Interrogation user loss prediction method and device, electronic equipment and storage medium
By obtaining user characteristic indicators and disease labels, combining the churn prediction model, and determining the target churn prediction results, the problem of low accuracy of the existing online consultation churn prediction model is solved, and higher churn prediction accuracy is achieved.
Patent Information
- Application Number
- CN202411009558.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-07-25
AI Technical Summary
The existing online consultation churn prediction model has low accuracy and is difficult to effectively prevent user churn.
By obtaining user characteristic indicators and disease labels, combining the churn prediction model, the user's target churn prediction results are determined, and the accuracy of churn prediction is improved.
It improves the accuracy of the prediction of churn by consultation users, and can more accurately determine whether users will be lost, thereby taking corresponding measures to reduce user churn.
Smart Images

Figure CN119007952A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of digital medicine, and in particular to a method, device, electronic device and storage medium for predicting user churn during consultation. Background Art
[0002] With the rise of Internet medical technology, online medical consultation has gradually become one of the main ways for people to see a doctor. In the competition among various corporate Internet platforms, how to improve user loyalty to their platforms and prevent user churn has become a crucial link in the online medical consultation business.
[0003] In the related art, the user churn of the platform is mostly predicted by a churn prediction model, and a prediction result is directly outputted through the churn prediction model. However, the accuracy of directly predicting user churn through the churn prediction model is currently low. Summary of the invention
[0004] The main purpose of the embodiments of the present application is to propose a method, device, electronic device and storage medium for predicting the churn of consultation users, aiming to improve the accuracy of churn prediction for consultation users.
[0005] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application proposes a method for predicting user churn in a medical consultation, the method comprising:
[0006] Acquire a user characteristic index, where the user characteristic index is used to indicate the user's medical consultation behavior;
[0007] Inputting the user characteristic index into a churn prediction model to obtain a first churn prediction result;
[0008] Acquire a disease label of the user, where the disease label is used to indicate a need for a follow-up visit of the user;
[0009] A target churn prediction result for the user is determined according to the disease label and the first churn prediction result.
[0010] In some embodiments, the disease tag indicates that the user needs to have a follow-up visit;
[0011] The step of determining the target churn prediction result of the user according to the disease label and the first churn prediction result includes:
[0012] Acquire the user's follow-up visit behavior data, where the follow-up visit behavior data is used to indicate whether the user has had a follow-up visit within the follow-up visit period;
[0013] In a case where the user has a follow-up consultation within the follow-up consultation period, using the first churn prediction result as the target churn prediction result for the user;
[0014] In a case where the user does not have a follow-up visit within the follow-up visit period, a target churn prediction result for the user is determined based on the first churn prediction result.
[0015] In some embodiments, determining a target churn prediction result of the user according to the first churn prediction result includes:
[0016] In a case where the first churn prediction result predicts churn of the user, using the first churn prediction result as a target churn prediction result for the user;
[0017] In the case that the first churn prediction result predicts that the user has not churned, the first churn prediction result is corrected to obtain a second churn prediction result, and the second churn prediction result is used as a target churn prediction result.
[0018] In some embodiments, the disease label indicates that the user does not need a follow-up visit;
[0019] The step of determining the target churn prediction result of the user according to the disease label and the first churn prediction result includes:
[0020] The first churn prediction result is used as the target churn prediction result.
[0021] In some embodiments, obtaining the disease label of the user includes:
[0022] Obtain a follow-up visit prediction model based on big data learning;
[0023] Acquiring the diagnosis and treatment data of the user, where the diagnosis and treatment data is used to indicate the disease information of the user;
[0024] The diagnosis and treatment data of the user is input into the follow-up visit prediction model to obtain the disease label of the user.
[0025] In some embodiments, obtaining the user characteristic index includes:
[0026] Get user ID data;
[0027] Obtain the user consultation behavior data corresponding to each user ID according to the user ID data;
[0028] The user consultation behavior data corresponding to each user ID is used as a user characteristic indicator.
[0029] In some embodiments, the churn prediction model is a decision tree model;
[0030] Before inputting the user characteristic index into the churn prediction model to obtain the first churn prediction result, the method further includes:
[0031] Obtain a user history medical consultation data set including multiple user history medical consultation features;
[0032] Calculate the information gain rate of each of the user's historical medical consultation features, perform feature selection according to the information gain rate, and obtain a feature selection result;
[0033] A decision tree model is constructed based on the feature selection results.
[0034] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application proposes a device for predicting user churn in a consultation, the device comprising:
[0035] A user feature acquisition module, used to acquire user feature indicators, wherein the user feature indicators are used to indicate the user's medical consultation behavior;
[0036] A first churn prediction module, used to perform churn prediction processing according to the user characteristic index to obtain a first churn prediction result;
[0037] A disease label acquisition module, used to acquire the disease label of the user, wherein the disease label is used to indicate the user's need for a follow-up visit;
[0038] The second churn prediction module is used to determine a target churn prediction result of the user according to the disease label and the first churn prediction result.
