data with the actual medication data to obtain a matching result comprises:

Acquiring at least one recommended medication category based on the historical recommended medication data, and acquiring at least one actual medication category based on the actual medication data;

If all the recommended medication categories match all the actual medication categories, the matching result is in compliance with the model recommendation;

If there is at least one mismatch between the recommended medication category and the actual medication category, the matching result is that it does not meet the model recommendation.
The artificial intelligence-based medication recommendation method according to claim 2, wherein the retrospective data further includes short-term effect data and long-term effect data;

According to the retrospective data in the first data set and the second data set, the short-term effectiveness index compliance rate and the long-term complication rate are respectively counted, and the significance check is performed to obtain the effect analysis result, including:

Based on the retrospective data in the first data set, obtaining the first short-term effectiveness index compliance rate and the first long-term complication rate;

Based on the retrospective data in the second data set, obtaining the second short-term effectiveness index compliance rate and the second long-term complication rate;

Perform a significant check on the first short-term effectiveness index compliance rate, the first long-term complication rate, the second short-term effectiveness indicator compliance rate, and the second long-term complication rate to obtain the effect Analyze the results.
The artificial intelligence-based medication recommendation method according to claim 2, wherein the difference analysis is performed on the retrospective data in the first data set and the second data set to obtain target short-term confounding factors and target long-term confounding Factors include:

Determine the confounding factors to be analyzed;

Based on the confounding factors to be analyzed, from the retrospective data in the first data set and the second data set, the first analysis subset and the second analysis sub-set corresponding to the confounding factors to be analyzed are filtered out set;

For the retrospective data in the first analysis subset and the second analysis subset, the short-term effectiveness index compliance rate and the long-term complication rate are respectively counted, and the significance check is performed, and the obtained effect analysis results are determined Is the confounding analysis result corresponding to the confounding factor to be analyzed;

According to the confounding analysis result corresponding to the confounding factor to be analyzed, the target short-term confounding factor and the target long-term confounding factor are obtained.
The artificial intelligence-based medication recommendation method according to claim 2, wherein the review of the first data set and the second data set is based on the target short-term confounding factor and the target short-term confounding factor Propensity analysis is performed on sexual data, and the target analysis results of the original drug recommendation model are obtained and stored, including:

Stratifying the first data set and the second data set based on the target short-term confounding factor and the target short-term confounding factor to obtain a hierarchical data set;

Performing a tendency analysis on the retrospective data in the hierarchical data set, and obtaining a tendency result corresponding to each of the hierarchical data sets;

Based on the tendency result corresponding to the hierarchical data set, the target analysis result corresponding to the original medication recommendation model is acquired and stored.
The artificial intelligence-based medication recommendation method according to claim 1, wherein the determining the recommendation tendency type based on the user portrait data comprises:

Judging whether the user portrait data includes an existing tendency type;

If the user portrait data includes an existing tendency type, the existing tendency type is determined as the recommended tendency type;

If the user portrait data does not contain the existing tendency type, determine similar groups based on the user portrait data, and determine the common tendency type corresponding to the similar group as the recommended tendency type; or, obtain the association based on the user portrait data The webpage data is to perform statistics on the feature tags of the associated webpage data to obtain the recommendation tendency type.
An artificial intelligence-based medication recommendation device, which includes:

A recommendation request acquisition module for acquiring a medication recommendation request, the medication recommendation request including disease type, current disease data, and user profile data;

The recommendation model determination module is configured to obtain at least two target drug recommendation models with good target analysis results according to the target analysis result of the original drug recommendation model corresponding to the disease type;

The original recommended medication acquisition module is used to analyze and process the current disease data by using at least two of the target medication recommendation models to obtain at least two original recommended medications;

The characteristic analysis value acquisition module is configured to perform characteristic analysis on the original recommended medication, and acquire characteristic analysis values corresponding to at least two characteristics to be analyzed;

The feature weight obtaining module is configured to determine a recommendation tendency type based on the user portrait data, and obtain the feature weights corresponding to at least two features to be analyzed according to the recommendation tendency type;

The target recommended medication acquisition module is configured to acquire at least two recommended evaluation values of the original recommended medications based on the feature analysis values corresponding to the at least two features to be analyzed and the feature weights, and compare the recommended evaluation values The largest of the original recommended medication is determined to be the target recommended medication.
A computer device includes a memory, a processor, and a computer program that is stored in the memory and can run on the processor, wherein the processor implements the following steps when the processor executes the computer program:

Obtain a medication recommendation request, the medication recommendation request including disease type, current disease data, and user profile data;

