Immune planning visual monitoring method and device, electronic equipment and storage medium

Abstract

The invention discloses an immune planning visual monitoring method and device, electronic equipment and a storage medium, and relates to the technical field of immune monitoring. The method comprises the steps that a first instruction is received, and a first display interface containing a multi-modal data set is generated; performing synchronous prediction on the multi-modal data set to obtain inventory gap probability distribution and expiration risk distribution, and displaying a corresponding visual layer in a first display interface; when any distribution meets a preset early warning condition, popping up an interpretable early warning card in the first display interface; and after binding the early warning vector corresponding to the interpretable early warning card with the current outpatient service identifier, displaying an early warning event record in a rolling manner in a time axis form in a first display interface. In this way, the problems that in the prior art, the service data clock granularity is different, early warning only stays in static characters, risk reasons cannot be explained, manual screenshot is needed in post-event auditing, and evidences are prone to being tampered can be solved.

Description

Technical Field

[0001] This application relates to the field of immune surveillance technology, and in particular to an immune program visualization surveillance method, device, electronic device and storage medium. Background Technology

[0002] Currently, most county-level immunization program information systems in China still rely on static reports combined with manual verification as their core operating mode. Data such as vaccine inventory, cold chain temperature, vaccination records, and video supervision are stored in different business modules. Due to inconsistencies in data format, update frequency, and time base among these modules, administrators cannot simultaneously observe key information such as vaccine availability, temperature status, vaccination progress, and on-site compliance within a unified interface, resulting in a typical data silo problem.

[0003] For example, when the system detects a risk of low stock or near expiration for a particular vaccine, it typically notifies the system via a pop-up text alert or an exported Excel list. This static alert lacks a quantifiable risk level and fails to explain the cause of the risk. Supervisory personnel can only rely on experience to judge the allocation volume, often resulting in a structural imbalance in inventory levels between the receiving and receiving parties. Furthermore, the existing platforms use inconsistent time bases: inventory reporting is done hourly, cold chain temperature is measured in seconds, and vaccination records are compiled daily. This misalignment leads to logical conflicts such as temperatures exceeding limits without inventory movement, or sudden drops in inventory while temperatures return to normal, which cannot be automatically identified and must be manually verified afterward—inefficient and prone to omissions. Additionally, the existing system only provides a simple list of risk events. If supervisory personnel need to review a near-expiration warning, they must manually take screenshots, export reports, and stitch together timelines—a cumbersome process where evidence is easily modified.

[0004] Therefore, the industry urgently needs an immune planning visualization monitoring method that can unify multimodal data, quantify the risks of stockouts and expiration, provide interpretable decision-making basis, and automatically generate an tamper-proof audit chain. Summary of the Invention

[0005] In view of this, embodiments of this application disclose an immunization program visualization monitoring method, device, electronic device and storage medium to solve the problems of inconsistent clock granularity, warnings that are only static text and cannot explain the cause of risk, the need for manual screenshots for post-event auditing, and the ease with which evidence can be tampered with.

[0006] On one hand, embodiments of this application provide a method for visual monitoring of immunization programs, including: Upon receiving the first instruction, a first display interface containing a multimodal dataset is generated. The multimodal dataset includes at least vaccine inventory information, cold chain information, vaccination information, and outpatient video supervision information. Simultaneous prediction is performed on the multimodal dataset to obtain the probability distribution of inventory shortage and the risk distribution of expiration date, and the corresponding visualization layers are displayed in the first display interface; When any distribution meets the preset warning conditions, an interpretable warning card pops up in the first display interface. The warning card is an interactive floating window that includes the vaccine name, remaining inventory, remaining days of expiration, key feature weight graphs, and a collapsible / expandable area. After binding the warning vector corresponding to the interpretability warning card to the current outpatient clinic identifier, the warning event records are displayed in a scrolling timeline format on the first display interface. The warning event records include at least the vaccine name, outpatient clinic identifier, timestamp, and color dynamic QR code.

[0007] In one possible embodiment, the video supervision information includes real-time images from various outpatient video monitoring points that have been connected to the current system, wherein the monitoring points can be switched.

[0008] In one possible embodiment, the first display interface further includes: the number of people served on the day, the number of people served on the month, the number of people served throughout the year, and the number of people vaccinated on the day in the whole district; as well as the number of dog bite clinics, the number of adult clinics, the number of children's clinics, the number of obstetric clinics, and the number of tetanus clinics.

[0009] On one hand, embodiments of this application provide an immunization program visualization monitoring device, including: The receiving module is used to receive the first instruction and generate a first display interface containing a multimodal dataset. The multimodal dataset includes at least vaccine inventory information, cold chain information, vaccination information, and outpatient video supervision information. The prediction module is used to simultaneously predict the multimodal dataset, obtain the probability distribution of inventory gap and the risk distribution of expiration date, and display the corresponding visualization layers in the first display interface; The early warning module is used to pop up an interpretable early warning card in the first display interface when any distribution meets the preset early warning conditions. The early warning card is an interactive floating window that includes the vaccine name, remaining inventory, remaining days of expiration, key feature weight graphs, and a collapsible and expandable area. The display module is used to bind the warning vector corresponding to the interpretable warning card to the current outpatient identifier and then display the warning event records in a scrolling timeline format on the first display interface. The warning event records include at least the vaccine name, outpatient identifier, timestamp, and color dynamic QR code.

[0010] In one possible embodiment, the inventory information includes: net inflow, inventory, usage, and loss of immunization program vaccines or non-immunization program vaccines during the vaccine statistical period; and real-time vaccine inventory status of each outpatient department; wherein the vaccine information corresponding to immunization program vaccines and non-immunization program vaccines can be switched back and forth.

[0011] In one possible embodiment, the video supervision information includes real-time images from various outpatient video monitoring points that have been connected to the current system, wherein the monitoring points can be switched.

[0012] In one possible embodiment, the vaccination information includes: the vaccination rate of eight vaccines for children aged 2 to 6 years in each vaccination unit that has been connected to the current system; and the vaccination status on the day, including the recipients, the number of vaccinations, and the specific names of the vaccines administered; wherein the recipients include children, adults, dog bite victims, and tetanus infected individuals.

[0013] In one possible embodiment, the first display interface further includes: the number of people served on the day, the number of people served on the month, the number of people served throughout the year, and the number of people vaccinated on the day in the whole district; as well as the number of dog bite clinics, the number of adult clinics, the number of children's clinics, the number of obstetric clinics, and the number of tetanus clinics.