[0039] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, which includes a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory. When the program is executed by the processor, the steps of the method for predicting user churn after consultation are implemented as described above.
[0040] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a storage medium, which is a computer-readable storage medium used for computer-readable storage, and the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the above-mentioned consultation user churn prediction method.
[0041] The method, device, electronic device and storage medium proposed in the present application for predicting the churn of consultation users obtain user characteristic indicators, which are used to indicate the consultation behavior of the user; input the user characteristic indicators into the churn prediction model to obtain a first churn prediction result; obtain the disease label of the user, which is used to indicate the user's need for follow-up consultation; and determine the target churn prediction result of the user based on the disease label and the first churn prediction result. The present application combines the disease label and the first churn prediction result to predict the churn probability of the user, which can improve the accuracy of churn prediction for consultation users. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] Figure 1 is a flow chart of a method for predicting user churn in consultation provided by an embodiment of the present application;
[0043] Figure 2 yes Figure 1 Flow chart of step S104;
[0044] Figure 3 yes Figure 2 Step S203 flow chart;
[0045] Figure 4 When the user does not need a follow-up visit Figure 1 Flow chart of step S104;
[0046] Figure 5 yes Figure 1 Flow chart of step S103;
[0047] Figure 6 yes Figure 1 Flow chart of step S101 in FIG.
[0048] Figure 7 is a flowchart of constructing a decision tree model provided by an embodiment of the present application;
[0049] Figure 8 It is a structural schematic diagram of a device for predicting user churn in consultation provided in an embodiment of the present application;
[0050] Fig. 9 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION
[0051] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0052] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.
[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.
[0054] In addition, the described features, structures or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known methods, devices, implementations or operations are not shown or described in detail to avoid blurring the various aspects of the present disclosure.
[0055] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.
[0056] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.
[0057] First, some nouns involved in this application are analyzed:
[0058] Artificial Intelligence (AI) is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. AI is a branch of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Research in this field includes robots, language recognition, image recognition, natural language processing and expert systems. AI can simulate the information process of human consciousness and thinking. AI is also a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.
[0059] Natural language processing (NLP): NLP uses computers to process, understand and apply human languages (such as Chinese and English). NLP is a branch of artificial intelligence and an interdisciplinary subject between computer science and linguistics. It is often referred to as computational linguistics. Natural language processing includes grammatical analysis, semantic analysis, and text understanding. Natural language processing is often used in technical fields such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information intent recognition, information extraction and filtering, text classification and clustering, public opinion analysis and opinion mining. It involves data mining related to language processing, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computing.
[0060] Information Extraction: A text processing technology that extracts specified types of entity, relationship, event and other factual information from natural language text and forms structured data output. Information extraction is a technology that extracts specific information from text data. Text data is composed of some specific units, such as sentences, paragraphs, and chapters. Text information is composed of some small specific units, such as characters, words, phrases, sentences, paragraphs, or a combination of these specific units. Extracting noun phrases, names, place names, etc. from text data are all text information extraction. Of course, the information extracted by text information extraction technology can be various types of information.
[0061] Churn Prediction Model: is a data analysis tool that is mainly used to predict the possibility of customers or users leaving or stopping using a product or service in the future. In many industries, such as telecommunications, banking, subscription services, e-commerce, etc., customer churn is one of the key factors affecting corporate revenue and growth. By predicting which customers are likely to churn, companies can take targeted measures to reduce churn rates and increase customer retention rates, thereby optimizing marketing strategies and improving customer satisfaction.
[0062] With the rise of Internet medical technology, online medical consultation has gradually become one of the main ways for people to see a doctor. In the competition among various corporate Internet platforms, how to improve user loyalty to their platforms and prevent user churn has become a crucial link in the online medical consultation business.
[0063] In the related art, the user churn of the platform is mostly predicted by a churn prediction model, and a prediction result is directly outputted through the churn prediction model. However, the accuracy of directly predicting user churn through the churn prediction model is currently low.
[0064] Based on this, the embodiments of the present application provide a method, device, electronic device and storage medium for predicting user churn during consultation, aiming to improve the accuracy of user churn prediction.
[0065] The method, device, electronic device and storage medium for predicting user churn in consultation provided in the embodiments of the present application are specifically described through the following embodiments. First, the method for predicting user churn in consultation in the embodiments of the present application is described.
[0066] The method, device, electronic device and storage medium proposed in the present application for predicting the churn of consultation users obtain user characteristic indicators, which are used to indicate the consultation behavior of the user; input the user characteristic indicators into the churn prediction model to obtain a first churn prediction result; obtain the disease label of the user, which is used to indicate the user's need for follow-up consultation; and determine the target churn prediction result of the user based on the disease label and the first churn prediction result. The present application combines the disease label and the first churn prediction result to predict the churn probability of the user, which can improve the accuracy of churn prediction for consultation users.
[0067] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.
