A collaborative screen information dynamic aggregation method fusing user intention prediction

Abstract

The application discloses a kind of collaborative screen information dynamic aggregation methods of fusing user intention prediction, it is related to data analysis technical field.The method includes obtaining intention performance index based on first time series interaction data;If the intention performance index is greater than first preset value, then obtain a plurality of second time series interaction data based on hospital information service platform;Based on each second time series interaction data, the operation guide of the first user is generated.The application calculates the intention performance index that can represent the possibility of user operation reflecting real intention by first time series interaction data, only in intention performance index greater than first preset value (that is, the intention of user is clear) when calling the historical second time series interaction data of other users in platform to generate operation guide, effectively solve the technical problem that operation guide is separated from actual operation demand of user caused by scene as core, fixed process push in traditional hospital information display.

Description

A method for dynamic aggregation of collaborative screen information based on user intent prediction Technical Field

[0001] This application relates to the field of data analysis technology, specifically a method for dynamically aggregating collaborative screen information by integrating user intent prediction. Background Technology

[0002] Hospitals are typical information interaction environments characterized by "multiple roles, multiple scenarios, and high timeliness," encompassing various roles including doctors, nurses, patients, and medical technicians such as lab technicians and pharmacists. Their information needs differ significantly across scenarios such as outpatient consultations, ward rounds, surgical procedures, medication dispensing, and patient waiting. However, traditional multi-screen information display models have obvious shortcomings and struggle to meet the specific needs of medical settings. Specifically, the information required differs fundamentally for users with different identities. For example, doctors need to quickly access professional information such as patient medical history, examination reports, and medication records; patients, on the other hand, want to know basic information such as their waiting stage and attending physician. Traditional multi-screen information display models often push a large amount of irrelevant content, making it difficult for users to quickly locate and obtain the information they truly need.

[0003] To address this issue, traditional data aggregation methods are often scenario-based. For example, when a user picks up medication at a pharmacy, they need to scan a code to obtain a pickup number. However, this model heavily relies on the user's complete familiarity with the process. Taking the medication pickup scenario as an example, users must understand the prerequisite that "scanning a code to obtain a number is required before picking up medication." However, in reality, patients and their families often lack understanding of the specific procedures, rendering information pushed to them in this manner completely useless. Therefore, it is necessary to analyze user actions to clarify their core intentions and deliver precise operational guidance based on those intentions. However, current mainstream information pushes mostly adopt fixed processes, while user needs vary greatly at different times (for example, a patient's core needs are drastically different before and after surgery), making it difficult to obtain the user's true intentions and resulting in generally low accuracy of pushed information. Summary of the Invention

[0004] The purpose of this application is to provide a collaborative screen information dynamic aggregation method that integrates user intent prediction to solve the technical problem that existing methods cannot accurately push operation guidance.

[0005] To achieve the above objectives, this application provides the following technical solution:

[0006] A method for dynamically aggregating collaborative screen information based on user intent prediction includes:

[0007] Based on a hospital information service platform, first-sequence interaction data of a first user is obtained; the hospital information service platform is pre-established and includes multiple operation module interfaces; the first user is any user in the hospital information service platform; the first-sequence interaction data includes at least the first user's click information and dwell time information on each operation module interface in time.

[0008] Based on the first time-series interaction data, an intent performance index is obtained; the intent performance index is at least used to characterize the likelihood that the interaction operation of the first user reflects the true intent of the first user.

[0009] If the intent performance index is greater than the first preset value, then multiple second time-series interaction data are obtained based on the hospital information service platform; the second time-series interaction data are the historical time-series interaction data of any user other than the first user in the hospital information service platform;

[0010] Based on the second time-series interaction data, the operation guide for the first user is generated.

[0011] As a specific solution in the technical solution of this application, the step of obtaining the intent performance index based on the first time-series interaction data includes:

[0012] Based on the first time-series interaction data, the anomaly index of each operation module interface is obtained; the anomaly index is used to characterize at least the number of times the first user clicks the corresponding operation module interface and the length of time the first user stays on the corresponding operation module interface.

[0013] Based on various anomaly indices, the intent performance index is obtained.

[0014] As a specific solution in this application, the step of obtaining the anomaly index of each operation module interface based on the first time-series interaction data includes:

[0015] Based on the interfaces of each operation module, obtain the first operation module interface; the first operation module interface is the operation module interface that the first user clicked in the first time series interaction data, and the current time series needs to obtain the abnormal index corresponding to the first operation module interface;

[0016] Based on the first time-series interaction data, the number of clicks and the dwell time are obtained; the number of clicks is the number of times the first user clicks on the interface of the first operation module; the dwell time is the average dwell time after the first user clicks on the interface of the first operation module multiple times.

[0017] Based on the number of clicks and the dwell time, an anomaly index of the first operation module interface is obtained; the anomaly index is positively correlated with the number of clicks and negatively correlated with the dwell time.

[0018] As a specific solution in the technical solution of this application, the step of obtaining the intent performance index based on various anomaly indices includes:

[0019] Based on each anomaly index, a first value and a second value are obtained; the first value is used to characterize at least the average value of each anomaly index; the second value is used to characterize at least the difference between each anomaly index.

[0020] Based on the first value and the second value, the intention performance index is obtained; the intention performance index is negatively correlated with the first value and positively correlated with the second value.

[0021] As a specific solution in this application, each operation module interface includes a parent operation module interface and a sub-operation module interface, wherein the level of the sub-operation module interface is lower than the level of the corresponding parent operation module interface; after obtaining the intent performance index based on the first value and the second value, the method further includes:

[0022] Based on the first time-series interaction data, an operation level time-series sequence is obtained; each sequence value in the operation level time-series sequence corresponds to the level of each operation module interface clicked by the first user.