[0014] In one possible embodiment, the multimodal dataset is a multimodal dataset calibrated for spatiotemporal consistency. The calibration process includes: upon receiving a second instruction or detecting that the deviation between the inventory reporting frame timestamp and the cold chain temperature probe timestamp is greater than a preset threshold, a calibration progress bar overlay pops up at the bottom of the first display interface; wherein, the overlay uses the cold chain temperature probe timestamp as a reference to back-align the inventory reporting frame timestamp and displays the alignment progress percentage in real time; within the set alignment window, if the vaccine circulation time is greater than the inventory change allowable window, or the cold chain temperature is normal and the inventory drop is greater than a preset range, the overlay is stopped in the red zone and the spatiotemporal data source node to be corrected is highlighted with a red pulse until the anomaly is resolved or a shutdown instruction is received.

[0015] In one possible embodiment, the prediction module is used to: obtain vaccine inventory, allocation volume, allocation relationship, remaining days of expiration, transportation mileage, and shared transportation routes based on a multimodal dataset; construct an inventory propagation graph using vaccine inventory and allocation volume as input features for outpatient nodes and vaccine allocation relationship as hyperedges, and generate an inventory hidden vector through the first channel convolution of a dual-channel hypergraph neural network (Dual-HGNN); construct an expiration decay graph using remaining days of expiration and transportation mileage as input features for vaccine batch number nodes and shared transportation routes as hyperedges, and generate an expiration hidden vector through the second channel convolution of a dual-channel hypergraph neural network (Dual-HGNN); input the inventory hidden vector and the expiration hidden vector into a cross-attention module, fuse them to obtain a joint hidden state, and present the fusion process in real time on the first display interface with a 3D hyperedge shrinkage animation; perform a one-time regression on the joint hidden state using a decoding head, and output the inventory gap probability distribution and the expiration risk distribution.

[0016] In one possible embodiment, the early warning module is configured to: receive a second instruction and generate a second display interface, the second display interface including a heat map layer constructed in real time using street-level polygons as base map units, the heat map layer including the heat map vectors of the daily vaccination doses of all clinics in the current street and the real-time inventory heat map vectors of the corresponding vaccines; perform element-level difference calculations on the heat map vectors of the daily vaccination doses and the real-time inventory heat map vectors to obtain a street heat map difference matrix; when the absolute value of any element in the street heat map difference matrix is ​​greater than a preset threshold, an early warning float pops up at the edge of the layer; obtain a click signal triggered by the early warning float point, the second display interface automatically scales to the center coordinates of the unbalanced street, and calls the interface of the interpretable early warning card to inject the unbalanced street identifier, vaccine name and corresponding difference vector into the interpretable early warning card to complete the pop-up display.

[0017] In one possible embodiment, the early warning module is configured to: when any distribution meets a preset early warning condition, display an interpretable early warning card with a slide-in animation in the lower right corner of the first display interface; inject an interactive Shapley value Sankey diagram component into the inner layer of the interpretable early warning card, the component displaying the marginal contribution percentage of each feature to the inventory gap probability in a single flow bandwidth from the left feature node to the right inventory gap probability node; the entire bandwidth flow is calculated by the Shapley algorithm based on the vaccine inventory, allocation quantity, transportation days, and remaining expiration days; and inject a voice broadcast button into the top layer of the interpretable early warning card, wherein... When the voice broadcast button is clicked, the vaccine name, shortage quantity, and suggested allocation path are broadcast in natural language template, and male / female / dialect switching is supported; the one-click collapse / expand area is injected into the sidebar of the explainability warning card. In the collapsed state, only the title bar is retained, and in the expanded state, the Shapley value Sankey diagram component and the voice broadcast button are fully displayed; the one-click evidence storage button is injected into the bottom of the explainability warning card. When the one-click evidence storage button is clicked, the current card content, the snapshot of the Shapley value Sankey diagram component, and the current timestamp are encapsulated into an audit package with Merkle hash and appended to the timeline log of the first display interface.

[0018] On one hand, embodiments of this application provide an electronic device including a processor and a memory, wherein the memory stores program code, and when the program code is executed by the processor, the processor executes any of the above-described immune program visualization monitoring methods.

[0019] On the one hand, this application provides a computer-readable storage medium including program code, which, when the storage medium is run on an electronic device, causes the electronic device to perform any of the above-mentioned immunization program visualization monitoring methods.

[0020] The beneficial effects of this application are as follows: This application provides an immunization program visualization monitoring method, device, electronic device, and storage medium. First, a multimodal dataset is loaded all at once via a first instruction, presenting four types of information simultaneously on separate cards to eliminate data silos. Then, inventory reporting frames are back-aligned based on synchronous prediction, automatically eliminating timestamp discrepancies and enabling real-time detection of logical conflicts through the interface. Next, dual-probability heatmap layers are overlaid on the same screen to quantify the risk level of inventory shortage probability and expiration probability. When any probability exceeds a threshold, an interpretable pop-up card slides out in the lower right corner, allowing the user to click on the Shapley value Sankey diagram to see the percentage contribution of each feature to the shortage, achieving root cause quantification. Finally, the warning vector is bound to the outpatient identifier, and the interface generates an audit package with Merkle hash and QR code with one click, completing tamper-proof evidence storage. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of an application scenario in an embodiment of this application; Figure 2 This is a flowchart illustrating the implementation of an immunization program visualization monitoring method in this application embodiment; Figure 3 This is a schematic diagram of a first display interface in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of an immunization program visualization monitoring device in an embodiment of this application; Figure 5 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below 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. 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. Unless otherwise specified, the embodiments and features in the embodiments of this application can be arbitrarily combined with each other. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0024] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in sequences other than those illustrated or described herein.