[0068] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.
[0069] The method for predicting the loss of medical consultation users provided in the embodiment of the present application relates to the field of artificial intelligence technology. The method for predicting the loss of medical consultation users provided in the embodiment of the present application can be applied to a terminal, can be applied to a server side, or can be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the method for predicting the loss of medical consultation users, etc., but is not limited to the above forms.
[0070] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.
[0071] Figure 1 is an optional flowchart of the method for predicting user churn provided in the embodiment of the present application. Figure 1 The method may include but is not limited to steps S101 to S104.
[0072] Step S101, obtaining a user characteristic index, where the user characteristic index is used to indicate the user's medical consultation behavior;
[0073] Step S102: Input the user characteristic index into the churn prediction model to obtain a first churn prediction result;
[0074] Step S103: Obtain the user's disease label, where the disease label is used to indicate the user's need for follow-up consultation;
[0075] Step S104: Determine a target churn prediction result for the user according to the disease label and the first churn prediction result.
[0076] Steps S101 to S104 shown in the embodiment of the present application can improve the accuracy of churn prediction for consulting users by obtaining user feature indicators, inputting the user feature indicators into a churn prediction model, obtaining a first churn prediction result, obtaining the user's disease label, and determining the user's target churn prediction result based on the disease label and the first churn prediction result.
[0077] In step S101 of some embodiments, the user characteristic index is first explained. The user characteristic index is a series of data points used in user churn prediction (Churn Prediction) to analyze and predict whether a user will leave a product, service or platform. In the embodiment of the present application, the user's consultation behavior is used as a user characteristic index, that is, the user's various consultation behaviors are used as data points for churn prediction to facilitate subsequent churn prediction. Exemplarily, the user characteristic indicators in the embodiment of the present application include but are not limited to the number of consultations of the user, the consultation time period, the major category of the consultation department, the proposed diagnosis label, whether a prescription is issued, whether payment is made, the classification of drugs, the major product category to which non-drug goods belong, the average customer price, etc.
[0078] It should be noted that the various types of medical consultation behavior data included in the user characteristic indicators can be obtained through the user's medical data, through the user's natural language text, and can also be obtained through screening through artificial intelligence models.
[0079] In one possible implementation, the medical data may be personal health records, prescriptions, examination reports, and other data.
[0080] In a possible implementation, the natural language text is a medical text, and the medical text may be an electronic medical record (Electronic Healthcare Record), an electronic personal health record, including a series of electronic records with preservation value such as medical records, electrocardiograms, and medical images.
[0081] In one possible implementation, various types of user consultation behaviors can be screened out from the online conversation between the user and the doctor based on an artificial intelligence model.
[0082] In step S102 of some embodiments, inputting the user characteristic index into the churn prediction model means taking the user characteristic index as input and obtaining a preliminary prediction result of user churn, i.e., a first churn prediction result, through the churn prediction model. The churn prediction model may adopt a decision tree model, a random forest model, a neural network model, etc.
[0083] Exemplarily, the process of churn prediction using a decision tree model is as follows:
[0084] First, the user characteristic index obtained in step S101 is subjected to min-max normalization processing to achieve dimensionless normalization of the data.
[0085] Then, the standardized user feature index data is input into the decision tree model, and feature selection is performed on each feature to find the optimal feature subset, thereby obtaining the first churn prediction result. In the embodiment of the present application, the C4.5 decision tree algorithm is used to select features, and the C4.5 decision tree algorithm performs feature selection by information gain rate. The information gain calculation formula of feature A for data set D is: Gain(A)=Info(D)-Info_A(D), where data set D is a set of user feature indicators, and feature A is any feature in the user feature indicators.
[0086] Information gain rate (Information Gain Rate) is used to measure the ratio of the amount of information brought by an attribute when dividing a data set relative to the inherent uncertainty of the attribute. The information gain rate is obtained by dividing the information gain by the inherent value. Its purpose is to standardize the information gain to prevent bias towards attributes with more values. An attribute with a higher information gain rate means that even taking into account the uncertainty of the attribute itself, it can still provide a relatively high information gain, that is, its contribution to classification is significant relative to its inherent uncertainty. The information gain rate is calculated as follows: Gainr(A) = Gain(A) / H(A). Among them, H(A) is the entropy of A, and entropy is used to measure the uncertainty of samples in a data set. The higher the entropy, the greater the uncertainty. The calculation formula of H(A) is as follows:
[0087]
[0088] In the above formula, n is the number of categories of feature A, D j is the number of samples corresponding to the i-th value of feature A, and |D| is the number of samples.
[0089] The above is an example of obtaining the first churn prediction result according to the decision tree model based on the C4.5 decision tree algorithm in this application. In addition to the C4.5 decision tree algorithm, a decision tree model constructed by the ID3 or CART decision tree algorithm can also be used.