[0023] Based on the operation level time sequence, multiple sequence strings are obtained;

[0024] Based on each sequence string, a correction coefficient is obtained; the correction coefficient is used to at least characterize the quantity of each sequence string.

[0025] The intention performance index is corrected based on the correction coefficient.

[0026] As a specific solution in this application, the step of obtaining multiple sequence strings based on the operation level time series includes:

[0027] Traverse the operation level time sequence to obtain the current first sequence value and the current second sequence value; the current second sequence value is greater than the current first sequence value; the current first sequence value and the current second sequence value are time-adjacent, and the current second sequence value is located after the current first sequence value.

[0028] All sequence values ​​between the current first sequence value and the historical second sequence value, the current first sequence value, and the historical second sequence value are combined into a sequence string; the acquisition time of the historical second sequence value is adjacent to the current second sequence value, and the current second sequence value is located after the time of the historical second sequence value; the historical second sequence value includes an initial second sequence value, which is the first sequence value in the operation level time sequence;

[0029] Multiple sequence strings are obtained after the operation level time sequence traversal is completed.

[0030] As a specific solution in this application, the step of obtaining the correction coefficient based on each sequence string includes:

[0031] Based on each sequence string, multiple clusters are obtained; the parent operation module interface corresponding to each sequence string in each cluster has the same level;

[0032] Based on each cluster, a first quantity is obtained; the first quantity is the number of sequence strings in the first cluster; the first cluster is the cluster with the largest number of corresponding sequence strings among all clusters.

[0033] Based on each sequence string, obtain the second quantity; the second quantity is the total number of all sequence strings.

[0034] Based on the first quantity and the second quantity, a correction coefficient is obtained; the correction coefficient is positively correlated with the first quantity and negatively correlated with the second quantity.

[0035] As a specific solution in this application, the step of generating the operation guide for the first user based on various second time-series interaction data includes:

[0036] Based on each second time-series interaction data, a third time-series interaction data is obtained; the third time-series interaction data is the time-series interaction data that has the highest similarity to the first time-series interaction data among each second time-series interaction data.

[0037] Based on the third time-series interaction data, operation instructions for the first user are generated.

[0038] As a specific solution in this application, the step of generating the operation guide for the first user based on various second time-series interaction data includes:

[0039] Based on each second time-series interaction data, multiple fourth time-series interaction data are obtained; the fourth time-series interaction data is any time-series interaction data among the second time-series interaction data whose similarity to the first time-series interaction data is greater than or equal to a second preset value.

[0040] The frequent pattern growth algorithm generates multiple frequent item sets from each fourth time-series interaction data; the frequent item set is the operation sequence of the operation module interface combination that meets the preset minimum support in each fourth time-series interaction data.

[0041] The current operation state of the first user is determined based on the first time-series interaction data; the current operation state includes at least the operation module interface currently operated by the first user.

[0042] Filter out a target set of frequent items that matches the current operation state from the multiple sets of frequent items;

[0043] Based on the target set of frequent items, operation instructions are generated for the first user.

[0044] As a specific solution in this application, the step of filtering out a target set of frequent items that matches the current operation state from multiple sets of frequent items includes:

[0045] Based on each frequent item set, multiple first frequent item sets are obtained; the first frequent item set is any frequent item set that has the current operation state in each frequent item set.

[0046] Based on each first frequent item set, multiple second frequent item sets are obtained; the second frequent item set is any set of frequent items in each first frequent item set whose similarity to the historical operation state of the first user is greater than a third preset value.

[0047] Based on each of the second frequent item sets, the target frequent item set is obtained.

[0048] Compared with the prior art, the beneficial effects of this application are:

[0049] This application obtains first-sequence interaction data from a pre-established hospital information service platform, including the timing of clicks and dwell times on the interfaces of each operation module of the first user. It then calculates an intent performance index based on this first-sequence interaction data, which characterizes the likelihood that the user's actions reflect their true intent. Only when the intent performance index exceeds a first preset value (i.e., the user's intent is clear) is the application retrieved from historical second-sequence interaction data of other users within the platform to generate operation guidance. This effectively solves the technical problem in traditional hospital information displays where scenario-based, fixed-process pushes lead to operation guidance that deviates from the user's actual operational needs. It achieves accurate intent matching and dynamic guidance based on the user's actual operational behavior, ensuring the accuracy and practicality of information interaction in medical scenarios, avoiding useless guidance interference, and meeting the hospital's information interaction needs of "multi-role, multi-scenario, and high-timeliness." Attached Figure Description

[0050] Figure 1 is a flowchart illustrating a collaborative screen information dynamic aggregation method that integrates user intent prediction, as proposed in an embodiment of this application.

[0051] Figure 2 is a flowchart illustrating a method for obtaining an anomaly index according to an embodiment of this application;

[0052] Figure 3 is a flowchart illustrating a method for correcting the intention performance index proposed in an embodiment of this application. Detailed Implementation

[0053] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0054] The terms "first," "second," etc., used in the specification and accompanying drawings of this application embodiments are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. For example, the first timing interaction data and the second timing interaction data mentioned below belong to different timing interaction data. It should be understood that such names can be used interchangeably where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or modules is not necessarily limited to those explicitly listed, but may include other steps or modules not explicitly listed or inherent to these processes, methods, products, or devices. The module divisions appearing in the embodiments of this application are merely logical divisions; in actual applications, there may be other division methods. For example, multiple modules may be combined into or integrated into another system, or some features may be ignored or not performed. In addition, the shown or discussed mutual couplings or direct couplings or communication connections may be through some interfaces, and the indirect couplings or communication connections between modules may be electrical or other similar forms, none of which are limited in the embodiments of this application. Moreover, the modules or sub-modules described as separate components may or may not be physically separated, may or may not be physical modules, or may be distributed among multiple circuit modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the embodiments of this application.

