An intelligent interaction method and system for improving in-hospital patient service and information collaboration

Abstract

The present application relates to the technical field of patient service management, and specifically includes an intelligent interaction method and system for improving in-hospital patient service and information collaboration, comprising: constructing a three-dimensional feature vector of in-hospital service state, deploying a demand guiding response strategy, setting a double-delay gradient under a service collaboration framework, combining with a visit constraint condition, loading the optimized interaction response parameters to a terminal cluster, configuring a communication interaction protocol and a collaborative semantic mapping rule corresponding to an information collaboration network. The present application solves the technical problems of existing service response mode solidification, difficulty in accurately mining the coupling relationship between patient core demand and medical resource supply, and resource scheduling decision lagging behind service dynamic changes, and achieves the technical effects of constructing a three-dimensional feature vector containing patients, medical staff and environment, extracting the coupling relationship between patient core demand features and medical resource supply features, dynamically deploying a demand guiding strategy and collecting interaction parameters, and improving the response efficiency and information collaboration reliability of in-hospital service.

Description

Technical Field

[0001] This invention relates to the field of patient service management technology, specifically to an intelligent interactive method and system for improving in-hospital patient services and information collaboration. Background Technology

[0002] The aging population and the increasing number of patients with chronic diseases have led to a continuous rise in outpatient traffic in hospitals, and the demand for multi-departmental collaboration and cross-scenario services is becoming more frequent. However, the current in-hospital patient service and information collaboration mechanisms are imperfect and cannot adapt to the service needs under digital transformation. Existing service carriers such as self-service terminals and mobile applications mostly adopt fixed interaction modes, resulting in lagging resource scheduling and insufficient data processing capabilities at the edge. This makes it difficult to achieve continuous improvement in service capabilities while ensuring data security, thus restricting the improvement of in-hospital service quality and information collaboration efficiency.

[0003] In summary, existing technologies suffer from several technical problems, including rigid service response models with poor adaptability, difficulty in accurately identifying the coupling relationship between patients' core needs and the supply of medical resources, and resource scheduling decisions lagging behind dynamic changes in services. Summary of the Invention

[0004] This application provides an intelligent interactive method and system to improve in-hospital patient services and information collaboration. It aims to solve the technical problems in the existing technology, such as the rigid and poorly adaptable service response model, the difficulty in accurately identifying the coupling relationship between the core needs of patients and the supply of medical and nursing resources, and the lag of resource scheduling decisions behind the dynamic changes in services.

[0005] In view of the above problems, the technical solution to achieve the present application is as follows:

[0006] In a first aspect, this application provides an intelligent interaction method to improve in-hospital patient services and information collaboration. The method includes: collecting patient visit information, medical resource status, and in-hospital environmental parameters to construct a three-dimensional feature vector of in-hospital service status; deploying a demand-guided response strategy based on the three-dimensional feature vector of in-hospital service status, tracking patient interaction behavior, and collecting interaction response parameters; aligning visit time-series data with patient service log streams at in-hospital edge computing nodes, extracting the coupling relationship between core patient demand features and medical resource supply features, and setting a dual-delay gradient under a service collaboration framework in conjunction with the demand-guided response strategy; dynamically optimizing the interaction response parameters based on the dual-delay gradient under the service collaboration framework and in conjunction with visit constraints; loading the optimized interaction response parameters to a terminal cluster; and configuring the communication interaction protocol and collaborative semantic mapping rules corresponding to the information collaboration network.

[0007] Preferably, the generator built into the service collaboration framework is used to simulate the interaction response sequence under the standard medical treatment process; the discriminator built into the service collaboration framework is used to identify service abnormal deviation events by comparing the distribution differences between the interaction response parameters of the in-hospital patient service interface and the interaction response sequence.

[0008] Preferably, a patient behavior dynamics analysis unit is set up. The patient behavior dynamics analysis unit is used to extract the task progress rate and operation path deviation from the multimodal interaction log; track the patient's cross-device service jump trajectory between self-service terminals, mobile application service interfaces and ward screens, and extract the waiting tolerance threshold of key nodes; construct a service load profile based on the task progress rate and operation path deviation, combined with the waiting tolerance threshold, and dynamically adjust the content push frequency, information density and guidance dialogue complexity in the interaction response parameters based on the service load profile.

[0009] Preferably, a correlation mapping model between patient behavior characteristics and the intensity of medical resource occupancy is established. When the frequency of task interruption or rollback operation is detected to suddenly increase beyond a preset safety threshold, a soothing intervention strategy in the waiting area is automatically triggered. At the same time, a federated edge learning architecture is adopted. Under the condition that the edge data is encrypted and the original log does not leave the domain, the interaction satisfaction scores and behavioral residual data of patients in multiple wards are aggregated, and the attention mechanism weight parameters of the patient behavior dynamics analysis unit are iteratively optimized.

[0010] Preferably, the time window denoising parameters of the patient service log stream are dynamically adjusted according to outpatient traffic, and a service event synchronization control node is embedded in the log preprocessing stage; at the hospital edge computing node, the service event synchronization control node is used to align the registration completion time, examination appointment trigger time and the terminal's first interaction response start time.

[0011] Preferably, a dynamic dimensionality reduction encoder is deployed to adaptively select the feature dimensionality reduction dimension based on the patient's historical visit frequency, patient visit type, and disease chronicity. At the same time, the baseline interaction template library is updated, and if a service request combination with multiple overlapping items across departments and dynamically escalating urgency is detected, online reconstruction of the service semantic feature space is triggered.

[0012] Preferably, an indoor positioning device is used to obtain the patient's real-time movement trajectory within the hospital; the real-time movement trajectory is cross-correlation analysis is performed with the service request timing, and by locking the arrival times of key service nodes, including the examination room door and the pharmacy window, collaborative timing instructions for multi-terminal content push are dynamically set.

[0013] Preferably, when a spatiotemporal misalignment is detected between the patient's location status and the terminal service response, a timing compensation strategy is activated to perform projection correction on the three-dimensional feature vector of the in-hospital service status.