[0025] The design concept of the embodiments of this application is briefly introduced below: Currently, most county-level immunization program information systems in China still rely on static reports combined with manual verification as their core operating mode. Data such as vaccine inventory, cold chain temperature, vaccination records, and video supervision are stored in different business modules. Due to inconsistencies in data format, update frequency, and time base across these modules, managers cannot simultaneously observe key information such as remaining vaccine quantity, temperature exceeding limits, vaccination progress, and on-site compliance within a unified interface, creating a typical data silo problem. For example, when the system detects a risk of low vaccine inventory or nearing its expiration date, it typically notifies the system via pop-up text alerts or exported Excel lists. These static alerts lack quantifiable risk levels and cannot explain the cause of the risk. Supervisory personnel can only rely on experience to judge the allocation volume, often resulting in a structural imbalance in inventory levels between the receiving and receiving parties. Furthermore, the existing platforms suffer from inconsistent time bases. Inventory reporting is done at the hourly level, cold chain temperature at the second level, and vaccination records at the daily level. This misalignment leads to logical conflicts such as temperatures exceeding limits without inventory changes, or sudden inventory drops while temperatures return to normal, which cannot be automatically identified and must be manually verified afterward, resulting in low efficiency and a high risk of omissions. Additionally, the existing system only provides a simple list of risk events. If regulators need to review a near-expiration warning, they must manually take screenshots, export reports, and stitch together timelines—a cumbersome process that makes evidence susceptible to modification. Therefore, the industry urgently needs a visual monitoring method for immunization programs that can unify multimodal data, quantify stockout and expiration risks, provide interpretable decision-making support, and automatically generate an immutable audit chain.

[0026] In view of this, embodiments of this application provide an immunization program visualization monitoring method, device, electronic device, and storage medium. The immunization program visualization monitoring method includes: receiving a first instruction and generating a first display interface containing a multimodal dataset, the multimodal dataset including at least vaccine inventory information, cold chain information, vaccination information, and outpatient video supervision information; synchronously predicting the multimodal dataset to obtain the probability distribution of inventory shortage and the risk distribution of expiration date, and displaying the corresponding visualization layers on the first display interface; when any distribution meets a preset warning condition, popping up an interpretable warning card on the first display interface, the warning card being an interactive floating window containing the vaccine name, remaining inventory, remaining days of expiration date, key feature weight graphics, and a collapsible / expandable area; binding the warning vector corresponding to the interpretable warning card to the current outpatient identifier, and then displaying warning event records in a scrolling timeline format on the first display interface, the warning event records including at least the vaccine name, outpatient identifier, timestamp, and a color dynamic QR code. Thus, the system first loads the multimodal dataset in one go via the first instruction, presenting the four types of information on the same screen in separate cards, eliminating data silos. Then, based on the synchronous prediction process, the inventory reporting frames are back-aligned, automatically eliminating timestamp deviations, so that logical conflicts can be detected by the interface in real time. Next, dual-probability heatmap layers are used to overlay the probability of inventory shortage and the probability of expiration on the same screen to quantify the risk level. When any probability exceeds the threshold, an interpretable pop-up card slides out in the lower right corner. Users can click on the Shapley value Sankey diagram to see the percentage contribution of each feature to the shortage, realizing root cause quantification. Finally, the warning vector is bound to the clinic identifier, and the interface generates an audit package with Merkle hash and QR code with one click, completing tamper-proof evidence storage.

[0027] The preferred embodiments of this application will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0028] like Figure 1 The diagram shown illustrates an application scenario according to an embodiment of this application. The application scenario diagram includes a terminal device 101 and a server 102. The terminal device 101 and the server 102 communicate via a communication network.

[0029] Terminal device 101 is an electronic device used by the target user. This electronic device can be a personal computer, mobile phone, tablet computer, laptop, e-book reader, vehicle-mounted terminal, etc. Furthermore, terminal device 101 can have a client application related to immunization program visualization monitoring installed. This client application can be software (e.g., an app, browser), a webpage, a mini-program, etc. The target user (user end) can use the aforementioned client application related to immunization program visualization monitoring through terminal device 101 to perform operations related to immunization program visualization monitoring.

[0030] Server 102 can be a standalone physical server, an edge device 102 in the field of cloud computing, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud storage, cloud functions, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0031] There is no limit to the number of the aforementioned terminal devices 101 and / or servers 102.

[0032] It should be noted that the immunization program visualization monitoring method in this embodiment can be executed by the terminal device 101 or the server 102 alone, or by both the terminal device 101 and the server 102. For example, when executed by both the terminal device 101 and the server 102, the terminal device 101 receives a first instruction and generates a first display interface containing a multimodal dataset. The multimodal dataset includes at least vaccine inventory information, cold chain information, vaccination information, and outpatient video supervision information. The server 102 performs synchronous prediction on the multimodal dataset to obtain the probability distribution of inventory shortage and the risk distribution of expiration date, and displays the corresponding visualization layers in the first display interface. When any distribution meets the preset warning conditions, the terminal device 101 pops up an interpretable warning card in the first display interface. The warning card is an interactive floating window that includes the vaccine name, remaining inventory, remaining days of expiration date, key feature weight graph, and a collapsible / expandable area. After binding the warning vector corresponding to the interpretable warning card with the current outpatient identifier, the warning event record is displayed in a scrolling timeline format in the first display interface. The warning event record includes at least the vaccine name, outpatient identifier, timestamp, and color dynamic QR code.

[0033] The following describes the immunization program visualization monitoring method provided by the exemplary embodiments of this application in conjunction with the above application scenarios and with reference to the accompanying drawings. It should be noted that the above application scenarios are only shown to facilitate understanding of the spirit and principles of this application, and the embodiments of this application are not limited in any way in this respect.

[0034] It is particularly important to emphasize that all information and data obtained in this application embodiment have been with the consent of relevant departments and personnel and have been carried out in a legal and compliant manner. The information and data include, but are not limited to, video data, personal information of vaccine recipients, information of vaccination units, information of medical staff, and vaccine information.

[0035] refer to Figure 2 This is a flowchart illustrating the implementation of an immunization program visualization monitoring method provided in this application embodiment. The method is described here with server 102 as the execution entity, and the specific implementation process includes: S201, Receive the first instruction and generate a first display interface containing a multimodal dataset. The multimodal dataset includes at least vaccine inventory information, cold chain information, vaccination information, and outpatient video supervision information.

[0036] In this embodiment, the first instruction represents the start instruction of the immunization program visualization monitoring system. After the target object triggers the first instruction, the server generates a first display interface and displays it on the terminal device. The first display interface includes a multimodal dataset, which includes inventory information, cold chain information, vaccination information, and outpatient video supervision information.

[0037] Inventory information includes: net inflow, inventory, usage, and loss of immunization program vaccines or non-immunization program vaccines during the vaccine statistics period; and real-time vaccine inventory status of each outpatient department; among which, vaccine information corresponding to immunization program vaccines and non-immunization program vaccines can be switched back and forth.

[0038] The vaccination information includes: the vaccination rate of eight vaccines for children aged 2-6 years in each vaccination unit that has been connected to the current system; and the vaccination status of the day, including the recipients, the number of doses, and the specific names of the vaccines administered; among which, the recipients include children, adults, dog bite victims, and tetanus infected individuals.