[0090] In step S103 of some embodiments, the disease label refers to a label related to the follow-up visit given to each user, through which it is possible to directly know whether the user needs a follow-up visit, and the follow-up visit period and other contents related to the follow-up visit. The disease label can be a label directly given to the user by the consulting doctor according to the user's physical condition, disease type or medication frequency, or it can be a label automatically given to the user by the system after identifying the user's medical data and medical text through big data or artificial intelligence.
[0091] In one possible implementation, the medical data may be personal health records, prescriptions, examination reports, and other data.
[0092] In a possible implementation, the natural language text is a medical text, and the medical text may be an electronic medical record (Electronic Healthcare Record), an electronic personal health record, including a series of electronic records with preservation value such as medical records, electrocardiograms, and medical images.
[0093] In step S104 of some embodiments, the target churn prediction result refers to the final churn prediction conclusion. According to the disease label and the first churn prediction result, the target churn prediction result of the user is determined. It means that the prediction of whether the user will churn is made based on the need for follow-up visits in the disease label. If the user does not have a follow-up visit within the follow-up period, the user churn is directly judged as lost regardless of the first churn prediction result. At this time, regardless of the first churn prediction result, the user's target churn prediction result is user churn. That is, the user churn situation is determined based on the user's disease label first. If the disease label shows that the user has no need for follow-up visits or the user has no disease label, which is insufficient to determine whether the user will have a follow-up visit, the first churn prediction result is used as the target churn prediction result, replacing the traditional method of directly predicting based on the model results, which can improve the accuracy of churn prediction for consulting users.
[0094] See also Figure 2 , Figure 2 is a flowchart of the specific method of step S104 in some embodiments of the present application. In some embodiments of the present application, when the disease label indicates that the user has a need for a follow-up visit, step S104 specifically includes but is not limited to steps S201 to S203. Figure 2 These three steps are described in detail:
[0095] Step S201, obtaining the user's follow-up visit behavior data, where the follow-up visit behavior data is used to indicate whether the user has had a follow-up visit within the follow-up visit period;
[0096] Step S202: if the user has a follow-up visit within the follow-up visit period, use the first churn prediction result as the target churn prediction result for the user;
[0097] Step S203: When the user does not have a follow-up visit within the follow-up visit period, determine a target churn prediction result for the user based on the first churn prediction result.
[0098] In the steps S201 to S203 shown in the embodiment of the present application, when the disease label indicates that the user has a need for a follow-up visit, the follow-up visit behavior data of the user is obtained; if the user has a follow-up visit within the follow-up visit period, the first churn prediction result is used as the target churn prediction result of the user; if the user has not had a follow-up visit within the follow-up visit period, the target churn prediction result of the user is determined according to the first churn prediction result. The accuracy of user churn prediction can be further improved.
[0099] In step S201 of some embodiments, obtaining the user's follow-up behavior data means first determining whether the user has a follow-up need based on the disease label. If so, querying the user's historical consultation records based on the follow-up deadline in the disease label to determine whether the user has had a follow-up visit within the follow-up deadline.
[0100] It should be noted that the user's follow-up behavior data can be obtained directly by querying the user's registration / consultation records during the follow-up period, or it can be given after analyzing the user's medical data or medical text through artificial intelligence.
[0101] In step S202 of some embodiments, when the user has a follow-up visit within the follow-up period, the first churn prediction result is used as the target churn prediction result for the user. This means that when the follow-up behavior data of the user shows that the user has a follow-up visit within the follow-up period, and it is impossible to determine whether the user has churned based on the disease label, the first churn prediction result is used as the target churn prediction result for the user, that is, the output result of the churn prediction model is used as the final churn prediction conclusion.
[0102] In step S203 of some embodiments, when the user does not have a follow-up visit within the follow-up period, the target churn prediction result of the user is determined based on the first churn prediction result, which means that when the follow-up behavior data of the user shows that the user has not had a follow-up visit within the follow-up period, the first churn prediction result is used as the target churn prediction result of the user. It should be understood that at this time, the target churn prediction result of the user is to judge the churn of the user, that is, the conclusion of user churn is used as the final churn prediction conclusion.
[0103] See also Figure 3 , Figure 3is a flowchart of a specific method of step S203 in some embodiments of the present application. In some embodiments of the present application, step S203 specifically includes but is not limited to step S301 and step S302. Figure 3 These two steps are described in detail:
[0104] S301, when a first churn prediction result predicts user churn, using the first churn prediction result as a target churn prediction result for the user;
[0105] S302: When the first churn prediction result predicts that the user has not churned, the first churn prediction result is corrected to obtain a second churn prediction result, and the second churn prediction result is used as a target churn prediction result.
[0106] In the steps S201 to S202 shown in the embodiment of the present application, when the first churn prediction result predicts user churn, the first churn prediction result is used as the target churn prediction result for the user; when the first churn prediction result predicts that the user has not churned, the first churn prediction result is corrected to obtain a second churn prediction result, and the second churn prediction result is used as the target churn prediction result. This can further improve the accuracy of user churn prediction.