[0014] Preferably, under load service conditions, the collection granularity and reporting frequency of the patient service log stream are dynamically determined based on the remaining power of the equipment; when the reconstruction residual of the three-dimensional feature vector of the in-hospital service status exceeds the dynamic threshold jointly set based on the power status and service criticality level, the sparse interactive sampling mode at the corresponding collection granularity is activated; in the information collaboration network, under the joint optimization objective of terminal energy consumption constraints and service quality assurance index, a piecewise linear scheduling strategy is executed using the sparse interactive sampling mode.

[0015] A second aspect of this application provides an intelligent interactive system for improving in-hospital patient services and information collaboration. The system includes: a three-dimensional feature vector construction module: collecting patient visit information, medical resource status, and in-hospital environmental parameters to construct a three-dimensional feature vector of in-hospital service status; an interaction response parameter collection module: based on the three-dimensional feature vector of the in-hospital service status, deploying a demand-guided response strategy, tracking patient interaction behavior, and collecting interaction response parameters; a coupling relationship extraction module: at in-hospital edge computing nodes, aligning visit time-series data with patient service log streams, extracting the coupling relationship between core patient demand features and medical resource supply features, and setting a dual-delay gradient under the service collaboration framework in conjunction with the demand-guided response strategy; and a terminal cluster loading module: based on the dual-delay gradient under the service collaboration framework and combined with visit constraints, dynamically optimizing the interaction response parameters, loading the optimized interaction response parameters to the terminal cluster, and configuring the communication interaction protocol and collaborative semantic mapping rules corresponding to the information collaboration network.

[0016] In summary, one or more technical solutions provided in this application achieve the technical effect of constructing a three-dimensional feature vector containing patients, medical staff, and environment, extracting the coupling relationship between the core needs of patients and the supply characteristics of medical resources, dynamically deploying demand guidance strategies and collecting interaction parameters, thereby improving the response efficiency of in-hospital services and the reliability of information collaboration. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0018] Figure 1 This application provides a flowchart illustrating an intelligent interactive method for improving in-hospital patient services and information collaboration.

[0019] Figure 2 This application provides a schematic diagram of the structure of an intelligent interactive system that enhances in-hospital patient services and information collaboration.

[0020] Figure labeling: 3D feature vector construction module M100, interactive response parameter collection module M200, coupling relationship extraction module M300, terminal cluster loading module M400. Detailed Implementation

[0021] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0022] Example 1: The present application will be described in detail below with reference to the accompanying drawings, as follows... Figure 1 As shown, this application provides an intelligent interaction method for improving in-hospital patient services and information collaboration, wherein the method includes:

[0023] S1: Collect patient visit information, medical and nursing resource status, and hospital environment parameters to construct a three-dimensional feature vector of hospital service status; S2: Based on the three-dimensional feature vector of hospital service status, deploy demand-guided response strategies, track patient interaction behavior, and collect interaction response parameters.

[0024] Specifically, a three-dimensional feature vector is a vector that integrates data from three different dimensions: patient visit information, medical resource status, and hospital environment parameters. It can comprehensively reflect the multifaceted characteristics of hospital service status and provide a comprehensive data foundation for subsequent service optimization. A demand-guided response strategy is a strategy that dynamically adjusts service responses based on patient needs. By analyzing patients' real-time needs and behavioral patterns, it guides the system to provide more personalized and precise service responses, thereby improving service adaptability and satisfaction. Interaction response parameters refer to various parameters of the system response during patient-system interaction, such as response time, information push frequency, and guidance language. These parameters reflect the quality and efficiency of service response and are important indicators for measuring service effectiveness.

[0025] Execution steps: By collecting patient visit information, medical resource status, and hospital environmental parameters, a three-dimensional feature vector of hospital service status is constructed. Patient visit information includes medical records, examination reports, visit history, and feedback. Medical resource status includes staff scheduling, equipment usage, and equipment occupancy rate. Hospital environmental parameters include waiting area population density and department distribution. Patient interaction behavior includes consultation frequency and question type. Multi-source data fusion technology is used to integrate scattered data to comprehensively reflect the hospital service status. The congestion level of the waiting area is acquired in real time through a sensor network, and combined with patients' electronic medical records and medical staff scheduling information, a three-dimensional feature vector is constructed.

[0026] Based on the constructed three-dimensional feature vector, a demand-guided response strategy is deployed. This strategy can dynamically adjust service responses according to patients' specific needs and behavioral patterns. Furthermore, when it is detected that a patient is waiting for too long in the waiting area and is emotionally anxious, the response strategy will be automatically adjusted to prioritize pushing waiting information and reassuring prompts to them. At the same time, the patient's interaction behavior is tracked, and interaction response parameters are collected, including response time and information push frequency. The collection of interaction response parameters provides data support for subsequent service optimization. Through the above steps, changes in patient needs can be perceived in real time, and service responses can be dynamically adjusted, thereby improving the accuracy and satisfaction of services.

[0027] S3: At the edge computing node within the hospital, align the time-series data of patient visits with the patient service log stream, extract the coupling relationship between the core needs of patients and the supply characteristics of medical and nursing resources, and set a dual-delay gradient under the service collaboration framework in conjunction with the demand-guided response strategy; S4: Based on the dual-delay gradient under the service collaboration framework and in conjunction with the constraints of patient visits, dynamically optimize the interaction response parameters, load the optimized interaction response parameters into the terminal cluster, and configure the communication interaction protocol and collaborative semantic mapping rules corresponding to the information collaboration network.

[0028] Specifically, edge computing nodes refer to computing nodes deployed at the edge, close to the data source or where data processing needs arise, to enable local data processing and analysis, reduce data transmission latency, and improve system real-time performance and response speed; medical appointment time-series data refers to data generated chronologically during a patient's medical treatment process, including registration time, appointment time, and consultation time, reflecting the patient's medical process and time nodes; patient service log streams refer to detailed logs of patient-service interactions recorded by the system, including operation time, operation content, and response time, used to analyze patient behavior patterns and service response; coupling relationship refers to the relationship between the core needs of patients and the supply of medical resources. The interrelationships and dependencies between features are analyzed to optimize resource allocation and service response. Dual delay gradients refer to two delay control strategies set for different service links and resource allocations within the service collaboration framework. These strategies reflect the dynamic adaptation lag of patient demand response delay within the service interaction scale and the supply matching lag of medical resource scheduling delay within the system collaboration scale. The dual delay gradients correspond to demand response delay and information synchronization delay, respectively, and are used to balance service response speed and resource scheduling accuracy. An information collaboration network refers to a network architecture that enables information sharing and collaborative work among multiple terminals and service nodes through communication interaction protocols and collaborative semantic mapping rules.