[0055] Before understanding the embodiments of this application, it is necessary to clearly understand the definitions of the following technical terms:

[0056] Collaboration Display: A collaboration display is an intelligent display terminal centered on cross-device collaboration and multi-scenario interaction. It is not a traditional single monitor, but rather an intelligent interactive system that integrates hardware, software, and network technologies, aiming to break down device barriers and achieve efficient linkage between "people, devices, and content." Simply put, while traditional monitors are only responsible for "displaying content," the core value of a collaboration display is "facilitating collaboration." It enables seamless connection between multiple terminals such as mobile phones, computers, and tablets, supporting real-time content sharing, annotation, and control by multiple users, and is widely used in office, education, and medical scenarios.

[0057] Hospital Information Service Platform: A hospital information service platform is a comprehensive digital service system that integrates internal and external medical resources, business processes, and data information within a hospital. Its core objective is to break down information barriers at each stage of medical services, providing efficient, convenient, and secure end-to-end service support for patients, medical staff, hospital administrators, and medical collaborating institutions. It is a core infrastructure for hospitals to achieve "smart healthcare" transformation. Currently, the construction of hospital information service platforms is a mature technology and will not be elaborated upon further.

[0058] Operation Module Interface: The operation module interface is a visual interactive entry point designed by the hospital information service platform for different user roles (e.g., patients, doctors, nurses, and administrators) to perform specific medical or management tasks. It is not a single interface, but rather a set of independent module interfaces focused on multiple functions, based on core scenarios such as "patient treatment," "clinical diagnosis and treatment," and "hospital operations." Essentially, the operation module interface "transforms the platform's backend business logic and data into graphical tools that users can directly operate," ensuring that different roles can efficiently complete their specific tasks (e.g., doctors writing medical records, patients paying fees, or retrieving test reports).

[0059] Frequent Pattern Growth (FP-Growth) algorithm is a classic and efficient algorithm in the field of frequent pattern mining. Its core objective is to quickly discover "frequent itemsets" (i.e., combinations of frequently occurring sequence values) from large-scale datasets based on a preset "minimum support" (the lowest frequency threshold that an itemset must meet). The key innovation of this algorithm lies in avoiding the efficiency bottlenecks of the traditional Apriori algorithm, which involves "generating a large number of candidate sets" and "scanning the database multiple times." It condenses the frequent itemset information of the dataset into a compressed tree structure called an "FP-tree" using only two data scans. Then, using a bottom-up recursive strategy, it gradually mines all frequent itemsets containing the target items by extracting "conditional pattern bases" and constructing "conditional FP-trees," ultimately achieving efficient frequent pattern extraction.

[0060] To address the technical problem that existing technologies cannot accurately push operation guidance, this application proposes a method for dynamically aggregating collaborative screen information based on user intent prediction. As shown in Figure 1, this method includes steps S100 to S400.

[0061] Step S100: Based on the hospital information service platform, obtain the first time-series interaction data of the first user.

[0062] In this embodiment, the hospital information service platform is pre-established. As mentioned above, establishing a hospital information service platform is a mature technology and will not be elaborated upon here. In this embodiment, the hospital information service platform includes multiple operation module interfaces, which are pre-designed based on the needs of each user. For example, for patients, multiple operation module interfaces may include an outpatient registration module interface, a payment module interface, a report query module interface, a waiting progress viewing module interface, and a consultation guidance module interface. Among them, the outpatient registration module interface allows patients to select a department (e.g., internal medicine or surgery), doctor, and appointment time to complete online registration, and displays a successful registration notification, consultation number, and consultation room location. The payment module interface displays the patient's outstanding fees (e.g., consultation fees, examination fees, and medication fees), supports WeChat and / or Alipay and / or medical insurance payments, and provides payment record queries. The report query module interface displays the patient's laboratory (e.g., urinalysis, blood routine) and examination (e.g., CT or ultrasound) reports sorted by time, supports viewing report details (e.g., values, conclusions, and doctor's interpretation), and allows downloading and / or printing of reports. The waiting progress viewing module displays the current patient queue in the consultation room in real time (e.g., "Internal Medicine 3 Consultation Room, current call number: 025, your number: 030, estimated wait time: 20 minutes"), and provides waiting reminders (e.g., SMS and / or in-platform notifications). The consultation guidance module displays route navigation (e.g., hospital floor map + walking directions, such as "From Internal Medicine Consultation Room → Ultrasound Room: Take elevator 3 to the 2nd floor, turn left and walk 50 meters") based on the patient's current consultation stage (e.g., "Waiting for ultrasound" or "Waiting for medication"). It also provides suggestions on what items to bring or other reminders (e.g., "Fasting is required for abdominal ultrasound"). For doctors, multiple operation modules can include a patient medical record management module, a prescription issuance module, an examination result review module, an outpatient reception and scheduling module, and a consultation application module. The patient medical record management module is used to access patients' historical medical records (outpatient / inpatient records, past medical history, allergy history); create new current medical records (supporting template input, such as a "cold consultation template," allowing quick filling in of symptoms, signs, and diagnostic conclusions). The prescription issuance module is divided into "examination prescriptions" (e.g., ordering CT scans, blood tests, etc., selecting the examination department and priority) and "medication prescriptions" (e.g., selecting medication, dosage, and usage, automatically linking to medical insurance reimbursement rules), supporting prescription modification and / or cancellation, and displaying the execution status (e.g., "Examination order confirmed, awaiting patient execution"). The examination result review module is used to receive patients' examination reports in real time (e.g., blood test results uploaded by the laboratory, CT images uploaded by the radiology department), supporting highlighting key points in the report (e.g., circling abnormal values), and linking them to the medical record conclusion.The outpatient appointment scheduling module displays the doctor's appointment schedule for the day and / or week (e.g., patient name, appointment time, appointment type, etc.), supports temporary adjustments to the schedule (e.g., "cancel the 14:00 appointment, mark it as 'temporary consultation', etc."), and updates the patient waiting screen accordingly. The consultation application module is used to initiate multidisciplinary consultations when encountering difficult cases (e.g., selecting consultation departments, doctors, or consultation times, etc.), upload patient medical records and examination reports as the basis for consultation, and display the consultation application progress (e.g., "Orthopedics has accepted the order, consultation scheduled for tomorrow at 10:00, etc.").