[0039] The video supervision information includes real-time footage from each outpatient video monitoring point that has been connected to the current system, and the monitoring points can be switched.

[0040] In one embodiment, the first display interface further includes: the number of people served on the day, the number of people served in the month, the number of people served throughout the year, and the number of people vaccinated in the district on the day; as well as the number of dog bite clinics, the number of adult clinics, the number of children's clinics, the number of obstetric clinics, and the number of tetanus clinics.

[0041] like Figure 3 The diagram shown is a schematic representation of a first display interface provided in an embodiment of this application, wherein: The video supervision section primarily displays real-time video feeds from various outpatient clinics in the region that are currently connected to the A-zone exemption information system video supervision platform. The list of unit names shows the townships and outpatient clinics within the region that are covered by video surveillance. You can select a township / street to view the monitoring locations of outpatient clinics in that township / street.

[0042] The section on vaccination rates for eight vaccines for children aged 2-6 years primarily uses electronic monitoring codes for upstream tracking, displaying vaccine information, vaccine distribution traceability nodes, and a map showing the vaccine's journey. Clicking on a vaccine type displays the ranking of that vaccine among vaccination clinics within the region.

[0043] The service volume indicator display section mainly shows the number of people vaccinated in the region on the same day, the number of people served on the same day, the number of people served in the same month, and the number of people served throughout the year.

[0044] The outpatient status section displays the number of people vaccinated, the number of people served, the number of people served in the month, and the number of people served throughout the year for the entire region.

[0045] The outpatient clinic display subsystem mainly shows the number of vaccination clinics of various types within the area. Specifically, clicking on the rabies clinic displays the clinics within the area that can administer rabies vaccines; clicking on the adult clinic displays the clinics within the area that can administer adult vaccines; clicking on the pediatric clinic displays the clinics within the area that can administer pediatric vaccines; clicking on the obstetrics clinic displays the clinics within the area that can administer neonatal vaccines; and clicking on the tetanus clinic displays the clinics within the area that can administer tetanus vaccines.

[0046] The vaccine inventory section primarily displays the inventory of immunized and non-immunized vaccines at various clinics within the region. Clicking on "immunized / non-immunized" in the vaccine inventory module, selecting a date range and vaccine name (multiple selections supported), and clicking "view" will show the net inflow, inventory, usage, and loss of that vaccine within the region during the selected date range. Clicking on "immunized / non-immunized" in the vaccine inventory module, selecting a date range and vaccine name (multiple selections supported), and clicking "view" will show the real-time inventory of that vaccine at clinics within the region during the selected date range.

[0047] The "Daily Vaccination Status" section primarily displays the vaccination status of various types of vaccination clinics within the district. Clicking on "Children's Vaccination" allows you to view the total number of children's vaccination doses and the number of doses for each children's vaccine in the district that day; clicking on "Adult Vaccination" allows you to view the total number of adults' vaccination doses and the number of doses for each adults' vaccine in the district that day; clicking on "Obstetric Vaccination" allows you to view the total number of newborn vaccination doses and the number of doses for newborn vaccines in the district that day; clicking on "Dog Bite Vaccination" allows you to view the total number of rabies vaccination doses and the number of doses for rabies vaccines in the district that day; and clicking on "Tetanus Vaccination" allows you to view the total number of tetanus vaccination doses and the number of doses for tetanus vaccines in the district that day.

[0048] The Cold Chain Data section primarily displays the cold chain data monitoring status within this area. Based on this section, users can view the monitoring data of each probe in the cold chain monitoring system of each clinic within this area that has been connected to the Area A Exemption Information System platform, including the clinic's name, warehouse name, and real-time temperature.

[0049] The Vaccine Inventory and Expiry Date Warning section primarily displays data on low-stock and near-expiry vaccines within the region. This section provides a rolling display of information on low-stock and near-expiry vaccines and their corresponding clinics within the region.

[0050] S202, synchronously predict the multimodal dataset to obtain the probability distribution of inventory gap and the distribution of expiration risk, and display the corresponding visualization layers in the first display interface.

[0051] In the application embodiment, after obtaining the multimodal dataset, the multimodal dataset is further subjected to spatiotemporal consistency calibration. The calibration process includes: upon receiving the second instruction, or detecting that the deviation between the inventory reporting frame timestamp and the cold chain temperature probe timestamp is greater than a preset threshold, a calibration progress bar overlay pops up at the bottom of the first display interface; wherein, the overlay uses the cold chain temperature probe timestamp as a reference to back-align the inventory reporting frame timestamp and displays the alignment progress percentage in real time; within the set alignment window, if the vaccine circulation time is greater than the inventory change allowable window, or the cold chain temperature is normal and the inventory drop is greater than a preset range, the overlay is stopped in the red zone and the spatiotemporal data source node to be corrected is highlighted with a red pulse until the anomaly is resolved or a closing instruction is received.

[0052] Thus, the calibration process uses the probe clock as a reference, aligning the inventory reporting frames back to the same second window to eliminate time drift in multimodal data; the interface progress bar provides real-time feedback on the alignment rate, stopping the red pulse immediately upon an anomaly. If the processing time within the same window exceeds the allowable duration, or if the temperature is normal but inventory drops sharply, the system immediately highlights the spatiotemporal data source node, transforming logical conflicts that previously required manual verification into readily visible interface events. This enables anomaly capture and misjudgment correction, ensuring that subsequent predictions and early warnings are based on a time-consistent and logically self-consistent multimodal dataset, thereby improving the accuracy of predictions and early warnings.

[0053] Furthermore, synchronous predictions are performed on the multimodal dataset that has undergone spatiotemporal consistency calibration to obtain the probability distribution of inventory shortages and the distribution of expiration risk, including: First, based on the multimodal dataset, we obtained the vaccine inventory, allocation amount, allocation relationship, remaining days of expiration, transportation mileage, and shared transportation routes.

[0054] Then, using vaccine inventory and allocation volume as input features for outpatient nodes and vaccine allocation relationships as hyperedges, an inventory propagation graph is constructed. The inventory hidden vector is generated through the first channel convolution of a dual-channel hypergraph neural network (Dual-HGNN). The inventory propagation graph is a hypergraph data structure used for interface-side risk prediction. It uses outpatient nodes as nodes, disclosed vaccine allocation relationships as hyperedges, and vaccine inventory and allocation volume as node input features. Through a single hypergraph convolution, the inventory hidden vector of the outgoing node is simultaneously transmitted to all incoming nodes, generating an interface-level quantitative prediction of future shortage probabilities.