[0029] Execution Steps: By aligning the patient visit time-series data with the patient service log stream at the hospital's edge computing nodes, the coupling relationship between the core needs of patients and the supply characteristics of medical resources can be accurately extracted. For example, analyzing the waiting time between patient registration and examination appointment, as well as the resource allocation of medical staff during this time period, reveals the correlation between the two. This alignment and analysis process utilizes time series analysis and data mining techniques, enabling the processing of large amounts of data in a short time and providing real-time support for service optimization. Combined with a demand-driven response strategy, a dual-delay gradient is set under the service collaboration framework. Furthermore, a shorter delay gradient is set during the patient's waiting period after registration to quickly respond to the patient's needs; a longer delay gradient is set during the examination result analysis phase to ensure the accuracy of resource allocation. Through this dual-delay gradient setting, dynamic balance can be achieved between different service stages, improving overall service efficiency.

[0030] Based on dual-delay gradients and consultation constraints, including patient urgency and departmental resource limitations, interactive response parameters are dynamically optimized and loaded into the terminal cluster. For example, while patients are waiting for examinations, the frequency and content of information pushes are dynamically adjusted according to the optimized parameters to reduce patient anxiety. Simultaneously, based on heterogeneous terminals such as AI guidance screens in the outpatient hall, bedside interactive terminals in wards, mobile nursing PDAs (Personal Digital Assistants), pharmacy medication reminder screens, and in-hospital navigation robots, an information collaboration network is established for each heterogeneous terminal. Furthermore, the mobile nursing PDA is a small handheld device integrating data collection, information entry, barcode scanning, and wireless communication functions. The communication interaction protocol and collaborative semantic mapping rules of the information collaboration network are configured to ensure information consistency and collaboration between different terminals and service nodes. For example, through unified semantic mapping rules, examination results are quickly synchronized to patients' mobile applications and medical staff's terminals, avoiding information lag and errors, and significantly improving the response efficiency and information collaboration capabilities of in-hospital services.

[0031] Furthermore, the method of this application also includes:

[0032] The generator built into the service collaboration framework is used to simulate the interaction response sequence under the standard medical treatment process; the discriminator built into the service collaboration framework is used to identify service abnormal deviation events by comparing the distribution differences between the interaction response parameters of the in-hospital patient service interface and the interaction response sequence.

[0033] Specifically, the service collaboration framework is used to integrate and coordinate different service links and resources within the hospital to achieve efficient service response and collaborative work. Within this framework, the generator is an algorithm module used to simulate the patient interaction response sequence under a standard medical process. By generating interaction patterns that conform to the conventional process, it provides a benchmark reference for the system. The discriminator is another algorithm module in the service collaboration framework, used to compare the distribution differences between the actual interaction response parameters and the standard interaction response sequence simulated by the generator. Through this comparison, the discriminator can identify service deviation events, i.e., deviations between the actual interaction and the standard process. An interaction response sequence refers to the sequence of interactive behaviors performed by the patient and the system in a preset order within a standard medical process, such as the interaction responses in registration, waiting, examination, and medication dispensing. Distribution differences refer to the statistical differences between the actual interaction response parameters and the standard interaction response sequence, usually measured by statistical indicators such as probability distribution, mean, and variance. Service deviation events refer to events where the actual interaction response deviates significantly from the standard process, indicating problems such as service interruption, response delay, or poor patient experience.

[0034] Execution Steps: The generator built into the service collaboration framework provides an ideal interaction pattern for the system by simulating the interaction response sequence under the standard medical process. For example, the generator can simulate the standard interaction sequence of a patient after registration, including appointment for examination, waiting, consultation, and medication pickup, based on the hospital's routine procedures. The generator's output is a sequence containing timestamps and interaction content, such as: registration 08:00, appointment for examination 08:10, waiting for consultation 08:20, consultation 08:40, medication pickup 08:50. The discriminator is used to monitor the interaction response parameters of the patient service interface in real time and compare them with the standard interaction response sequence simulated by the generator. For example, if the system records that a patient's registration time is 08:00, but the appointment time for examination is delayed to 08:30 and the waiting time is extended to 08:50, this delay is significantly different from the standard sequence simulated by the generator. The discriminator identifies service deviation events by calculating this distribution difference.

[0035] By working together with the generator and discriminator, abnormal situations in the service process can be detected in real time, and timely measures can be taken to make adjustments. For example, when the discriminator detects that the waiting time is too long, it can automatically adjust the guidance strategy in the waiting area, increase the allocation of medical staff, or push reassuring information to patients, thereby improving the stability and reliability of patient services. By introducing the generator and discriminator, the recognition rate of abnormal service events is improved, thus enhancing the overall medical experience of patients.

[0036] Furthermore, in addition to tracking patient interactions and collecting interaction response parameters, the method of this application also includes:

[0037] A patient behavior dynamics analysis unit is set up, which is used to extract task progress rate and operation path deviation from multimodal interaction logs; track the patient's cross-device service jump trajectory between self-service terminals, mobile application service interfaces and ward screens, and extract the waiting tolerance threshold of key nodes; construct a service load profile based on the task progress rate and operation path deviation, combined with the waiting tolerance threshold, and dynamically adjust the content push frequency, information density and guidance dialogue complexity in the interaction response parameters based on the service load profile.

[0038] Specifically, the patient behavior dynamics analysis unit is used to analyze and interpret patients' behavioral patterns and dynamic changes during the medical treatment process. By processing multimodal interaction logs, it extracts key features of patient behavior, such as task progress rate and operation path deviation. Multimodal interaction logs refer to log data recording patients' interactive behaviors on different devices and service interfaces, including various information such as operation time, operation type, and dwell time. Different devices and service interfaces include self-service terminals, mobile applications, and ward screens. Task progress rate refers to the speed at which patients complete various tasks in the medical treatment process, such as the time interval from registration to examination appointment, reflecting the efficiency of patients in the medical treatment process.