[0063] It should be noted that, since it is impossible to exhaustively list all the operation module interfaces, the above-mentioned operation module interfaces are merely illustrative examples of the "multiple operation module interfaces" in this application, and do not represent that the multiple operation module interfaces of the hospital information service platform in this application can only be as shown in the above-mentioned operation module interfaces. It should be understood that the collaborative screen information dynamic aggregation method integrating user intent prediction proposed in this application is applicable to hospital information service platforms with arbitrary operation module interfaces. This application does not involve any improvement to the operation module interfaces, that is, the embodiments of the various operation module interfaces listed above are all mature technologies, and will not be elaborated further hereafter.

[0064] In this embodiment, the first user can be any user within the hospital information service platform. That is, the first user can be a doctor, patient, or nurse, etc. The first time-series interaction data includes at least the first user's click information and dwell time information on each operation module interface in time.

[0065] Step S200: Based on the first time-series interaction data, obtain the intent performance index.

[0066] In this embodiment, the intent performance index is used at least to characterize the likelihood that the first user's interaction reflects the first user's true intent.

[0067] It's important to note that after a user logs into the hospital information service platform, the platform actively records the user's clicks on relevant operation modules and the time spent on each module. Since users may not be familiar with the various operation modules on the platform and may not know which module contains the information they want to query, their actions on these modules may not reflect their true intentions. Therefore, it's necessary to analyze the user's true intentions based on the time spent on each module. For example, if a user doesn't find the information they want, their time spent on the corresponding module will be relatively short. For instance, if a user wants to make an appointment at the hospital, and they are familiar with the outpatient registration module, they will complete the registration by following the steps of "information entry - subject selection - date selection - doctor selection." However, if they are unfamiliar with the outpatient registration module (e.g., don't know which department to book an appointment with), they will frequently switch between modules to find the desired appointment, and their time spent on each module will be relatively short. Based on this, step S200 involves obtaining the intent performance index based on the first time-series interaction data, including steps S210 and S220.

[0068] Step S210: Based on the first time-series interaction data, obtain the anomaly index of each operation module interface.

[0069] In this embodiment, the anomaly index is used to characterize at least the number of times the first user clicks the corresponding operation module interface and the length of time the first user stays on the corresponding operation module interface.

[0070] As mentioned above, if a user frequently switches back to a certain operation module interface, it indicates that the user has not found the information they want in the operation module interface of the hospital information service platform, and the user's dwell time on that operation module interface is also short each time. In other words, in this embodiment, the number of clicks or the dwell time on a certain operation module interface can be used as the anomaly index of that operation module interface. In other words, in this embodiment, the anomaly index is used to characterize whether the user's actions on the corresponding operation module interface reflect the user's true intention. To obtain an accurate anomaly index, in one embodiment of this application, step S210, based on the first time-series interaction data, obtains the anomaly index of each operation module interface, including steps S211 to S213, as shown in Figure 2.

[0071] Step S211: Based on the interfaces of each operation module, obtain the interface of the first operation module.

[0072] In this embodiment, the first operation module interface is the operation module interface clicked by the first user in the first time-series interaction data, and the current time series needs to obtain the anomaly index corresponding to the first operation module interface. That is to say, in this embodiment, the method for obtaining the anomaly index corresponding to each operation module interface clicked by the user is the same as the method for obtaining the anomaly index corresponding to the first operation module interface.

[0073] Step S212: Based on the first time-series interaction data, obtain the number of clicks and the dwell time.

[0074] In this embodiment, the number of clicks refers to the number of times the first user clicks the interface of the first operation module. The dwell time is the average dwell time after the first user clicks the interface of the first operation module multiple times.

[0075] Step S213: Based on the number of clicks and the dwell time, obtain the anomaly index of the first operation module interface.

[0076] In this embodiment, the anomaly index is positively correlated with the number of clicks and negatively correlated with the dwell time. That is, in this embodiment, any reasonable method can be used to obtain the anomaly index of the first operation module interface based on the number of clicks and the dwell time. For example, in one embodiment of this application, step S213, the formula for calculating the anomaly index of the first operation module interface based on the number of clicks and the dwell time, can be as follows:

[0077]

[0078] in, This indicates the anomaly index of the first operation module interface; Indicates the duration of stay; This represents the number of clicks. Alternatively, in another embodiment of this application, step S213, the formula for calculating the anomaly index of the first operation module interface based on the number of clicks and the dwell time, can be as follows:

[0079]

[0080] in, This indicates the anomaly index of the first operation module interface; Indicates the duration of stay; Indicates the number of clicks; This represents an exponential function with the natural constant e as its base.

[0081] In this embodiment, a higher number of clicks results in a higher anomaly index, and a lower number of clicks results in a lower anomaly index. Similarly, a longer dwell time results in a lower anomaly index, and a shorter dwell time results in a higher anomaly index. In this embodiment, a higher anomaly index for a particular operation module interface indicates a more anomalous user action on that interface. In other words, the user's actions on that operation module interface are less likely to reflect their true intentions.

[0082] Step S220: Obtain the intent performance index based on each anomaly index.