[0055] Simultaneously, using the remaining days of expiration and transportation mileage as input features for vaccine batch number nodes, and the shared transportation route as hyperedges, an expiration decay graph is constructed. The expiration date hidden vector is generated through the second channel convolution of a dual-channel hypergraph neural network (Dual-HGNN). The expiration decay graph is a hypergraph data structure used for interface-side risk prediction. It uses vaccine batch numbers as nodes, shared transportation routes as hyperedges, and the remaining days of expiration and transportation mileage as node input features. Through a single hypergraph convolution, the time-temperature decay effect along the transportation route is synchronously transmitted to all batch number nodes, generating an interface-level quantitative prediction of future expiration probability.

[0056] Furthermore, the hidden vectors of inventory and expiration date are input into the cross-attention module and fused to obtain a joint hidden state. The fusion process is presented in real time on the first display interface using a 3D hyperedge contraction animation. The cross-attention module is the weight allocator of the interface-side fusion engine. It uses the hidden vectors of inventory and expiration date as the query and key respectively to calculate cross-channel attention weights and generate a joint weight matrix in one go. This matrix determines the proportion of stockout risk and expiration risk in the final joint hidden state, achieving interface-level quantitative fusion of which is more important and which occupies bandwidth, and driving the real-time flow amplitude of the 3D hyperedge contraction animation. The joint hidden state is the quantitative vector of the interface-side fusion result. Through the cross-attention module, the hidden vectors of inventory output from the inventory propagation graph and the hidden vectors of expiration date output from the expiration date decay graph are weighted and fused to generate a dual-channel quantitative representation that simultaneously includes stockout risk and expiration risk. This vector serves as the sole input to the decoder and is directly regressed to the probability distribution of inventory shortage and the distribution of expiration risk, achieving interface-level synchronous prediction of one vector and two probabilities.

[0057] Finally, the joint hidden state is regressed once by the decoding head to output the probability distribution of inventory shortage and the risk distribution of expiration date. Specifically, the joint hidden state (which already contains the dual channel weights of shortage and expiration) is obtained, and the joint hidden state is decoded by a single-layer fully connected (or lightweight MLP) to output two neurons; further, the two neurons are activated to obtain the probability values ​​in the interval [0, 1], where the value of the first neuron is the probability of inventory shortage and the value of the second neuron is the probability of expiration date.

[0058] Using the above methods, multimodal data is loaded all at once on the first interface. Inventory frames are back-aligned based on the cold chain clock, eliminating time drift in the background technology. Dual-channel hypergraph parallel convolution transforms allocation relationships and transportation paths into a simultaneous 3D hyper-edge flow on the same screen, generating dual probabilities in the [0, 1] interval in a single forward pass, making stockout risk and expiration risk appear simultaneously on the screen. Furthermore, a single regression compresses the joint hidden state into two neurons. Activation of these neurons yields the probability of inventory shortage and the probability of expiration, achieving real-time quantification of dual probabilities. This transforms the previously scattered, qualitative, and manual risk assessment into a simultaneous, quantitative, and one-click interface operation, systematically solving the four major pain points of data silos, clock deviation, lack of quantification, and lack of auditing.

[0059] S203 When any distribution meets the preset warning conditions, an interpretable warning card pops up in the first display interface. The warning card is an interactive floating window that includes the vaccine name, remaining inventory, remaining days of expiration, key feature weight graphs, and a collapsible area.

[0060] In this embodiment of the application, after obtaining the probability distribution of inventory shortage and the risk distribution of expiration date, it is further determined whether the probability distribution of inventory shortage and the risk distribution of expiration date meet the early warning conditions, that is, whether the probability of inventory shortage is greater than or equal to the preset shortage threshold or whether the probability of expiration date is greater than or equal to the preset expiration threshold.

[0061] When either the inventory shortage probability distribution or the expiration risk distribution meets the preset warning conditions, an interpretable warning card will pop up with a slide-in animation in the lower right corner of the first display interface. An interactive Shapley Sankey diagram component will be injected into the inner layer of the interpretable warning card. This component will display the marginal contribution percentage of each feature to the inventory shortage probability in a single flow bandwidth from the left feature node to the right inventory shortage probability node. The entire bandwidth flow will be calculated by the Shapley algorithm based on vaccine inventory, allocation volume, transportation days, and remaining expiration days. A voice broadcast button will be injected into the top layer of the interpretable warning card. Clicking the button will announce the vaccine name, shortage quantity, and suggested allocation path in natural language, supporting multiple voice options including male, female, English, Chinese, and dialects. A one-click collapsible / expandable area is injected into the sidebar of the explainability warning card. In the collapsed state, only the title bar is retained; in the expanded state, the Shapley value Sankey diagram component and the voice announcement button are fully displayed. A one-click evidence storage button is injected into the bottom of the explainability warning card. When clicked, the current card content, a snapshot of the Shapley value Sankey diagram component, and the current timestamp are encapsulated into an audit package with a Merkle hash and appended to the timeline log of the first display interface. The Merkle hash involves first cutting the card content, Shapley diagram, and timestamp into blocks, then performing SHA-256 hashes level by level, finally synthesizing a unique root hash value. This root hash is written to a PDF and a QR code, which can be scanned to compare with the on-chain anchor value, achieving tamper-proof auditing.

[0062] Thus, after obtaining the probability distribution of inventory shortages and the risk distribution of expiration dates, if the system determines that either distribution meets the warning conditions, it triggers the following interface chain: slide in a card in the lower right corner → pause heatmap → inject the overall Shapley Sankey diagram. During this process, clicking on feature nodes can recalculate the contribution percentage and refresh the bandwidth; the top voice button also supports multiple voice broadcasts, thus achieving dual-channel explanation through viewing and listening. In addition, the sidebar can be collapsed with one click, collapsing to leave only the title, and expanding to display the full weights and broadcasts; the bottom has one-click evidence storage, encapsulating the card, weight graph, and timestamp into an audit package with Merkle hash, and appending the QR code to the timeline log to achieve an immutable chain. The entire process can be completed simultaneously on the same screen without switching pages, including risk identification, root cause location, voice verification, and one-click evidence storage, systematically solving the pain points of existing technologies such as lack of quantification, explanation, and auditing.

[0063] S204 After binding the warning vector corresponding to the interpretability warning card with the current outpatient clinic identifier, the warning event record is displayed in the first display interface in the form of a timeline. The warning event record includes at least the vaccine name, outpatient clinic identifier, timestamp, and color dynamic QR code.