[0107] In step S301 of some embodiments, when the first churn prediction result predicts user churn, the first churn prediction result is used as the target churn prediction result for the user; this means that if the follow-up history data shows that the user has not had a follow-up visit within the follow-up period, the prediction result based on the disease label should be user churn, and at this time, if the first churn prediction result obtained according to the churn prediction model also predicts user churn, there is no need to correct the first churn prediction result, and the first churn prediction result can be directly used as the target churn prediction result, because at this time the prediction results of the two methods are consistent, that is, the first churn prediction result is the same as the result predicted according to the disease label and the follow-up history data, both of which judge user churn, and at this time the target churn prediction result is user churn.
[0108] In step S302 of some embodiments, when the first churn prediction result predicts that the user has not churned, the first churn prediction result is corrected to obtain a second churn prediction result, and the second churn prediction result is used as the target churn prediction result. The second churn prediction result means that if the follow-up history data shows that the user has not had a follow-up visit within the follow-up period, the prediction result based on the disease label should be user churn, but at this time the first churn prediction result predicts that the user has not churned. Because the prediction results of the two methods are inconsistent at this time, the first churn prediction result needs to be corrected. Obviously, when the user has not had a follow-up visit within the follow-up period, it can be concluded that the user has churned. At this time, the first churn prediction result obtained by the churn prediction model is wrong, so the first churn prediction result is corrected, and the prediction result based on the disease label is used as the second churn prediction result. At this time, the target churn prediction result is user churn.
[0109] See also Figure 4 , Figure 4 is a flowchart of the specific method of step S104 in some embodiments of the present application. In some embodiments of the present application, when the disease label indicates that the user does not need a follow-up visit, step S104 specifically includes but is not limited to step S401. Figure 4 This step is described in detail:
[0110] Step S401: Use the first churn prediction result as the target churn prediction result.
[0111] In step S401 of the embodiment of the present application, when the disease label indicates that the user does not need a follow-up visit, the first churn prediction result is used as the target churn prediction result, which can reduce the program processing flow and save computing resources.
[0112] In step S401 of some embodiments, if the disease label indicates that the user does not need a follow-up visit, the first churn prediction result output by the churn prediction model is directly used as the target churn prediction result. That is, if the user does not need a follow-up visit, the churn situation of the user cannot be judged based on whether or not the user has a follow-up visit, so no relevant judgment or prediction is performed, and the first churn prediction result output by the churn prediction model is directly used as the final churn prediction conclusion.
[0113] See also Figure 5 , Figure 5 is a flowchart of a specific method of step S103 in some embodiments of the present application. In some embodiments of the present application, step S103 specifically includes but is not limited to steps S501 to S503. Figure 5 These three steps are described in detail:
[0114] Step S501, obtaining a follow-up visit prediction model based on big data learning;
[0115] Step S502: Obtain the user's diagnosis and treatment data, where the diagnosis and treatment data is used to indicate the user's disease information;
[0116] Step S503: input the user's diagnosis and treatment data into the follow-up prediction model to obtain the user's disease label.
[0117] In the steps S501 to S503 shown in the embodiment of the present application, the diagnosis and treatment data of the user is obtained by obtaining a follow-up visit prediction model based on big data learning, and the diagnosis and treatment data is used to indicate the disease information of the user. The diagnosis and treatment data of the user is input into the follow-up visit prediction model to obtain the disease label of the user. The disease label of the user can be automatically generated without the need for the doctor to add it manually, which saves the doctor's consultation time and improves the consultation efficiency.
[0118] In step S501 of some embodiments, the follow-up prediction model based on big data learning refers to the ability to predict follow-up related content such as whether the user has follow-up behavior and follow-up deadline based on the user's medical data. The follow-up prediction model based on big data learning can adopt models such as logistic regression, random forest, support vector machine, gradient boosting tree, neural network, etc. The follow-up prediction model can adopt existing related models or be constructed according to needs.
[0119] As an example, a construction process of a follow-up visit prediction model is given below:
[0120] The historical diagnosis and treatment data of multiple users are obtained to form a historical diagnosis and treatment data set; in a possible implementation, the historical diagnosis and treatment data may be personal health records, prescriptions, examination reports and other data; in a possible implementation, the historical diagnosis and treatment data may be an electronic medical record (Electronic Healthcare Record), an electronic personal health record, including a series of electronic records with preservation value such as medical records, electrocardiograms, and medical images.
[0121] It should be noted that the multiple users include users who have a need for a follow-up visit and users who do not have a need for a follow-up visit.
[0122] Then the historical diagnosis and treatment data set is preprocessed and divided into a training set, a validation set, and a test set;
[0123] Define the features associated with the need for follow-up visits in the dataset and build a preliminary neural network model based on the relevant features;
[0124] The preliminary neural network model was trained using training sets, validation sets, and test sets, and finally a trained revisit prediction model was obtained.