[0083] In this embodiment, the intent performance index can be obtained based on each abnormal index in any reasonable manner. For example, the intent performance index can be the sum of each abnormal index; or, step S220, obtaining the intent performance index based on each abnormal index, includes steps S221 to S222.

[0084] Step S221: Based on each abnormal index, obtain the first value and the second value.

[0085] In this embodiment, the first value is used to characterize at least the average value of each anomaly index. The second value is used to characterize at least the difference between each anomaly index. That is, in this embodiment, the first value can be the average value of each anomaly index; the second value can be the variance or standard deviation of each anomaly index.

[0086] Step S222: Based on the first value and the second value, obtain the intention performance index.

[0087] In this embodiment, a larger first value indicates that every user operation is abnormal and cannot represent their true operational intent; a larger second value indicates that some of the user's operations are less abnormal, meaning that some of the user's operations can represent their true operational intent. Based on this, the intent performance index can be negatively correlated with the first value and positively correlated with the second value. For example, in step S222, based on the first and second values, the formula for calculating the intent performance index can be as follows:

[0088]

[0089] in, Indicator of intent; Indicates the first numerical value; Indicates the second numerical value; Zero-prevention coefficient, used to avoid a denominator of 0. It can be any value close to 0, such as 0.01 or 0.001. Alternatively, in step S222, based on the first value and the second value, the calculation formula for the intention performance index can be as follows:

[0090]

[0091] in, Indicator of intent; Indicates the first numerical value; Indicates the second numerical value; This represents an exponential function with the natural constant e as its base.

[0092] In this embodiment, the larger the intention expression index, the more the interaction between the first user and the hospital information service platform reflects the true intention of the first user; the smaller the intention expression index, the less the interaction between the first user and the hospital information service platform reflects the true intention of the first user.

[0093] In this embodiment, each operation module interface includes a parent operation module interface and a sub-operation module interface, where the sub-operation module interface is at a lower level than the corresponding parent operation module interface. For example, taking the first user as a patient, after logging into the hospital information service platform, the first user's operation interface (hereinafter referred to as the first operation interface) displays "Outpatient Registration," "Payment," and "Report Inquiry," etc. If "Outpatient Registration" is clicked, the user enters the outpatient registration module interface (hereinafter referred to as the second operation interface). The outpatient registration module interface displays "Internal Medicine," "Surgery," and "ENT," etc. If "Internal Medicine" is clicked, the user enters the internal medicine registration module interface (hereinafter referred to as the third operation interface). In this embodiment, for the first and second operation interfaces, the first operation interface is the parent operation module interface of the second operation interface, and the second operation interface is a sub-operation module interface of the first operation interface; for the second and third operation interfaces, the second operation interface is the parent operation module interface of the third operation interface, and the third operation interface is a sub-operation module interface of the second operation interface. That is to say, in this embodiment, the parent operation module interface is the higher-level operation interaction entry point in the hospital information service platform, carrying core functional categories. The sub-operation module interface is an interactive interface in the hospital information service platform that is at a lower level than the corresponding parent operation module interface and provides a more detailed extension of the parent interface's functions. In this embodiment, numbers can be used to represent the levels of the corresponding operation module interfaces. For example, the level number corresponding to the first operation interface can be 100; the level number corresponding to the second operation interface can be 20; the level number corresponding to the third operation interface can be 5, and so on. Further details will not be provided below.

[0094] It is important to note that different users or different operation processes can lead to differences in the time users spend on different operation module interfaces. For example, older users generally spend more time on the same module interface than younger users; the same user spends significantly more time on the information entry operation module interface than on the date selection operation module interface. If the accuracy of determining whether a user's interaction can represent the user's true operation intention is determined solely by the time spent on the interaction, its accuracy is relatively low. Therefore, after obtaining the intent performance index based on the first and second values ​​in step S222, the method further includes steps S230 to S260, as shown in Figure 3.

[0095] Step S230: Based on the first time-series interaction data, obtain the operation level time-series sequence.

[0096] In this embodiment, each sequence value in the operation level time sequence corresponds to the level of each operation module interface clicked by the first user.

[0097] It's important to note that if a user's actions are normal and without errors, the process will proceed sequentially according to the hierarchy of each operation module interface. For example, in a medical appointment process, the appointment information will be entered first, followed by the selection of the relevant department, without frequent retrievals from the sub-module interface to the parent module interface. This is because the parent module interface has a higher hierarchy than the sub-module interfaces. Therefore, if a user's actions are normal and without errors, the sequence values ​​in the corresponding operation hierarchy time sequence will gradually decrease. In other words, if the sequence values ​​in the operation hierarchy time sequence frequently increase, it indicates that the user's actions are abnormal.

[0098] Step S240: Based on the operation level time sequence, obtain multiple sequence strings.

[0099] In this embodiment, each sequence string represents a user's operation of returning from the sub-operation module interface to the parent operation module interface. Specifically, step S240 involves obtaining multiple sequence strings based on the operation level time sequence, including steps S241 to S243.

[0100] Step S241: Traverse the operation level time sequence to obtain the current first sequence value and the current second sequence value.

[0101] In this embodiment, the current second sequence value is greater than the current first sequence value. The current first sequence value and the current second sequence value are sequentially adjacent, and the current second sequence value is sequentially following the current first sequence value.

[0102] Step S242: Combine all sequence values ​​between the current first sequence value and the historical second sequence value, the current first sequence value, and the historical second sequence value into a sequence string.

[0103] In this embodiment, the acquisition time of the historical second sequence value is adjacent to the current second sequence value, and the current second sequence value is located after the historical second sequence value. The historical second sequence value includes an initial second sequence value, which is the first sequence value in the operation level time sequence.

[0104] Step S243: After the operation level time sequence traversal is completed, obtain multiple sequence strings.