[0064] In this embodiment, after the content of the interpretability warning card is confirmed, the warning vector generated based on the interpretability warning card is further bound to the current clinic's unique identifier to form an unalterable audit record. Next, the audit record is first written to the local database, and then pushed to the timeline log area at the bottom of the first display interface, presented in real-time as a colored scrolling bar. Each record contains at least the generic name of the vaccine, the clinic number, a timestamp, and a colored dynamic QR code. This QR code encapsulates the record's Merkle root hash and public blockchain TxID, ensuring that what is seen is what is verified.

[0065] The timeline features infinite scrolling and lazy loading, allowing for quick navigation by day, week, and month. Long-pressing any record brings up a side window displaying the Shapley weight graph, 3D super-edge animation frames, and original data fingerprint for the current alert, enabling review without switching pages. After regulators scan the QR code, the mobile app automatically redirects to a verification page, comparing the on-chain anchor value. If the hashes don't match, they are immediately highlighted in red, enabling rapid tamper detection and significantly improving regulatory efficiency and evidence credibility.

[0066] In one possible embodiment, the immunization program visualization monitoring provided in this application further includes: receiving a second instruction and generating a second display interface, wherein the second display interface includes a heat map layer constructed in real time using street-level polygons as base map units, the heat map layer including the heat map vectors of the daily vaccination doses of all clinics in the current street and the real-time inventory heat map vectors of the corresponding vaccines; performing element-level difference calculations on the heat map vectors of the daily vaccination doses and the real-time inventory heat map vectors to obtain a street heat map difference matrix; when the absolute value of any element in the street heat map difference matrix is ​​greater than a preset threshold, a warning float pops up at the edge of the layer; obtaining a click signal triggered by the warning float point, the second display interface automatically scales to the center coordinates of the unbalanced street, and calls the interface of the interpretable warning card to inject the unbalanced street identifier, vaccine name and corresponding difference vector into the interpretable warning card to complete the pop-up display.

[0067] In this solution, the second display interface uses a street polygon as its base, statically overlaying inoculation and inventory heat vectors on the same screen. The element-level difference matrix calculates the location of imbalances in seconds, and a red buoy pops up when any element exceeds the threshold. After clicking, the interface automatically zooms to the center of the imbalanced street, injects an interpretable warning card with one click, and uses Shapley weights and voice quantification contributions to quickly complete the closed loop between viewing, calculation, and interpretation, achieving the technical effects of simultaneous on-screen quantification, interpretability, and one-click auditing.

[0068] Based on the same inventive concept, embodiments of this application also provide an immunization program visualization monitoring device. For example... Figure 4 As shown, this is a schematic diagram of the structure of an immunization program visualization monitoring device, which may include: The receiving module 401 is used to receive the first instruction and generate a first display interface containing a multimodal dataset. The multimodal dataset includes at least vaccine inventory information, cold chain information, vaccination information, and outpatient video supervision information. Prediction module 402 is used to perform synchronous prediction on multimodal datasets, obtain the probability distribution of inventory gap and the distribution of expiration risk, and display the corresponding visualization layers in the first display interface; The early warning module 403 is used to pop up an interpretable early warning card in the first display interface when any distribution meets the preset early warning conditions. The early warning card is an interactive floating window that includes the vaccine name, remaining inventory, remaining days of expiration, key feature weight graphs, and a collapsible and expandable area. The display module 404 is used to bind the warning vector corresponding to the interpretable warning card to the current outpatient identifier and then display the warning event record in the form of a timeline in the first display interface. The warning event record includes at least the vaccine name, outpatient identifier, timestamp, and color dynamic QR code.

[0069] In one possible embodiment, the inventory information includes: net inflow, inventory, usage, and loss of immunization program vaccines or non-immunization program vaccines during the vaccine statistical period; and real-time vaccine inventory status of each outpatient department; wherein the vaccine information corresponding to immunization program vaccines and non-immunization program vaccines can be switched back and forth.

[0070] In one possible embodiment, the video supervision information includes real-time images from various outpatient video monitoring points that have been connected to the current system, wherein the monitoring points can be switched.

[0071] In one possible embodiment, the vaccination information includes: the vaccination rate of eight vaccines for children aged 2 to 6 years in each vaccination unit that has been connected to the current system; and the vaccination status on the day, including the recipients, the number of vaccinations, and the specific names of the vaccines administered; wherein the recipients include children, adults, dog bite victims, and tetanus infected individuals.

[0072] In one possible embodiment, the first display interface further includes: the number of people served on the day, the number of people served on the month, the number of people served throughout the year, and the number of people vaccinated on the day in the whole district; as well as the number of dog bite clinics, the number of adult clinics, the number of children's clinics, the number of obstetric clinics, and the number of tetanus clinics.

[0073] In one possible embodiment, the multimodal dataset is a multimodal dataset calibrated for spatiotemporal consistency. The calibration process includes: upon receiving a second instruction or detecting that the deviation between the inventory reporting frame timestamp and the cold chain temperature probe timestamp is greater than a preset threshold, a calibration progress bar overlay pops up at the bottom of the first display interface; wherein, the overlay uses the cold chain temperature probe timestamp as a reference to back-align the inventory reporting frame timestamp and displays the alignment progress percentage in real time; within the set alignment window, if the vaccine circulation time is greater than the inventory change allowable window, or the cold chain temperature is normal and the inventory drop is greater than a preset range, the overlay is stopped in the red zone and the spatiotemporal data source node to be corrected is highlighted with a red pulse until the anomaly is resolved or a shutdown instruction is received.

[0074] In one possible embodiment, the prediction module 402 is used to: obtain vaccine inventory, allocation volume, allocation relationship, remaining days of expiration, transportation mileage, and shared transportation route based on a multimodal dataset; construct an inventory propagation graph using vaccine inventory and allocation volume as input features for outpatient nodes and vaccine allocation relationship as hyperedges, and generate an inventory hidden vector through the first channel convolution of a dual-channel hypergraph neural network (Dual-HGNN); construct an expiration decay graph using remaining days of expiration and transportation mileage as input features for vaccine batch number nodes and shared transportation route as hyperedges, and generate an expiration hidden vector through the second channel convolution of a dual-channel hypergraph neural network (Dual-HGNN); input the inventory hidden vector and the expiration hidden vector into a cross-attention module, fuse them to obtain a joint hidden state, and present the fusion process in real time on the first display interface with a 3D hyperedge shrinkage animation; perform a one-time regression on the joint hidden state using a decoding head, and output the inventory gap probability distribution and the expiration risk distribution.