[0125] It should be noted that the above is only a process for constructing a follow-up prediction model based on a neural network. This application can select different algorithms to construct a follow-up prediction model according to actual needs.
[0126] In step S502 of some embodiments, the user's medical data is obtained. The medical data may be the user's specific symptoms, physical condition, medication frequency, and drug information, etc., which can reflect the user's disease information.
[0127] For example, if a user has a specific chronic disease, or based on the historical medication history, it is found that the user needs regular follow-up visits and re-prescriptions to take medications, these can all be used as information on whether the user needs follow-up visits; in addition to the above diagnosis and treatment data, some users in special periods also need regular follow-up visits. For example, if the past history includes respiratory diseases, mental illnesses, immune system diseases, urinary system diseases, peri-pregnancy and perinatal conditions, these all belong to the user's diagnosis and treatment data.
[0128] In one possible implementation, the diagnosis and treatment data may be data such as personal health records, prescriptions, and examination reports;
[0129] In one possible implementation, the diagnosis and treatment data may be an electronic medical record (Electronic Healthcare Record), an electronic personal health record, including a series of electronic records with preservation value such as medical records, electrocardiograms, and medical images.
[0130] In step S503 of some embodiments, the diagnosis and treatment data of the user is input into the follow-up visit prediction model to obtain the disease label of the user. This means that the diagnosis and treatment data obtained in step S502 is used as input, and the follow-up visit prediction model is used to predict whether the user has a need for a follow-up visit, and if there is a need for a follow-up visit, the follow-up visit deadline of the user is predicted. By generating the disease label of the user through the follow-up visit prediction model based on big data, there is no need for the doctor to add it manually, which can save the doctor's consultation time and improve the consultation efficiency.
[0131] See also Figure 6 , Figure 6 is a flowchart of a specific method of step S101 in some embodiments of the present application. In some embodiments of the present application, step S101 specifically includes but is not limited to steps S601 to S603. Figure 6 These three steps are described in detail:
[0132] Step S601, obtaining user ID data;
[0133] Step S602: Obtain user consultation behavior data corresponding to each user ID according to the user ID data;
[0134] Step S603: Use the user consultation behavior data corresponding to each user ID as a user characteristic indicator.
[0135] In the steps S601 to S603 shown in the embodiment of the present application, by obtaining user ID data, the user consultation behavior data corresponding to each user ID is obtained according to the user ID data, and the user consultation behavior data corresponding to each user ID is used as a user feature indicator. Based on the particularity of Internet medical consultation users, the present application removes the basic attributes of users used in traditional predictions when selecting user features, and uses the user ID as an independent churn prediction object, thereby reducing the possibility of deviation in the prediction results of the churn prediction model and further improving the accuracy of churn prediction.
[0136] In step S601 of some embodiments, obtaining user ID data refers to obtaining a unique identifier corresponding to each user ID of the Internet medical platform. The unique identifier can be a nickname or login name of the user ID, a mobile phone number or email address bound to the user ID, or a UID number corresponding to the user ID in the background database, etc.
[0137] It should be understood that a nickname can be selected as the unique identifier of a user ID only when users on the Internet platform are not allowed to have the same user nickname.
[0138] In step S602 of some embodiments, the user consultation behavior data corresponding to each user ID is obtained based on the user ID data, which means that no matter whether the user ID is associated with one or several patients, all the consultation behavior data under each user ID are counted only in units of user ID, without distinguishing whether the user ID corresponds to one or several patients.
[0139] In one possible implementation, the consultation behavior data may be all personal health records, prescriptions, examination reports, and other data under the user ID;
[0140] In one possible implementation, the consultation behavior data may be all electronic medical records (Electronic Healthcare Record) under the user ID, electronic personal health records, including medical records, electrocardiograms, medical images, and other electronic records that are worth preserving for future reference.
[0141] In step S603 of some embodiments, due to the particularity of online consultations on the Internet, multiple patients may be bound to one user ID, and in actual consultations, a single user ID may consult multiple patients, and these multiple patients are usually family members of the registered user ID. Since some Internet medical platforms do not have mandatory identity authentication, the consulting doctor and the system cannot distinguish which patient is the one undergoing a particular consultation, and the basic attributes of each patient may be very different.
[0142] Therefore, in the embodiment of the present application, the user consultation behavior data corresponding to each user ID is used as the user feature index, which means that when selecting user features, the traditional selection of user basic attribute features is abandoned to reduce the impact on the prediction results. That is, when the churn prediction model is used for prediction, the user ID is used as an independent individual for prediction. Regardless of whether the user ID corresponds to one or more patients, only the churn probability of each user ID is predicted, which can improve the accuracy of the first churn prediction result output by the churn prediction model.