[0105] In this embodiment, the number of sequence strings indicates how many times the user (i.e., the first user) returned to the parent operation module interface from the sub-operation module interface. In other words, the more sequence strings there are, the more abnormal the user's interaction with the hospital information service platform is, meaning the interaction between the user and the hospital information service platform is less likely to reflect the user's true operational intentions.

[0106] Step S250: Obtain correction coefficients based on each sequence string.

[0107] In this embodiment, the correction coefficient is used at least to characterize the quantity of each sequence string. In this embodiment, the correction coefficient may be negatively correlated with the quantity of each sequence string. For example, step S250, based on each sequence string, obtains the correction coefficient, including steps S251 to S254.

[0108] Step S251: Based on each sequence string, obtain multiple clusters.

[0109] In this embodiment, the parent operation module interface corresponding to each sequence string in each cluster has the same level.

[0110] Step S252: Obtain the first quantity based on each cluster.

[0111] In this embodiment, the first quantity refers to the number of sequence strings in the first cluster; the first cluster is the cluster with the largest number of corresponding sequence strings among all clusters. It is important to note that the first quantity represents the number of times a user returns to a specific core function (e.g., "outpatient registration"), indicating that the user is only unfamiliar with that function (e.g., unable to find the entry point for the department they need to register for), rather than being confused about all functions. That is, the larger the first quantity, the more concentrated the user's return operations (confusion only with a single function), and the higher the probability that their overall operation still reflects their true intention (e.g., the user clearly wants to register, but is repeatedly trying within the "outpatient registration" function). Therefore, the intention performance index should be "less weakened" when calculating the correction coefficient later.

[0112] Step S253: Obtain the second quantity based on each sequence string.

[0113] In this embodiment, the second quantity is the total number of all sequence strings.

[0114] Step S254: Obtain the correction coefficient based on the first quantity and the second quantity.

[0115] In this embodiment, the correction coefficient is positively correlated with the first quantity and negatively correlated with the second quantity. The calculation method for the correction coefficient in this embodiment can be referred to in steps S213 and S222, and will not be repeated here.

[0116] Step S260: Correct the intention performance index based on the correction coefficient.

[0117] In this embodiment, the intent performance index can be corrected based on the correction coefficient in any reasonable way. For example, in step S260, the calculation formula for correcting the intent performance index based on the correction coefficient can be as follows:

[0118]

[0119] in, This indicates the performance index of the intended outcome after the correction. This indicates a correction to the previous intention performance index; This represents the correction coefficient. Alternatively, in step S260, the formula for calculating the correction of the intention performance index based on the correction coefficient can be as follows:

[0120]

[0121] in, This indicates the performance index of the intended outcome after the correction. This indicates a correction to the previous intention performance index; Indicates the correction factor; This indicates that the normalization function is used to normalize the values ​​within the parentheses to the range [0, 1].

[0122] Step S300: If the intent performance index is greater than the first preset value, then multiple second time-series interaction data are obtained based on the hospital information service platform.

[0123] In this embodiment, the first preset value can be set according to requirements, for example, the first preset value can be 0.7 or 0.8, etc. In this embodiment, the second time-series interaction data is the historical time-series interaction data of any user other than the first user in the hospital information service platform. That is to say, the second time-series interaction data includes the time-series interaction data of other users (i.e., users other than the first user) and the completion of related operations.

[0124] Step S400: Generate operation instructions for the first user based on each of the second time-series interaction data.

[0125] In the embodiments of this application, the operation instructions for the first user can be generated based on each of the second time-series interaction data in any reasonable manner. For example, the operation instructions for the first user can be generated based on each of the second time-series interaction data as shown in Embodiment 1 and Embodiment 2 below.

[0126] Example 1 of generating operation instructions

[0127] In this embodiment, step S400 generates operation guidance for the first user based on each second time-series interaction data, including steps S410 and S420.

[0128] Step S410: Obtain the third time series interaction data based on each of the second time series interaction data.

[0129] In this embodiment, the third temporal interaction data is the temporal interaction data with the highest similarity to the first temporal interaction data among all the second temporal interaction data. It should be noted that in the field of computer science, calculating the similarity between two data (i.e., the first temporal interaction data and the second temporal interaction data) is a mature technology, which will not be elaborated here.

[0130] Step S420: Based on the third time-series interaction data, generate operation instructions for the first user.

[0131] As mentioned above, since the third time-series interaction data is the interaction data most similar to the first time-series interaction data among all the second time-series interaction data, operation guidance for the first user can be generated based on the third time-series interaction data. For example, if the third time-series interaction data has a total of X operation steps, and the first user's current operation step is the Y-th operation step, where X is greater than Y, and both X and Y are positive integers, then operation guidance can be generated for the first user's next operation step according to the (Y+1)-th step of the third time-series interaction data. In hospital information service platforms, generating corresponding operation guidance based on certain information (i.e., the third time-series interaction data) is a mature technology, and will not be elaborated here.

[0132] Example 2 of generating operation instructions

[0133] It should be noted that Embodiment 1 relies solely on the single time-series interaction data (i.e., the third time-series interaction data) with the highest similarity to the first user's time-series interaction data to generate guidance. This single time-series interaction data is easily affected by accidental factors. For example, the operation of the most similar user may include personalized abnormal behaviors such as accidental clicks or repeated backtracking. Guidance generated based on such data lacks universal applicability and can only match the special operating habits of a single user, making it difficult to cover the operating patterns of most normal users. Therefore, in this embodiment, step S400, based on each second time-series interaction data, generates the operation guidance for the first user, including steps S430 to S470.

[0134] Step S430: Based on each second time-series interaction data, obtain multiple fourth time-series interaction data.

[0135] In this embodiment, the fourth temporal interaction data is any temporal interaction data among the various second temporal interaction data that has a similarity to the first temporal interaction data greater than or equal to a second preset value.