[0075] In one possible embodiment, the early warning module 403 is configured to: receive a second instruction and generate a second display interface, the second display interface including a heat map layer constructed in real time using street-level polygons as base map units, the heat map layer including the heat map vectors of the daily vaccination doses of all clinics in the current street and the real-time inventory heat map vectors of the corresponding vaccines; perform element-level difference calculations on the heat map vectors of the daily vaccination doses and the real-time inventory heat map vectors to obtain a street heat map difference matrix; when the absolute value of any element in the street heat map difference matrix is ​​greater than a preset threshold, an early warning float pops up at the edge of the layer; obtain a click signal triggered by the early warning float point, the second display interface automatically scales to the center coordinates of the unbalanced street, and calls the interface of the interpretable early warning card to inject the unbalanced street identifier, vaccine name and corresponding difference vector into the interpretable early warning card to complete the pop-up display.

[0076] In one possible embodiment, the early warning module 403 is configured to: when any distribution meets a preset early warning condition, display an interpretable early warning card with a slide-in animation in the lower right corner of the first display interface; inject an interactive Shapley value Sankey diagram component into the inner layer of the interpretable early warning card, the component displaying the marginal contribution percentage of each feature to the inventory shortage probability in a single flow bandwidth from the left feature node to the right inventory shortage probability node; the entire bandwidth flow is calculated by the Shapley algorithm based on the vaccine inventory, allocation, transportation days, and remaining expiration days; and inject a voice broadcast button into the top layer of the interpretable early warning card. The system includes features such as a voice broadcast button that, when clicked, broadcasts the vaccine name, shortage quantity, and suggested allocation path in natural language, supporting male / female voice / dialect switching; a one-click collapsible / expandable area is injected into the sidebar of the interpretability warning card, where only the title bar is retained in the collapsed state, and the Shapley value Sankey diagram component and voice broadcast button are fully displayed in the expanded state; and a one-click evidence storage button is injected into the bottom of the interpretability warning card. When clicked, this button encapsulates the current card content, a snapshot of the Shapley value Sankey diagram component, and the current timestamp into an audit package with Merkle hash, which is then appended to the timeline log of the first display interface.

[0077] The technical effects of the aforementioned immunization program visualization monitoring device can be referenced from the immunization program visualization monitoring method, and will not be elaborated here.

[0078] In some possible implementations, the immunization program visualization monitoring device according to this application may include at least a processor and a memory. The memory stores program code that, when executed by the processor, causes the processor to perform the steps in the immunization program visualization monitoring method according to various exemplary embodiments of this application described herein. For example, the processor may perform actions such as... Figure 2 The steps are shown in the figure.

[0079] Based on the same inventive concept, this application also provides an electronic device that can realize the functions of the aforementioned immunization program visualization monitoring method device. (Refer to...) Figure 5 Electronic devices include: At least one processor 501 and a memory 502 connected to at least one processor 501. In this embodiment, the specific connection medium between the processor 501 and the memory 502 is not limited. Figure 5 The example shown is the connection between processor 501 and memory 502 via bus 500. Bus 500 is... Figure 5 The connections between other components are indicated by thick lines and are for illustrative purposes only, not as limiting information. The Bus 500 can be divided into address bus, data bus, control bus, etc., for ease of representation. Figure 5The term 501 is represented by a single thick line, but this does not imply that there is only one bus or one type of bus. Alternatively, the processor 501 can also be called a controller; there is no restriction on the name.

[0080] In this embodiment, memory 502 stores instructions executable by at least one processor 501. By executing the instructions stored in memory 502, at least one processor 501 can perform the aforementioned immunization program visualization monitoring method. Processor 501 can implement... Figure 4 The functions of each module in the device shown.

[0081] The processor 501 is the control center of the device. It can connect to various parts of the control device through various interfaces and lines. By running or executing instructions stored in memory 502 and calling data stored in memory 502, the processor can perform various functions and process data, thereby monitoring the device as a whole.

[0082] In one possible design, processor 501 may include one or more processing units. Processor 501 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may also not be integrated into processor 501. In some embodiments, processor 501 and memory 502 may be implemented on the same chip; in some embodiments, they may also be implemented on separate chips.

[0083] Processor 501 can be a general-purpose processor, such as a central processing unit (CPU), digital signal processor, application-specific integrated circuit, field-programmable gate array or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the immunization program visualization monitoring method disclosed in the embodiments of this application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0084] Memory 502, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules. Memory 502 may include at least one type of storage medium, such as flash memory, hard disk, multimedia card, card-type memory, random access memory (RAM), static random access memory (SRAM), programmable read-only memory (PROM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), magnetic storage, magnetic disk, optical disk, etc. Memory 502 can be any other medium capable of carrying or storing desired program code in the form of instructions or data structures that can be accessed by a computer, but is not limited thereto. In the embodiments of this application, memory 502 can also be a circuit or any other device capable of implementing storage functions for storing program instructions and / or data.

[0085] By designing and programming the processor 501, the code corresponding to the immune planning visualization monitoring method described in the foregoing embodiments can be embedded into the chip, enabling the chip to execute the code during operation. Figure 2 The steps of the immune program visualization monitoring method shown in the embodiment are described below. How to design and program the processor 501 is a technique well-known to those skilled in the art and will not be elaborated upon here.

[0086] Based on the same inventive concept, embodiments of this application also provide a storage medium storing computer instructions that, when executed on a computer, cause the computer to perform the aforementioned method for visual monitoring of immune planning.

[0087] In some possible implementations, various aspects of the immunization program visualization monitoring method provided in this application can also be implemented in the form of a program product, which includes program code that, when the program product is run on a device, causes the control device to perform the steps in the immunization program visualization monitoring method according to the various exemplary embodiments of this application described above.

[0088] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0089] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0090] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0091] These computer program instructions can also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0092] 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 the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for visual monitoring of immunization programs, characterized in that, include: Upon receiving a first instruction, a first display interface containing a multimodal dataset is generated, wherein the multimodal dataset includes at least vaccine inventory information, cold chain information, vaccination information, and outpatient video supervision information; The multimodal dataset is simultaneously predicted to obtain the probability distribution of inventory shortage and the risk distribution of expiration date, and the corresponding visualization layers are displayed in the first display interface. When any distribution meets the preset warning conditions, an interpretable warning card pops up in the first display interface. The warning card is an interactive floating window that includes the vaccine name, remaining inventory, remaining days of expiration, key feature weight graphics, and a collapsible area. After binding the warning vector corresponding to the interpretable warning card with the current outpatient identifier, the warning event records are displayed in a scrolling timeline format on the first display interface. The warning event records include at least the vaccine name, outpatient identifier, timestamp, and color dynamic QR code.