[0039] Operation path deviation refers to the degree of deviation between the patient's actual operation path and the standard operation path, such as whether the patient frequently returns to the previous step or skips certain routine steps; waiting tolerance threshold refers to the longest waiting time that a patient can accept at different service nodes, including the waiting area and the entrance of the examination room, reflecting the patient's tolerance for waiting time; service load profile is a model that comprehensively describes the patient's behavioral characteristics and needs during the medical treatment process. By integrating information such as task progress rate, operation path deviation, and waiting tolerance threshold, it provides a basis for personalized service adjustments.

[0040] Execution steps: A patient behavior dynamics analysis unit is set up to process multimodal interaction logs, extracting task progress rate and operation path deviation. For example, by analyzing the operation logs of patients on self-service terminals, it was found that the average time from registration to examination appointment for a certain patient was 15 minutes. However, in actual operation, it took 25 minutes, indicating a high operation path deviation, suggesting that the patient may have encountered difficulties during the operation. At the same time, the cross-device service jump trajectory of patients between self-service terminals, mobile application service interfaces, and ward screens is tracked to extract the waiting tolerance threshold of key nodes. For example, it was found that the average waiting tolerance threshold of patients in the waiting area was 30 minutes. Beyond this time, patient satisfaction will decrease significantly.

[0041] Based on task progress rate, operation path deviation, and waiting tolerance threshold, a service load profile is constructed. For example, patients with slow task progress rate and high operation path deviation are identified as high-load patients; patients whose waiting time is close to the tolerance threshold are marked as high-anxiety patients. According to the service load profile, interaction response parameters, such as content push frequency, information density, and guidance script complexity, are dynamically adjusted. For example, for patients in a high-load state, the information push frequency is reduced and the guidance script is simplified to reduce the patient's information burden; for patients in a high-anxiety state, the push frequency of reassuring information is increased and the information density is improved to help patients quickly understand the waiting progress. This mechanism of dynamically adjusting service response based on patient behavior significantly improves the patient's medical experience.

[0042] Furthermore, the method of this application also includes:

[0043] A correlation mapping model between patient behavior characteristics and the intensity of medical resource occupancy is established. When the frequency of task interruption or rollback operation is detected to suddenly increase beyond the preset safety threshold, a soothing intervention strategy in the waiting area is automatically triggered. At the same time, a federated edge learning architecture is adopted. Under the condition that the edge data is encrypted and the original log does not leave the domain, the interaction satisfaction scores and behavioral residual data of patients in multiple wards are aggregated, and the attention mechanism weight parameters of the patient behavior dynamics analysis unit are iteratively optimized.

[0044] Specifically, the correlation mapping model between patient behavior characteristics and the intensity of medical and nursing resource occupancy is used to quantify the relationship between patient behavior characteristics and the intensity of medical and nursing resource occupancy. Patient behavior characteristics include task interruption frequency and rollback operations, while the intensity of medical and nursing resource occupancy includes the workload of medical staff and the frequency of equipment use. The correlation mapping model can predict the impact of patient behavior on medical and nursing resources. Task interruption frequency refers to the number of times a patient interrupts the current task during the medical treatment process for various reasons, such as interruption due to uncertain information when filling in medical records. Rollback operations refer to the behavior of patients returning to the previous step or canceling the current operation during the operation process, reflecting the patient's dissatisfaction during the operation process.

[0045] Reassurance intervention strategies refer to reassurance measures taken to alleviate patient anxiety in waiting areas or other service nodes, such as sending reassurance messages and providing immediate consultation services; Federated edge learning architecture is a distributed machine learning architecture that allows data processing and model training on local devices, including edge computing nodes in hospitals, while ensuring data security through encryption and privacy protection mechanisms to prevent raw log data from leaving the local machine; Behavioral residual data refers to the difference between the patient's actual behavior and the model's predicted behavior, reflecting the model's prediction error and can be used to optimize the model; Attention mechanism weight parameters refer to the parameters used in the patient behavior dynamics analysis unit to adjust the model's attention to different behavioral features. By optimizing these parameters, the model's ability to identify key features can be improved.

[0046] Execution steps: Establish a correlation mapping model between patient behavior characteristics and the intensity of medical resource occupancy. Through analysis, it was found that when the frequency of patient task interruption exceeds 3 times or the number of rollback operations suddenly increases to more than 5 times, the intensity of medical resource occupancy will increase significantly, because this usually means that the patient needs more guidance or assistance. Based on this correlation, a safety threshold is preset. When the patient's behavior is detected to exceed the threshold, the reassurance intervention strategy in the waiting area is automatically triggered. For example, when the system detects that a patient frequently interrupts the operation of the self-service terminal, the system will guide the patient to the manual service window, thereby alleviating the patient's anxiety and reducing the ineffective occupancy of medical resources.

[0047] Meanwhile, the system adopts a federated edge learning architecture. Under the condition that the edge data is encrypted and the original logs do not leave the domain, it aggregates the interaction satisfaction scores and behavioral residual data of patients from multiple wards. Through an encryption mechanism, the satisfaction scores and behavioral residual data of patients from each ward are summarized into a central model, which is used to iteratively optimize the attention mechanism weight parameters of the patient behavior dynamics analysis unit. In this way, it can continuously learn and adapt to the behavioral patterns of patients in different wards, improve the accuracy and robustness of the model, and optimize the allocation efficiency of medical resources.

[0048] Furthermore, at the hospital's edge computing nodes, the method for aligning visit time-series data with patient service log streams includes:

[0049] The time window denoising parameters of the patient service log stream are dynamically adjusted based on outpatient traffic, and a service event synchronization control node is embedded in the log preprocessing stage. At the hospital edge computing node, the service event synchronization control node is used to align the registration completion time, examination appointment trigger time, and the terminal's first interaction response start time.