[0136] In this embodiment, the second preset value can be a fixed value, such as 0.8 or 0.9. To avoid the inability to obtain the fourth time-series interaction data when the number of second time-series interaction data is small, the second preset value can also be dynamically changed in this embodiment. For example, the second preset value can be 0.8 times or 0.9 times the maximum similarity between each second time-series interaction data and the first time-series interaction data. Alternatively, the similarity between each second time-series interaction data and the first time-series interaction data can be arranged in descending order, and the average of the first 10% or 20% of similarity can be taken as the second preset value.

[0137] Step S440: Generate multiple frequent item sets from each of the fourth time series interaction data based on the frequent pattern growth algorithm.

[0138] In this embodiment, the frequent item set is the operation sequence of the operation module interface combination that satisfies the preset minimum support in each of the fourth time-series interaction data. It should be noted that the frequent pattern growth algorithm, which obtains the operation sequence of the operation module interface combination that satisfies the preset minimum support from each of the fourth time-series interaction data, is a mature technology and will not be elaborated upon here.

[0139] Step S450: Determine the current operation status of the first user based on the first time-series interaction data.

[0140] In this embodiment, the current operation state includes at least the operation module interface currently being operated by the first user. For example, the user may be on the payment module interface because they want to pay registration fees, examination fees, or medication fees.

[0141] Step S460: Select a target set of frequent items that matches the current operation state from the multiple sets of frequent items.

[0142] In this embodiment, the target frequent item set is any set of frequent items that includes the current operation state of the first user. In this embodiment, any reasonable method can be used to filter out the target frequent item set that matches the current operation state from multiple sets of frequent items. For example, one frequent item set with the current operation state can be randomly selected from each set of frequent items as the target frequent item set. In order to obtain the target frequent item set for generating accurate operation guidance for the first user, in this embodiment, step S460, filtering out the target frequent item set that matches the current operation state from multiple sets of frequent items, includes steps S461 to S463.

[0143] Step S461: Based on each frequent item set, obtain multiple first frequent item sets.

[0144] In this embodiment, the first frequent item set is any frequent item set that has the current operation state among the various frequent item sets.

[0145] Step S462: Based on each first frequent item set, obtain multiple second frequent item sets.

[0146] In this embodiment, the second set of frequent items is any set of frequent items from each of the first sets of frequent items whose similarity to the historical operation state of the first user is greater than a third preset value. The third preset value can be set according to requirements, for example, the third preset value can be 0.8 or 0.9, etc.

[0147] Step S463: Obtain the target frequent item set based on each of the second frequent item sets.

[0148] In this embodiment, the target frequent item set can be obtained based on each of the second frequent item sets using any reasonable method. For example, the second frequent item set with the highest similarity to the user's historical operation state among the first frequent item sets can be selected as the target frequent item set. Alternatively, a third frequent item set can be selected as the target frequent item set, where the third frequent item set is the frequent item set with the highest similarity to all frequent item sets except the third frequent item set.

[0149] Step S470: Based on the target frequent item set, generate operation instructions for the first user.

[0150] It is important to understand that generating corresponding operation guidelines based on certain information (i.e., the set of frequently occurring target items) in hospital information service platforms is a mature technology, which will not be elaborated on here.

[0151] In this embodiment, multiple fourth-series interaction data that meet the second preset similarity value are first selected to expand the range of reference samples and avoid the limitations of single data. Then, a frequent item set (i.e., the operation module interface sequence that is common to most similar users and meets the minimum support) is extracted from multiple sets of data through a frequent pattern growth algorithm to ensure that the generated operation guidance is based on universally applicable operation logic. Finally, the target frequent item set is selected by combining the current operation state of the first user, which can accurately match the process node where the user is in real time, and further improve the scenario fit of the generated operation guidance.

[0152] This concludes the description of Example 2 for generating operation instructions.

[0153] The embodiment of the collaborative screen information dynamic aggregation method that integrates user intent prediction proposed in this application obtains first-sequence interaction data of the first user, including the time sequence of clicks and dwell time on the interfaces of each operation module, from a pre-established hospital information service platform. It then calculates an intent performance index, which characterizes the likelihood that the user's operation reflects their true intent, based on the first-sequence interaction data. Operation guidance is generated only when the intent performance index is greater than a first preset value (i.e., the user's intent is clear), by calling historical second-sequence interaction data from other users within the platform. This effectively solves the technical problem in traditional hospital information display where operation guidance deviates from the user's actual operational needs due to scenario-centric, fixed-process push notifications. It achieves accurate intent matching and dynamic guidance based on the user's actual operational behavior, ensuring the accuracy and practicality of information interaction in medical scenarios, avoiding useless guidance interference, and meeting the hospital's information interaction needs of "multi-role, multi-scenario, and high-timeliness."

[0154] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0155] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the methods, apparatuses, and devices described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0156] In the several embodiments provided in this application, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between devices or modules may be electrical, mechanical, or other forms.

[0157] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0158] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0159] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0160] The computer program product includes one or more computer instructions. When the computer program is loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., digital video optical disc), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0161] Although embodiments of this application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles of this application.