Obtaining at least two target drug recommendation models with good target analysis results according to the target analysis result of the original medication recommendation model corresponding to the disease type;

Using at least two of the target medication recommendation models to analyze and process the current disease data to obtain at least two original recommended medications;

Perform feature analysis on the original recommended medication, and obtain feature analysis values corresponding to at least two features to be analyzed;

Determining a recommendation tendency type based on the user portrait data, and obtaining feature weights corresponding to at least two features to be analyzed according to the recommendation tendency type;

Based on the characteristic analysis values and the characteristic weights corresponding to at least two of the characteristics to be analyzed, the recommended evaluation values of at least two of the original recommended medications are obtained, and the original recommended medication with the largest recommended evaluation value is determined Recommend medication for the target.
The computer device according to claim 9, wherein, before the obtaining the medication recommendation request, the processor further implements the following steps when executing the computer program:

Obtaining retrospective data corresponding to the disease type, the retrospective data including historical disease data and actual medication data;

The original medication recommendation model corresponding to the disease type is used to analyze and process the historical illness data, obtain historical recommended medication data, match the historical recommended medication data with the actual medication data, and obtain a matching result, Dividing the retrospective data into a first data set that meets the model recommendation and a second data set that does not meet the model recommendation based on the matching result;

For the retrospective data in the first data set and the second data set, respectively calculate the short-term effectiveness index compliance rate and the long-term complication rate, and perform a significance check to obtain the effect analysis result;

If the result of the effect analysis meets the preset analysis condition, perform a difference analysis on the retrospective data in the first data set and the second data set to obtain target short-term confounding factors and target long-term confounding factors;

Based on the target short-term confounding factor and the target short-term confounding factor, perform a tendency analysis on the retrospective data in the first data set and the second data set, and obtain and store the target analysis of the original medication recommendation model result.
10. The computer device according to claim 10, wherein the matching process is performed on the historical recommended medication data and the actual medication data, and when the matching result is obtained, the following steps are specifically implemented:

Acquiring at least one recommended medication category based on the historical recommended medication data, and acquiring at least one actual medication category based on the actual medication data;

If all the recommended medication categories match all the actual medication categories, the matching result is in compliance with the model recommendation;

If there is at least one mismatch between the recommended medication category and the actual medication category, the matching result is that it does not meet the model recommendation.
10. The computer device of claim 10, wherein the retrospective data further includes short-term effect data and long-term effect data;

According to the retrospective data in the first data set and the second data set, the short-term effectiveness index compliance rate and the long-term complication rate are respectively counted, and the significance check is performed. When the effect analysis result is obtained, the specific Implement the following steps:

Based on the retrospective data in the first data set, obtaining the first short-term effectiveness index compliance rate and the first long-term complication rate;

Based on the retrospective data in the second data set, obtaining the second short-term effectiveness index compliance rate and the second long-term complication rate;

Perform a significant check on the first short-term effectiveness index compliance rate, the first long-term complication rate, the second short-term effectiveness indicator compliance rate, and the second long-term complication rate to obtain the effect Analyze the results.
The computer device according to claim 10, wherein when the difference analysis is performed on the retrospective data in the first data set and the second data set to obtain the target short-term confounding factors and the target long-term confounding factors, the specific realization The following steps:

Determine the confounding factors to be analyzed;

Based on the confounding factors to be analyzed, from the retrospective data in the first data set and the second data set, the first analysis subset and the second analysis sub-set corresponding to the confounding factors to be analyzed are filtered out set;

For the retrospective data in the first analysis subset and the second analysis subset, the short-term effectiveness index compliance rate and the long-term complication rate are respectively counted, and the significance check is performed, and the obtained effect analysis results are determined Is the confounding analysis result corresponding to the confounding factor to be analyzed;

According to the confounding analysis result corresponding to the confounding factor to be analyzed, the target short-term confounding factor and the target long-term confounding factor are obtained.
The computer device according to claim 10, wherein the retrospective data in the first data set and the second data set are biased based on the target short-term confounding factor and the target short-term confounding factor When analyzing, acquiring and storing the target analysis result of the original medication recommendation model, the following steps are specifically implemented:

Stratifying the first data set and the second data set based on the target short-term confounding factor and the target short-term confounding factor to obtain a hierarchical data set;

Performing a tendency analysis on the retrospective data in the hierarchical data set, and obtaining a tendency result corresponding to each of the hierarchical data sets;

Based on the tendency result corresponding to the hierarchical data set, the target analysis result corresponding to the original medication recommendation model is acquired and stored.
9. The computer device according to claim 9, wherein when the recommendation tendency type is determined based on the user portrait data, the following steps are specifically implemented:

Judging whether the user portrait data includes an existing tendency type;

If the user portrait data includes an existing tendency type, the existing tendency type is determined as the recommended tendency type;

If the user portrait data does not contain the existing tendency type, determine similar groups based on the user portrait data, and determine the common tendency type corresponding to the similar group as the recommended tendency type; or, obtain the association based on the user portrait data The webpage data is to perform statistics on the feature tags of the associated webpage data to obtain the recommendation tendency type.
A computer-readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the following steps:

Obtain a medication recommendation request, the medication recommendation request including disease type, current disease data, and user profile data;

Obtaining at least two target drug recommendation models with good target analysis results according to the target analysis result of the original medication recommendation model corresponding to the disease type;

Using at least two of the target medication recommendation models to analyze and process the current disease data to obtain at least two original recommended medications;

Perform feature analysis on the original recommended medication, and obtain feature analysis values corresponding to at least two features to be analyzed;

Determining a recommendation tendency type based on the user portrait data, and obtaining feature weights corresponding to at least two features to be analyzed according to the recommendation tendency type;

Based on the characteristic analysis values and the characteristic weights corresponding to at least two of the characteristics to be analyzed, the recommended evaluation values of at least two of the original recommended medications are obtained, and the original recommended medication with the largest recommended evaluation value is determined Recommend medication for the target.
15. The computer-readable storage medium of claim 16, wherein, before the obtaining the medication recommendation request, the computer program further implements the following steps when being executed by the processor:

Obtaining retrospective data corresponding to the disease type, the retrospective data including historical disease data and actual medication data;

The original medication recommendation model corresponding to the disease type is used to analyze and process the historical illness data, obtain historical recommended medication data, match the historical recommended medication data with the actual medication data, and obtain a matching result, Dividing the retrospective data into a first data set that meets the model recommendation and a second data set that does not meet the model recommendation based on the matching result;

For the retrospective data in the first data set and the second data set, respectively calculate the short-term effectiveness index compliance rate and the long-term complication rate, and perform a significance check to obtain the effect analysis result;

If the result of the effect analysis meets the preset analysis condition, perform a difference analysis on the retrospective data in the first data set and the second data set to obtain target short-term confounding factors and target long-term confounding factors;

Based on the target short-term confounding factor and the target short-term confounding factor, perform a tendency analysis on the retrospective data in the first data set and the second data set, and obtain and store the target analysis of the original medication recommendation model result.
17. The computer-readable storage medium of claim 17, wherein the retrospective data further includes short-term effect data and long-term effect data;

According to the retrospective data in the first data set and the second data set, the short-term effectiveness index compliance rate and the long-term complication rate are respectively counted, and the significance check is performed. When the effect analysis result is obtained, the specific Implement the following steps:

Based on the retrospective data in the first data set, obtaining the first short-term effectiveness index compliance rate and the first long-term complication rate;

Based on the retrospective data in the second data set, obtaining the second short-term effectiveness index compliance rate and the second long-term complication rate;

Perform a significant check on the first short-term effectiveness index compliance rate, the first long-term complication rate, the second short-term effectiveness indicator compliance rate, and the second long-term complication rate to obtain the effect Analyze the results.
17. The computer-readable storage medium of claim 17, wherein the retrospective data in the first data set and the second data set are analyzed based on the target short-term confounding factor and the target short-term confounding factor. When performing propensity analysis and obtaining and storing the target analysis result of the original drug recommendation model, the following steps are specifically implemented:

Stratifying the first data set and the second data set based on the target short-term confounding factor and the target short-term confounding factor to obtain a hierarchical data set;

Performing a tendency analysis on the retrospective data in the hierarchical data set, and obtaining a tendency result corresponding to each of the hierarchical data sets;

Based on the tendency result corresponding to the hierarchical data set, the target analysis result corresponding to the original medication recommendation model is acquired and stored.
16. The computer-readable storage medium according to claim 16, wherein when the recommendation tendency type is determined based on the user portrait data, the following steps are specifically implemented:

Judging whether the user portrait data includes an existing tendency type;

If the user portrait data includes an existing tendency type, the existing tendency type is determined as the recommended tendency type;

If the user portrait data does not contain the existing tendency type, determine similar groups based on the user portrait data, and determine the common tendency type corresponding to the similar group as the recommended tendency type; or, obtain the association based on the user portrait data The webpage data is to perform statistics on the feature tags of the associated webpage data to obtain the recommendation tendency type.