[0050] Specifically, outpatient traffic refers to the number of patients received by a hospital's outpatient department within a certain period of time, reflecting the busyness of the hospital's outpatient department; time window denoising parameters refer to the size of the time window and related parameters used to filter noisy data when processing patient service log streams. By adjusting the time window, unnecessary data interference is removed, improving the efficiency and accuracy of data processing; service event synchronization control nodes are used to ensure the synchronization and coordination between different service events, including registration, examination appointments, and interactive responses, avoiding time deviations and data inconsistencies; alignment refers to the temporal calibration of service events occurring at different times to ensure consistency in the time series, facilitating subsequent analysis and processing.

[0051] Execution steps: The time window denoising parameters of the patient service log stream are dynamically adjusted based on outpatient traffic. When outpatient traffic is high, the time window is narrowed to reduce interference from noise data and ensure the real-time performance and accuracy of the log data. When outpatient traffic is low, the time window can be appropriately widened to obtain more comprehensive data. The dynamic adjustment mechanism can effectively cope with fluctuations in hospital outpatient traffic and improve the flexibility and adaptability of data processing.

[0052] During the log preprocessing stage, a service event synchronization control node is embedded. The role of this node is to ensure time consistency between different service events. For example, the system uses the synchronization control node to align the registration completion time, the examination appointment trigger time, and the initial interaction response time of the terminal. Specifically, when a patient completes registration, this time point is recorded as T1; when an examination appointment is triggered, it is recorded as T2; and when the patient first interacts with the terminal, it is recorded as T3. The synchronization control node ensures that T1, T2, and T3 are accurately aligned in time sequence, avoiding analysis errors caused by time discrepancies. In the above steps, based on the aligned time points of registration, examination appointments, and interaction responses, patient waiting times and process efficiency can be analyzed more accurately, improving the hospital's service efficiency.

[0053] Furthermore, the method of this application also includes:

[0054] Deploy a dynamic dimensionality reduction encoder that adaptively selects the feature dimensionality reduction dimension based on the patient's historical visit frequency, patient visit type, and disease chronicity. At the same time, update the baseline interaction template library. If a service request combination with multiple overlapping items across departments and dynamically escalating urgency is detected, online reconstruction of the service semantic feature space is triggered.

[0055] Specifically, a dynamic dimensionality reduction encoder refers to an algorithm module that can dynamically adjust the dimensionality reduction strategy based on the characteristics of the input data. By analyzing the importance and complexity of the data, it adaptively selects the dimensionality reduction dimension to reduce data redundancy and improve processing efficiency. Feature dimensionality reduction dimension refers to the number of data features retained during the dimensionality reduction process. By selecting an appropriate dimensionality reduction dimension, key information can be retained while reducing data complexity. The benchmark interaction template library is a database that stores standard interaction templates to guide the system's interaction response in different scenarios. It is usually built based on historical data and common service scenarios.

[0056] Cross-departmental multi-item overlay refers to situations where patients involve multiple departments and require multiple examinations or treatments during their medical treatment. This complex combination of service requests places higher demands on the system's response and resource allocation. Dynamic escalation of the urgency of the condition refers to situations where a patient's condition deteriorates or other emergencies occur during their medical treatment, requiring the system to adjust service priorities and response strategies in a timely manner. Online reconstruction of the service semantic feature space refers to dynamically adjusting the representation of service semantic features based on real-time data during system operation to better adapt to new service requests and scenarios.

[0057] Execution steps: Deploy a dynamic dimensionality reduction encoder that adaptively selects feature dimensionality reduction dimensions based on the patient's historical visit frequency, visit type, and disease chronicity. For example, for patients with chronic diseases who visit frequently, more feature dimensions related to long-term disease management will be retained; for patients with acute diseases visiting for the first time, the focus will be on retaining features related to emergency diagnosis and treatment. In this way, patient data can be processed more efficiently, redundant information can be reduced, and key features can be retained to support precise services.

[0058] Simultaneously, the baseline interaction template library is updated. When a service request combination involving multiple overlapping items across departments and dynamically escalating urgency is detected, online reconstruction of the service semantic feature space is triggered. For example, if a patient needs to undergo examinations in multiple departments simultaneously during their visit and their condition undergoes an urgent change, the service semantic feature space will be automatically adjusted to better adapt to this complex situation. Specifically, semantic features related to emergency treatment will be added, and interaction templates will be optimized to provide more timely and accurate service responses. In the above steps, the response time for cross-departmental multi-item services is shortened through dynamic dimensionality reduction encoders and online reconstruction mechanisms, better addressing complex and ever-changing medical scenarios.

[0059] Furthermore, after triggering the online reconstruction of the service semantic feature space, the method of this application further includes:

[0060] The system uses an indoor positioning device to obtain the patient's real-time movement trajectory within the hospital; it performs cross-correlation analysis between the real-time movement trajectory and the service request timing, and dynamically sets collaborative timing instructions for multi-terminal content push by locking the arrival times of key service nodes, including the examination room door and the pharmacy window.

[0061] Specifically, indoor positioning devices refer to equipment or systems used to accurately locate patients in indoor environments, including Wi-Fi positioning, Bluetooth beacon positioning, and UWB (Ultra Wideband) positioning, which can obtain real-time location information of patients within the hospital; real-time movement trajectory refers to the path and location changes of patients within the hospital, continuously recorded by indoor positioning devices; cross-correlation analysis refers to assessing the correlation between two time series, and further, is used to analyze the relationship between the patient's real-time movement trajectory and the timing of service requests; key service nodes refer to important locations that patients must reach during their medical treatment, including the entrance to the examination room and the pharmacy window, and the arrival time of key service nodes is crucial for the coordination of service processes; collaborative timing instructions refer to instructions dynamically generated by the system based on the patient's real-time location and movement trajectory, used to coordinate content push and service response across multiple terminals, including self-service terminals, mobile applications, and ward screens.

[0062] Execution Steps: Indoor positioning devices are used to acquire patients' real-time movement trajectories within the hospital. For example, by deploying Bluetooth beacons within the hospital, patients' mobile devices can communicate with these beacons to accurately locate their positions and record their real-time movement trajectories, including path information such as entering the waiting area, heading towards the examination room, and reaching the pharmacy. Furthermore, the patient's mobile devices, including smartphones or smart bracelets provided by the hospital, are cross-correlation analyses performed between the patient's real-time movement trajectory and service request timing. For instance, the time interval from registration to examination appointment is analyzed, and combined with the patient's real-time location information, it is determined whether the patient has arrived at key service nodes within the expected time. Coordinated timing instructions for multi-terminal content push are dynamically set. Specifically, when the patient approaches the examination room door, examination notices are automatically pushed to the patient's mobile device; when the patient reaches the pharmacy window, medication pickup information is pushed. This real-time location-based service coordination mechanism improves the accuracy and timeliness of services.