Claims

1. A method for dynamically aggregating collaborative screen information based on user intent prediction, characterized in that, include: Based on the hospital information service platform, obtain the first time-series interaction data of the first user; The hospital information service platform is pre-established and includes multiple operation module interfaces; The first user can be any user in the hospital information service platform; the first time-series interaction data includes at least the first user's click information and dwell time information on each operation module interface in time sequence. Based on the first time-series interaction data, an intent performance index is obtained; the intent performance index is at least used to characterize the likelihood that the interaction operation of the first user reflects the true intent of the first user; if the intent performance index is greater than a first preset value, then multiple second time-series interaction data are obtained based on the hospital information service platform; the second time-series interaction data are the historical time-series interaction data of any user other than the first user in the hospital information service platform. Based on the various second-series interaction data, the operation instructions for the first user are generated; The step of obtaining the intent performance index based on the first time-series interaction data includes: obtaining an anomaly index for each operation module interface based on the first time-series interaction data; the anomaly index is used to characterize at least the number of times the first user clicks the corresponding operation module interface and the length of time the first user stays on the corresponding operation module interface; obtaining the intent performance index based on each anomaly index; the step of obtaining the anomaly index for each operation module interface based on the first time-series interaction data includes: obtaining a first operation module interface based on each operation module interface; the first operation module interface is the operation module interface that the first user has clicked in the first time-series interaction data, and the current time sequence needs to obtain the anomaly index corresponding to the first operation module interface; obtaining the number of clicks and the dwell time based on the first time-series interaction data; the number of clicks is the number of times the first user clicks the first operation module interface; the dwell time is the average dwell time after the first user clicks the first operation module interface multiple times; obtaining the anomaly index for the first operation module interface based on the number of clicks and the dwell time; the anomaly index is positively correlated with the number of clicks and negatively correlated with the dwell time.

2. The method for dynamic aggregation of collaborative screen information based on user intent prediction according to claim 1, characterized in that, The step of obtaining the intent performance index based on each abnormal index includes: obtaining a first value and a second value based on each abnormal index; the first value is used to at least characterize the average value of each abnormal index; the second value is used to at least characterize the difference between each abnormal index; obtaining the intent performance index based on the first value and the second value; the intent performance index is negatively correlated with the first value and positively correlated with the second value.

3. The collaborative screen information dynamic aggregation method based on user intent prediction according to claim 2, characterized in that, Each operation module interface includes a parent operation module interface and a child operation module interface, wherein the level of the child operation module interface is lower than the level of the corresponding parent operation module interface; after obtaining the intent performance index based on the first value and the second value, the method further includes: obtaining an operation level time sequence based on the first time-series interaction data; each sequence value in the operation level time sequence is the level corresponding to each operation module interface clicked by the first user; obtaining multiple sequence strings based on the operation level time sequence; obtaining a correction coefficient based on each sequence string; the correction coefficient is used to characterize at least the quantity of each sequence string; and correcting the intent performance index based on the correction coefficient.

4. The method for dynamic aggregation of collaborative screen information based on user intent prediction according to claim 3, characterized in that, The step of obtaining multiple sequence strings based on the operation level time sequence includes: traversing the operation level time sequence to obtain a current first sequence value and a current second sequence value; the current second sequence value is greater than the current first sequence value; the current first sequence value and the current second sequence value are temporally adjacent, and the current second sequence value is located after the temporal sequence of the current first sequence value; forming a sequence string by combining all sequence values ​​between the current first sequence value and a historical second sequence value, the current first sequence value, and the historical second sequence value; the acquisition time of the historical second sequence value is adjacent to the current second sequence value, and the current second sequence value is located after the temporal sequence of the historical second sequence value; the historical second sequence value includes an initial second sequence value, which is the first sequence value in the operation level time sequence; until the operation level time sequence is traversed, multiple sequence strings are obtained.

5. The method for dynamic aggregation of collaborative screen information based on user intent prediction according to claim 3, characterized in that, The step of obtaining the correction coefficient based on each sequence string includes: obtaining multiple clusters based on each sequence string; the parent operation module interface corresponding to each sequence string in each cluster has the same level; obtaining a first quantity based on each cluster; the first quantity is the number of sequence strings in the first cluster; the first cluster is the cluster with the largest number of corresponding sequence strings among all clusters; obtaining a second quantity based on each sequence string; the second quantity is the total number of each sequence string; obtaining a correction coefficient based on the first quantity and the second quantity; the correction coefficient is positively correlated with the first quantity and negatively correlated with the second quantity.

6. The method for dynamic aggregation of collaborative screen information based on user intent prediction according to any one of claims 1 to 5, characterized in that, The step of generating operation guidance for the first user based on each second time-series interaction data includes: obtaining third time-series interaction data based on each second time-series interaction data; the third time-series interaction data being the time-series interaction data with the highest similarity to the first time-series interaction data among the second time-series interaction data; and generating operation guidance for the first user based on the third time-series interaction data.

7. The method for dynamic aggregation of collaborative screen information based on user intent prediction according to any one of claims 1 to 5, characterized in that, The step of generating operation guidance for the first user based on each second time-series interaction data includes: acquiring multiple fourth time-series interaction data based on each second time-series interaction data; the fourth time-series interaction data being any time-series interaction data among the second time-series interaction data whose similarity to the first time-series interaction data is greater than or equal to a second preset value; generating multiple frequent item sets from each fourth time-series interaction data based on a frequent pattern growth algorithm; the frequent item set being an operation sequence of operation module interface combinations among each fourth time-series interaction data that satisfies a preset minimum support; determining the current operation state of the first user based on the first time-series interaction data; the current operation state including at least the operation module interface currently operated by the first user; filtering out a target frequent item set matching the current operation state from the multiple frequent item sets; and generating operation guidance for the first user based on the target frequent item set.

8. The method for dynamic aggregation of collaborative screen information based on user intent prediction according to claim 7, characterized in that, The step of selecting a target frequent item set that matches the current operation state from multiple frequent item sets includes: obtaining multiple first frequent item sets based on each frequent item set; the first frequent item set being any frequent item set among the frequent item sets that has the current operation state; obtaining multiple second frequent item sets based on each first frequent item set; the second frequent item set being any frequent item set among the first frequent item sets whose similarity to the historical operation state of the first user is greater than a third preset value; and obtaining the target frequent item set based on each second frequent item set.

Images