2. The method as described in claim 1, characterized in that, The inventory information includes: During the vaccine statistics period, the net inflow, inventory, usage, and loss of vaccines included in the immunization program or non-immunization program; In addition, the real-time vaccine inventory status of each outpatient department; The vaccine information corresponding to the immunization program vaccines and the non-immunization program vaccines can be switched back and forth.

3. The method as described in claim 1, characterized in that, The vaccination information includes: The vaccination rates of eight vaccines for children aged 2-6 years at each vaccination unit already connected to the current system; and, The vaccination schedule for the day includes the recipients, the number of doses, and the specific names of the vaccines administered; the recipients include children, adults, dog bite victims, and tetanus patients.

4. The method as described in claim 1, characterized in that, The multimodal dataset is a multimodal dataset that has undergone spatiotemporal consistency calibration. The calibration process includes: Upon receiving the second instruction, or upon detecting that the deviation between the timestamp of the inventory reporting frame and the timestamp of the cold chain temperature probe is greater than a preset threshold, a calibration progress bar overlay will pop up at the bottom of the first display interface. The floating layer uses the timestamp of the cold chain temperature probe as a reference to back-align the timestamp of the inventory reporting frame and displays the alignment progress percentage in real time. Within the set alignment window, if the vaccine circulation time exceeds the inventory change allowable window, or if the cold chain temperature is normal and the inventory drops sharply beyond the preset range, the floating layer will be stopped in the red zone and highlighted with a red pulse to indicate the spatiotemporal data source node to be corrected, until the anomaly is resolved or a shutdown command is received.

5. The method as described in claim 1, characterized in that, The simultaneous prediction of the multimodal dataset to obtain the probability distribution of inventory shortage and the distribution of expiration risk includes: Based on the multimodal dataset, the vaccine inventory, allocation amount, allocation relationship, remaining days of expiration, transportation mileage, and shared transportation routes are obtained. Using the vaccine inventory and the allocation amount as the outpatient node input features and the vaccine allocation relationship as the hyperedge, an inventory propagation graph is constructed, and the inventory hidden vector is generated by the first channel convolution of the dual-channel hypergraph neural network Dual-HGNN. Using the remaining days of the expiration date and the transportation mileage as input features for the vaccine batch number node and the shared transportation path as a hyperedge, an expiration date decay map is constructed, and an expiration date hidden vector is generated by the second channel convolution of the dual-channel hypergraph neural network Dual-HGNN. The inventory hidden vector and the expiration date hidden vector are input into the cross-attention module and fused to obtain a joint hidden state. The fusion process is presented in real time on the first display interface with a 3D hyperedge shrinkage animation. The joint hidden state is regressed once by the decoding head to output the probability distribution of the inventory gap and the distribution of the expiration risk.

6. The method as described in claim 1, characterized in that, Also includes: Upon receiving a second instruction, a second display interface is generated. The second display interface includes a heat map layer constructed in real time using street-level polygons as base map units. The heat map layer includes the heat vector of the daily vaccination doses of all clinics in the current street and the real-time inventory heat vector of the corresponding vaccines. Element-level difference calculations are performed on the heat vector of the daily vaccination dose and the heat vector of the real-time inventory to obtain the street heat difference matrix; When the absolute value of any element in the street thermal difference matrix is ​​greater than a preset threshold, a warning float will pop up at the edge of the layer. Upon receiving a click signal triggered by the warning buoy point, the second display interface automatically scales to the center coordinates of the unbalanced street and calls the interface of the interpretable warning card to inject the unbalanced street identifier, vaccine name, and corresponding difference vector into the interpretable warning card, thus completing the pop-up display.

7. The method as described in claim 1, characterized in that, When any distribution meets the preset warning conditions, an interpretable warning card pops up in the first display interface, including: When any distribution meets the preset warning conditions, the interpretable warning card will pop up in the lower right corner of the first display interface with a sliding animation. An interactive Shapley value Sankey diagram component is injected into the inner layer of the interpretable early warning card. The component displays the marginal contribution percentage of each feature to the inventory gap probability in one go, in the form of the flow bandwidth from the left feature node to the right inventory gap probability node. The entire bandwidth flow is calculated by the Shapley algorithm based on the vaccine inventory, allocation, transportation days, and remaining expiration days. A voice broadcast button is injected into the top layer of the interpretability warning card. When the voice broadcast button is clicked, the vaccine name, shortage quantity, and suggested allocation path are broadcast in natural language template, and multiple voice switching is supported. The one-click folding / expanding area is injected into the sidebar of the interpretable warning card. In the folded state, only the title bar is retained, and in the expanded state, the Shapley value Sankey diagram component and the voice broadcast button are fully displayed. A one-click evidence storage button is injected into the bottom of the interpretability warning card. When the one-click evidence storage button is clicked, the current card content, Shapley value Sankey diagram component snapshot, and current timestamp are encapsulated into an audit package with Merkle hash and appended to the timeline log of the first display interface.

8. An immunization program visualization monitoring device, characterized in that, include: The receiving module is used to receive a first instruction and generate a first display interface containing a multimodal dataset, wherein the multimodal dataset includes at least vaccine inventory information, cold chain information, vaccination information and outpatient video supervision information; The prediction module is used to simultaneously predict the multimodal dataset, obtain the probability distribution of inventory gap and the risk distribution of expiration date, and display the corresponding visualization layers in the first display interface; The early warning module is used to pop up an interpretable early warning card in the first display interface when any distribution meets the preset early warning conditions. The early warning card is an interactive floating window that includes the vaccine name, remaining inventory, remaining days of expiration, key feature weight graphics, and a collapsible and expandable area. The display module is used to bind the warning vector corresponding to the interpretable warning card with the current outpatient identifier, and then display the warning event record in the form of a timeline in the first display interface. The warning event record includes at least the vaccine name, outpatient identifier, timestamp, and color dynamic QR code.

9. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory stores program code that, when executed by the processor, causes the processor to perform the method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Includes program code that, when the storage medium is run on an electronic device, causes the electronic device to perform any of the methods described in claims 1 to 7.

Images