[0063] Furthermore, the method in this application also includes dynamically setting collaborative timing instructions for content push across multiple terminals:

[0064] When a spatiotemporal misalignment is detected between the patient's location status and the terminal service response, a timing compensation strategy is activated to perform projection correction on the three-dimensional feature vector of the in-hospital service status.

[0065] Specifically, spatiotemporal misalignment refers to a mismatch between the patient's actual location and the terminal service response in terms of time and space. For example, the patient may have arrived at the examination room door, but the terminal device may still be pushing waiting information, or the patient may have been waiting in the waiting area for too long, and the terminal device may not have updated the status in time. Temporal compensation strategies are mechanisms used to correct spatiotemporal misalignment by adjusting the time sequence of service responses to ensure that the information received by the patient matches their actual location and needs. Projection correction refers to processing the three-dimensional feature vector of the hospital service status and adjusting the feature vector to a position closer to the actual status through projection methods, thereby correcting the deviation caused by spatiotemporal misalignment.

[0066] Execution steps: Real-time monitoring of the matching between patient location status and terminal service response. When a spatiotemporal mismatch is detected, further, if the patient has arrived at the examination room door but the terminal device is still pushing waiting information, the time-series compensation strategy is activated. Specifically, based on the patient's actual location and time status, the service response content and time sequence of the terminal device are dynamically adjusted. If the indoor positioning device detects that the patient has arrived at the examination room door but the terminal device has not yet updated its status, the time-series compensation strategy is triggered, switching the terminal device's service response from waiting information to examination preparation information and adjusting the push time to ensure that the information received by the patient is consistent with the actual needs.

[0067] Simultaneously, the three-dimensional feature vector of the hospital's service status is projected and corrected. The three-dimensional feature vector includes the patient's location, the status of medical and nursing resources, and environmental parameters. The location information in the feature vector is updated to the actual examination room door where the patient arrives through the projection method. At the same time, the status of medical and nursing resources and environmental parameters are adjusted to match the new location status, effectively correcting the service deviation caused by spatiotemporal misalignment, improving the accuracy and timeliness of services. Through time-series compensation strategies and projection correction, the service error rate caused by spatiotemporal misalignment is reduced, and patient needs are matched more accurately.

[0068] Furthermore, the method of this application also includes:

[0069] Under load service conditions, the collection granularity and reporting frequency of patient service log streams are dynamically determined based on the remaining power of the equipment. When the reconstructed residual of the three-dimensional feature vector of the in-hospital service status exceeds the dynamic threshold jointly set based on power status and service criticality level, the sparse interactive sampling mode at the corresponding collection granularity is activated. In the information collaboration network, under the joint optimization objective of terminal energy consumption constraints and service quality assurance index, the sparse interactive sampling mode is used to execute a piecewise linear scheduling strategy.

[0070] Specifically, load service status refers to the system operating under resource constraints during high traffic or high load conditions, which may affect service efficiency and quality; collection granularity refers to the level of detail in data collection, i.e., the amount of data collected each time. Finer collection granularity means more detailed data, but increases energy consumption; coarser collection granularity has the opposite effect; reporting frequency refers to how often data is uploaded to the server, i.e., how often data is uploaded. High-frequency reporting can provide more real-time information, but also increases energy consumption; reconstruction residual refers to the difference between the original data and the reconstructed data during data processing, reflecting the accuracy of data collection and processing; sparse interactive sampling mode refers to a strategy to reduce energy consumption by reducing the sampling frequency or the number of sampling points during data collection, while maintaining data validity as much as possible; piecewise linear scheduling strategy is a dynamic resource allocation strategy that adjusts resource allocation in segments according to different system states, including energy consumption and service quality requirements, to achieve a balance between energy consumption and service quality.

[0071] Execution steps: Under load service conditions, the granularity of patient service log stream collection and reporting frequency are dynamically adjusted based on the remaining power of the equipment. When the equipment has sufficient power, a finer collection granularity and a higher reporting frequency are used to obtain more detailed and real-time data; when the power is limited, the collection granularity and reporting frequency are automatically reduced to reduce energy consumption. When the reconstructed residual of the three-dimensional feature vector of the in-hospital service status exceeds the dynamic threshold jointly set based on the power status and service criticality level, the sparse interactive sampling mode is activated. Furthermore, if the reconstructed residual exceeds the preset threshold, it indicates that the current collection granularity and reporting frequency may not meet the service quality requirements, and the system will switch to the sparse interactive sampling mode to reduce energy consumption by reducing the number of sampling points or extending the sampling interval, while maintaining the validity of the data as much as possible.

[0072] In information collaborative networks, a piecewise linear scheduling strategy is implemented based on a sparse interactive sampling pattern. This strategy dynamically adjusts resource allocation according to the terminal's energy consumption constraints and service quality assurance index. For example, the device's power consumption is divided into multiple intervals, each corresponding to different data collection granularity and reporting frequency. Specifically, when the power consumption is high, service quality is prioritized, using a higher data collection granularity and reporting frequency; when the power consumption is low, energy consumption control is emphasized, using a lower data collection granularity and reporting frequency. Through this piecewise linear scheduling strategy, a dynamic balance is achieved between energy consumption and service quality, improving the system's adaptability and flexibility. Furthermore, by dynamically adjusting the data collection granularity and reporting frequency, more accurate service responses can be provided when power is sufficient, while basic service quality can still be guaranteed when power is limited.