[0143] See also Figure 7 In the embodiment of the present application, the churn prediction model is a decision tree model. Before step S202 in some embodiments, the method for predicting churn of consultation users may also include but is not limited to steps S701 to S703. Figure 7 These three steps are described in detail:
[0144] Step S701, obtaining a user history medical consultation data set including multiple user history medical consultation features;
[0145] Step S702: Calculate the information gain rate of each user's historical medical consultation features, perform feature selection based on the information gain rate, and obtain feature selection results;
[0146] Step S703: construct a decision tree model based on the feature selection results.
[0147] In the steps S701 to S703 shown in the embodiment of the present application, a user history consultation data set including multiple user history consultation features is obtained, the information gain rate of each user history consultation feature is calculated, and feature selection is performed according to the information gain rate to obtain a feature selection result; a decision tree model is constructed according to the feature selection result. A decision tree model matching the user features of online medical consultations on the Internet can be obtained, further improving the accuracy of the first churn prediction result.
[0148] In step S701 of some embodiments, the user's historical consultation feature refers to the records of all historical consultation behaviors and loss situations of the user. All historical consultation behaviors and loss situations of multiple users are collected to form a user historical consultation data set. The historical consultation behavior includes but is not limited to collecting the user's medical records, including medical history, diagnosis, treatment, drug use, laboratory test results, doctor's advice, and patient's personal information. Personal information includes age, gender, lifestyle, etc. It should be understood that the more users there are in the user's historical consultation data set, the more accurate the prediction results that can be obtained by the constructed decision tree model. The user's historical consultation features can be obtained through the user's medical data, can also be obtained through the user's natural language text, and can also be obtained by screening through an artificial intelligence model.
[0149] In one possible implementation, the medical data may be personal health records, prescriptions, examination reports, and other data.
[0150] In a possible implementation, the natural language text is a medical text, and the medical text may be an electronic medical record (Electronic Healthcare Record), an electronic personal health record, including a series of electronic records with preservation value such as medical records, electrocardiograms, and medical images.
[0151] In one possible implementation, various historical consultation features of the user can be screened out from the historical conversation records between the user and the doctor based on an artificial intelligence model.
[0152] In step S702 of some embodiments, the information gain rate of each user's historical medical consultation feature is calculated, and feature selection is performed based on the information gain rate to obtain a feature selection result. This means that the information gain rate of each feature in the user's historical medical consultation feature is first calculated, and the feature with the highest information gain rate is selected from all features as the root node of the decision tree. The relevant content of the information gain rate has been introduced in detail in the previous text and will not be described in detail here.
[0153] In step S703 of some embodiments, building a decision tree model based on the feature selection result means that the information gain rate of each feature is calculated in step S702, and after obtaining a feature selection result, for each feature value, a child node is created, and the data set is divided according to different values of the feature, and the above process is repeated for each child node until the stop condition is met, such as the number of samples in the node is lower than a certain threshold or all samples belong to the same category, and then a training set and a data set are built according to the user's historical consultation data set, and a decision tree model is built by using the training set and the data set through a recursive splitting and pruning process, and the decision tree model is trained using the training data set. The decision tree model makes branch decisions based on the feature value, and finally achieves the prediction target. Since all the user's historical consultation features are derived from the online consultation of the Internet medical platform, a decision tree model that matches the user features of the Internet medical online consultation can be obtained through the above method, which meets the particularity of the consultation business and can further improve the accuracy of the first churn prediction result.
[0154] The present application also provides a device for predicting user churn. Figure 8 , the above-mentioned consultation user churn prediction method can be implemented, and the device includes: a user feature acquisition module 801, a first churn prediction module 802, a disease label acquisition module 803 and a second churn prediction module 804.
[0155] A user feature acquisition module 801 is used to acquire a user feature index, which is used to indicate the user's medical consultation behavior;
[0156] A first churn prediction module 802, configured to perform churn prediction processing according to user characteristic indicators to obtain a first churn prediction result;
[0157] The disease label acquisition module 803 is used to acquire the disease label of the user, and the disease label is used to indicate the user's need for follow-up consultation;
[0158] The second churn prediction module 804 is used to determine a target churn prediction result of the user according to the disease label and the first churn prediction result.
[0159] The medical consultation user churn prediction device of the embodiment of the present application is used to execute the medical consultation user churn prediction method in the above embodiment. Its specific processing process is the same as the medical consultation user churn prediction method in the above embodiment, and will not be repeated here.
[0160] The embodiment of the present application also provides an electronic device, the electronic device comprising: a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory, and the program is executed by the processor to realize the above-mentioned consultation user churn prediction method. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.
[0161] See also Fig. 9 , Fig. 9 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:
[0162] The processor 901 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;
[0163] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 902, and the processor 901 calls and executes the method for predicting the loss of consultation users in the embodiment of this application;
[0164] Input / output interface 903, used to implement information input and output;
[0165] Communication interface 904, used to realize communication interaction between the device and other devices, which can be realized through wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WI FI, Bluetooth, etc.);
[0166] A bus 905 that transmits information between various components of the device (e.g., the processor 901, the memory 902, the input / output interface 903, and the communication interface 904);
[0167] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .
[0168] An embodiment of the present application also provides a storage medium, which is a computer-readable storage medium used for computer-readable storage. The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the above-mentioned consultation user churn prediction method.
[0169] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0170] The method, device, electronic device and storage medium proposed in the present application for predicting the churn of consultation users obtain user characteristic indicators, which are used to indicate the consultation behavior of the user; input the user characteristic indicators into the churn prediction model to obtain a first churn prediction result; obtain the disease label of the user, which is used to indicate the user's need for follow-up consultation; and determine the target churn prediction result of the user based on the disease label and the first churn prediction result. The present application combines the disease label and the first churn prediction result to predict the churn probability of the user, which can improve the accuracy of churn prediction for consultation users.
[0171] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.
[0172] It can be understood by those skilled in the art that Figure 1-7 The technical solutions shown in the figure do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than those shown in the figure, or a combination of certain steps, or different steps.
[0173] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0174] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.
[0175] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0176] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can indicate: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can indicate: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.
[0177] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0178] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0179] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0180] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.
[0181] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.
Claims
1. A method for predicting user churn in a medical consultation, characterized in that: The method comprises: Acquire a user characteristic index, where the user characteristic index is used to indicate the user's medical consultation behavior; Inputting the user characteristic index into a churn prediction model to obtain a first churn prediction result; Acquire a disease label of the user, where the disease label is used to indicate a need for a follow-up visit of the user; A target churn prediction result for the user is determined according to the disease label and the first churn prediction result.
2. The method according to claim 1, characterized in that The disease label indicates that the user needs a follow-up visit; The step of determining the target churn prediction result of the user according to the disease label and the first churn prediction result includes: Acquire the user's follow-up visit behavior data, where the follow-up visit behavior data is used to indicate whether the user has had a follow-up visit within the follow-up visit period; In a case where the user has a follow-up consultation within the follow-up consultation period, using the first churn prediction result as the target churn prediction result for the user; In a case where the user does not have a follow-up visit within the follow-up visit period, a target churn prediction result for the user is determined based on the first churn prediction result.
3. The method according to claim 2, characterized in that Determining a target churn prediction result of the user according to the first churn prediction result includes: In a case where the first churn prediction result predicts churn of the user, using the first churn prediction result as a target churn prediction result for the user; In the case that the first churn prediction result predicts that the user has not churned, the first churn prediction result is corrected to obtain a second churn prediction result, and the second churn prediction result is used as a target churn prediction result.
4. The method according to claim 1, characterized in that: The disease label indicates that the user does not need a follow-up visit; The step of determining the target churn prediction result of the user according to the disease label and the first churn prediction result includes: The first churn prediction result is used as the target churn prediction result.
5. The method according to claim 1, characterized in that: The obtaining of the disease label of the user includes: Obtain a follow-up visit prediction model based on big data learning; Acquiring the diagnosis and treatment data of the user, where the diagnosis and treatment data is used to indicate the disease information of the user; The diagnosis and treatment data of the user is input into the follow-up visit prediction model to obtain the disease label of the user.
6. The method according to claim 1, characterized in that The obtaining of user characteristic indicators includes: Get user ID data; Obtain the user consultation behavior data corresponding to each user ID according to the user ID data; The user consultation behavior data corresponding to each user ID is used as a user characteristic indicator.
7. The method according to claim 1, characterized in that The churn prediction model is a decision tree model; Before inputting the user characteristic index into the churn prediction model to obtain the first churn prediction result, the method further includes: Obtain a user history medical consultation data set including multiple user history medical consultation features; Calculate the information gain rate of each of the user's historical medical consultation features, perform feature selection according to the information gain rate, and obtain a feature selection result; A decision tree model is constructed based on the feature selection results.
8. A device for predicting user churn during consultation, characterized in that: The device comprises: A user feature acquisition module, used to acquire user feature indicators, wherein the user feature indicators are used to indicate the user's medical consultation behavior; A first churn prediction module, used to perform churn prediction processing according to the user characteristic index to obtain a first churn prediction result; A disease label acquisition module, used to acquire the disease label of the user, wherein the disease label is used to indicate the user's need for a follow-up visit; The second churn prediction module is used to determine a target churn prediction result of the user according to the disease label and the first churn prediction result.
9. An electronic device, characterized in that: The electronic device includes a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory. When the program is executed by the processor, the steps of the method for predicting user churn as described in any one of claims 1 to 7 are realized.
10. A storage medium, the storage medium being a computer-readable storage medium, used for computer-readable storage, characterized in that: The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the method for predicting user churn according to any one of claims 1 to 7.
Citation Information
Patent Citations
Clinic doctor check evaluating system and evaluating process
CN101546418A
Method for screening attribute data and customer loss early warning method
CN110019166A
Loss prediction model training method and user loss prediction method
CN114239965A
Customer loss prediction method and device, computer equipment and storage medium
CN116049666A
Construction method of user loss prediction model, and user loss prediction method and device
CN116934385A