[0073] In summary, the beneficial effects of the embodiments of this application are:

[0074] This application provides an intelligent interaction method and system for improving in-hospital patient services and information collaboration. It constructs a three-dimensional feature vector of in-hospital service status by collecting patient visit information, medical resource status, and in-hospital environmental parameters. Based on this feature vector, a demand-guided response strategy is deployed, and patient interaction behavior is tracked and interaction response parameters are collected. At in-hospital edge computing nodes, the visit time-series data is aligned with the patient service log stream to extract the coupling relationship between core patient demand features and medical resource supply features. Combined with the demand-guided response strategy, a dual-delay gradient is set under the service collaboration framework. Based on the dual-delay gradient under the service collaboration framework and combined with visit constraints, the interaction response parameters are dynamically optimized. The optimized interaction response parameters are then loaded into the terminal cluster, and the communication interaction protocol and collaborative semantic mapping rules corresponding to the information collaboration network are configured. This application achieves the technical effect of constructing a three-dimensional feature vector containing patients, medical staff, and the environment; extracting the coupling relationship between core patient demand features and medical resource supply features; dynamically deploying a demand-guided strategy; and collecting interaction parameters, thereby improving the response efficiency and reliability of in-hospital services and information collaboration.

[0075] Example 2, based on the same inventive concept as the intelligent interaction method for improving in-hospital patient services and information collaboration in the foregoing examples, such as... Figure 2 As shown in the figure, this application provides an intelligent interactive system for improving in-hospital patient services and information collaboration, wherein the system includes:

[0076] The 3D feature vector construction module M100 collects patient visit information, medical and nursing resource status, and hospital environment parameters to construct a 3D feature vector of hospital service status.

[0077] Interactive response parameter collection module M200: Based on the three-dimensional feature vector of the in-hospital service status, it deploys demand-guided response strategies, tracks patient interaction behavior, and collects interactive response parameters.

[0078] Coupling Relationship Extraction Module M300: At the hospital edge computing node, the patient visit time series data is aligned with the patient service log stream to extract the coupling relationship between the core needs of patients and the supply characteristics of medical and nursing resources. Combined with the demand-guided response strategy, a dual-delay gradient is set under the service collaboration framework.

[0079] Terminal cluster loading module M400: Based on the dual-delay gradient under the service collaboration framework and combined with the medical treatment constraints, dynamically optimizes the interaction response parameters, loads the optimized interaction response parameters into the terminal cluster, and configures the communication interaction protocol and collaborative semantic mapping rules corresponding to the information collaboration network.

[0080] Furthermore, the intelligent interactive system for improving in-hospital patient services and information collaboration is also used to perform the following methods:

[0081] The generator built into the service collaboration framework is used to simulate the interaction response sequence under the standard medical treatment process; the discriminator built into the service collaboration framework is used to identify service abnormal deviation events by comparing the distribution differences between the interaction response parameters of the in-hospital patient service interface and the interaction response sequence.

[0082] Furthermore, the interactive response parameter collection module M200 is also used to perform the following method:

[0083] A patient behavior dynamics analysis unit is set up, which is used to extract task progress rate and operation path deviation from multimodal interaction logs; track the patient's cross-device service jump trajectory between self-service terminals, mobile application service interfaces and ward screens, and extract the waiting tolerance threshold of key nodes; construct a service load profile based on the task progress rate and operation path deviation, combined with the waiting tolerance threshold, and dynamically adjust the content push frequency, information density and guidance dialogue complexity in the interaction response parameters based on the service load profile.

[0084] Furthermore, the interactive response parameter collection module M200 is also used to perform the following method:

[0085] A correlation mapping model between patient behavior characteristics and the intensity of medical resource occupancy is established. When the frequency of task interruption or rollback operation is detected to suddenly increase beyond the preset safety threshold, a soothing intervention strategy in the waiting area is automatically triggered. At the same time, a federated edge learning architecture is adopted. Under the condition that the edge data is encrypted and the original log does not leave the domain, the interaction satisfaction scores and behavioral residual data of patients in multiple wards are aggregated, and the attention mechanism weight parameters of the patient behavior dynamics analysis unit are iteratively optimized.

[0086] Furthermore, the coupling relationship extraction module M300 is used to perform the following method:

[0087] The time window denoising parameters of the patient service log stream are dynamically adjusted based on outpatient traffic, and a service event synchronization control node is embedded in the log preprocessing stage. At the hospital edge computing node, the service event synchronization control node is used to align the registration completion time, examination appointment trigger time, and the terminal's first interaction response start time.

[0088] Furthermore, the coupling relationship extraction module M300 is also used to perform the following method:

[0089] Deploy a dynamic dimensionality reduction encoder that adaptively selects the feature dimensionality reduction dimension based on the patient's historical visit frequency, patient visit type, and disease chronicity. At the same time, update the baseline interaction template library. If a service request combination with multiple overlapping items across departments and dynamically escalating urgency is detected, online reconstruction of the service semantic feature space is triggered.

[0090] Furthermore, the coupling relationship extraction module M300 is also used to perform the following method:

[0091] The system uses an indoor positioning device to obtain the patient's real-time movement trajectory within the hospital; it performs cross-correlation analysis between the real-time movement trajectory and the service request timing, and dynamically sets collaborative timing instructions for multi-terminal content push by locking the arrival times of key service nodes, including the examination room door and the pharmacy window.

[0092] Furthermore, the coupling relationship extraction module M300 is also used to perform the following method:

[0093] When a spatiotemporal misalignment is detected between the patient's location status and the terminal service response, a timing compensation strategy is activated to perform projection correction on the three-dimensional feature vector of the in-hospital service status.

[0094] Furthermore, the coupling relationship extraction module M300 is also used to perform the following method:

[0095] Under load service conditions, the collection granularity and reporting frequency of patient service log streams are dynamically determined based on the remaining power of the equipment. When the reconstructed residual of the three-dimensional feature vector of the in-hospital service status exceeds the dynamic threshold jointly set based on power status and service criticality level, the sparse interactive sampling mode at the corresponding collection granularity is activated. In the information collaboration network, under the joint optimization objective of terminal energy consumption constraints and service quality assurance index, the sparse interactive sampling mode is used to execute a piecewise linear scheduling strategy.

[0096] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Figure 1 The intelligent interaction method and specific example for improving in-hospital patient services and information collaboration in Embodiment 1 are also applicable to the intelligent interaction system for improving in-hospital patient services and information collaboration in this embodiment. Through the foregoing detailed description of the intelligent interaction method for improving in-hospital patient services and information collaboration, those skilled in the art can clearly understand the intelligent interaction system for improving in-hospital patient services and information collaboration in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0097] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0098] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. An intelligent interactive method for improving in-hospital patient services and information collaboration, characterized in that, The method includes: Collect patient visit information, medical and nursing resource status, and hospital environment parameters to construct a three-dimensional feature vector of hospital service status; Based on the three-dimensional feature vector of the in-hospital service status, a demand-guided response strategy is deployed, and patient interaction behavior is tracked and interaction response parameters are collected. At the edge computing node within the hospital, the time-series data of patient visits is aligned with the patient service log stream. The coupling relationship between the core needs of patients and the supply characteristics of medical and nursing resources is extracted. Combined with the demand-guided response strategy, a dual-delay gradient is set under the service collaboration framework. Based on the dual-delay gradient under the service collaboration framework, combined with the medical treatment constraints, the interaction response parameters are dynamically optimized, the optimized interaction response parameters are loaded into the terminal cluster, and the communication interaction protocol and collaborative semantic mapping rules corresponding to the configuration information collaboration network are configured. This includes tracking patient interactions and collecting interaction response parameters, as well as: A patient behavior dynamics analysis unit is set up, which is used to extract the task progress rate and operation path offset from the multimodal interaction log; Track patients' cross-device service jump trajectories between self-service terminals, mobile application service interfaces, and ward screens, and extract the waiting tolerance thresholds of key nodes; Based on the task progress rate and operation path deviation, combined with the waiting tolerance threshold, a service load profile is constructed, and the content push frequency, information density and guidance script complexity in the interaction response parameters are dynamically adjusted based on the service load profile. This also includes: Establish a correlation mapping model between patient behavior characteristics and the intensity of medical and nursing resource occupancy. When the frequency of task interruption or rollback operation is detected to suddenly increase beyond the preset safety threshold, the comfort intervention strategy in the waiting area is automatically triggered. Meanwhile, using a federated edge learning architecture, under the conditions of encrypted edge data and original logs not leaving the domain, the interaction satisfaction scores and behavioral residual data of patients from multiple wards are aggregated, and the attention mechanism weight parameters of the patient behavior dynamics analysis unit are iteratively optimized. Among these, at the hospital's edge computing nodes, the alignment of visit time-series data with patient service log streams includes: The time window noise reduction parameters of the patient service log stream are dynamically adjusted based on outpatient traffic, and a service event synchronization control node is embedded in the log preprocessing stage. At the edge computing node within the hospital, the service event synchronization control node is used to align the registration completion time, the examination appointment trigger time, and the terminal's first interaction response start time.

2. The intelligent interaction method for improving in-hospital patient services and information collaboration as described in claim 1, characterized in that, The built-in generator of the service collaboration framework is used to simulate the interactive response sequence under the standard medical treatment process; The discriminator built into the service collaboration framework is used to identify service deviation events by comparing the distribution differences between the interaction response parameters of the in-hospital patient service interface and the interaction response sequence.

3. The intelligent interaction method for improving in-hospital patient services and information collaboration as described in claim 1, characterized in that, The method further includes: Deploy a dynamic dimensionality reduction encoder that adaptively selects the feature dimensionality reduction dimension based on the patient's historical frequency of visits, the type of patient visits, and the chronicity of the disease. At the same time, the baseline interaction template library is updated. If a combination of service requests involving multiple overlapping items across departments and dynamically escalating urgency of the illness is detected, the online reconstruction of the service semantic feature space is triggered.

4. The intelligent interaction method for improving in-hospital patient services and information collaboration as described in claim 3, characterized in that, Triggering online reconstruction of the service semantic feature space, the method further includes: Indoor positioning devices are used to obtain the real-time movement trajectory of patients within the hospital; The real-time movement trajectory is cross-correlation analysis is performed with the service request timing, and by locking the arrival time of key service nodes including the examination room door and the pharmacy window, the collaborative timing instructions for multi-terminal content push are dynamically set.

5. The intelligent interaction method for improving in-hospital patient services and information collaboration as described in claim 4, characterized in that, The method further includes dynamically setting collaborative timing instructions for content push across multiple terminals: When a spatiotemporal misalignment is detected between the patient's location status and the terminal service response, a timing compensation strategy is activated to perform projection correction on the three-dimensional feature vector of the in-hospital service status.

6. The intelligent interaction method for improving in-hospital patient services and information collaboration as described in claim 5, characterized in that, The method further includes: Under load service conditions, the collection granularity and reporting frequency of patient service log streams are dynamically determined based on the remaining power of the equipment. When the reconstructed residual of the three-dimensional feature vector of the service status in the hospital exceeds the dynamic threshold jointly set based on the power status and service criticality level, the sparse interactive sampling mode at the corresponding collection granularity is activated. In the information collaboration network, a piecewise linear scheduling strategy is executed under the joint optimization objective of terminal energy consumption constraints and service quality assurance index, using the sparse interactive sampling mode.

7. An intelligent interactive system for improving in-hospital patient services and information collaboration, characterized in that, The system is used to implement the intelligent interaction method for improving in-hospital patient services and information collaboration as described in any one of claims 1-6, wherein the system comprises: 3D Feature Vector Construction Module: Collects patient visit information, medical and nursing resource status, and hospital environment parameters to construct a 3D feature vector of hospital service status; Interactive response parameter collection module: Based on the three-dimensional feature vector of the in-hospital service status, deploy demand-guided response strategies, track patient interaction behavior, and collect interactive response parameters; Coupling Relationship Extraction Module: At the edge computing node within the hospital, the patient visit time series data is aligned with the patient service log stream to extract the coupling relationship between the core needs of patients and the supply characteristics of medical and nursing resources. Combined with the demand-guided response strategy, a dual-delay gradient is set under the service collaboration framework. Terminal cluster loading module: Based on the dual-latency gradient under the service collaboration framework and combined with the medical treatment constraints, dynamically optimize the interaction response parameters, load the optimized interaction response parameters into the terminal cluster, and configure the communication interaction protocol and collaborative semantic mapping rules corresponding to the information collaboration network.

Images