Methods, devices, equipment, and media for generating dynamic reports based on medical data

Abstract

This application discloses a method, apparatus, device, and medium for generating dynamic reports based on medical data, relating to the field of report processing technology. The method includes: acquiring user interaction instructions and report configuration information; assigning weights to data dimensions according to time granularity to obtain corresponding dynamic weights; inputting the report configuration information and dynamic weights into a cross-dimensional correlation engine to generate multimodal data; judging the multimodal data according to the warning level to obtain a judgment result; when the judgment result indicates that the inventory consumption data in the multimodal data exceeds the preset threshold of the corresponding warning level, generating risk-marked data; generating an initial report based on the risk-marked data and multimodal data; adjusting the initial report according to user interaction instructions to obtain a target report; and achieving automatic generation and synchronous warning of multi-dimensional linked reports, improving hospital operational decision-making efficiency and reducing inventory management risks.

Description

Technical Field

[0001] This invention relates to the field of report processing technology, and in particular to a method, apparatus, equipment and medium for generating dynamic reports based on medical data. Background Technology

[0002] Traditional medical data reporting and analysis methods often employ static reports or fixed-dimensional statistics, relying heavily on fixed analytical dimensions and weights. This fails to dynamically adjust the importance of each data dimension based on different time granularities (e.g., hourly, daily, weekly, monthly), leading to delayed or distorted results and hindering adaptability to dynamic analysis needs across different time periods, departments, and business scenarios. Furthermore, given the complex and ever-changing medical service processes, single-dimensional data analysis is insufficient for refined management and cross-dimensional insights. Therefore, a dynamic report generation method is urgently needed to dynamically generate multi-dimensional, interconnected reports and simultaneously provide inventory alerts. Summary of the Invention

[0003] The main purpose of this application is to provide a method, device, equipment and medium for generating dynamic reports based on medical data, aiming to solve the technical problem of how to dynamically generate multi-dimensional linked reports and simultaneously complete inventory early warning.

[0004] To achieve the above objectives, this application proposes a method for generating dynamic reports based on medical data, including:

[0005] Obtain user interaction commands and report configuration information, wherein the report configuration information includes time granularity, data dimension set and warning level, the time granularity includes hourly granularity, daily granularity, weekly granularity and monthly granularity, and the data dimensions include registration data dimension, billing data dimension and drug inventory dimension;

[0006] Based on the dynamic weighted aggregation algorithm, weights are assigned to the registration data dimension, fee data dimension and drug inventory dimension in the data dimension according to the time granularity to obtain the corresponding dynamic weights;

[0007] The report configuration information is combined with the dynamic weights and input into the cross-dimensional association engine to generate multimodal data, wherein the multimodal data includes registration data, revenue data and inventory consumption data, and the cross-dimensional association engine is built based on the registration data dimension, the fee data dimension and the drug inventory dimension.

[0008] The multimodal data is judged based on the warning level to obtain the judgment result;

[0009] When the judgment result indicates that the inventory consumption data in the multimodal data exceeds the preset threshold of the corresponding early warning level, risk marker data is generated;

[0010] An initial report is generated based on the risk labeling data and the multimodal data;

[0011] The initial report is adjusted according to the user interaction instructions to obtain the target report.

[0012] In one embodiment, the step of assigning weights to the registration data dimension, fee data dimension, and drug inventory dimension in the data dimension according to the time granularity based on a dynamic weighted aggregation algorithm to obtain the corresponding dynamic weights includes:

[0013] Departmental business priorities are obtained based on the aforementioned registration data dimensions;

[0014] The types of chargeable items are obtained based on the aforementioned cost data dimensions;

[0015] Drug efficacy data is obtained based on the aforementioned drug inventory dimension;

[0016] Based on the data dimensions and the time granularity, an initial first weight coefficient, an initial second weight coefficient, and an initial third weight coefficient are determined, wherein the registration data dimension corresponds to the initial first weight coefficient, the cost data dimension corresponds to the initial second weight coefficient, and the drug inventory dimension corresponds to the initial third weight coefficient.

[0017] The initial first weight coefficient is dynamically adjusted based on the department's business priority to obtain the first weight coefficient.

[0018] The initial second weight coefficient is adjusted in stages according to the type of the chargeable item to obtain the second weight coefficient.

[0019] The initial third weighting coefficient is compensated for by combining the drug expiration date data to obtain the third weighting coefficient.

[0020] The first weight coefficient, the second weight coefficient, and the third weight coefficient are used as the dynamic weights.

[0021] In one embodiment, before the step of combining the report configuration information with the dynamic weights and inputting them into a cross-dimensional correlation engine to generate multimodal data, wherein the multimodal data includes registration data, revenue data, and inventory consumption data, and the step of constructing the cross-dimensional correlation engine based on the registration data dimension, the fee data dimension, and the drug inventory dimension, the following steps are included:

[0022] A time series analysis model is constructed based on the aforementioned registration data dimensions;

[0023] Construct a department-fee item association matrix based on the aforementioned fee data dimensions;

[0024] Combining the drug inventory dimension and the registration data dimension, the departmental consumption correlation of each drug is calculated based on a graph neural network to construct an inventory consumption network graph, wherein the network nodes of the inventory consumption graph represent the drug inventory status and the edge weights represent the consumption rate.

[0025] By combining the time series analysis model, the department-fee item association matrix, and the inventory consumption network diagram, a cross-dimensional association engine is obtained.

[0026] In one embodiment, the step of combining the report configuration information with the dynamic weight input to the cross-dimensional association engine to generate multimodal data includes:

[0027] Input the report configuration information into the time series analysis model of the cross-dimensional correlation engine to generate registration volume data;

[0028] The report configuration information is input into the department-fee item association matrix of the cross-dimensional association engine and the revenue data is obtained by hierarchical clustering algorithm.

[0029] The report configuration information is input into the inventory consumption network diagram of the cross-dimensional association engine for calculation to obtain inventory consumption data;

[0030] Based on the dynamic weights, the registration data, revenue data and inventory consumption data are spatiotemporally aligned and normalized to obtain initial multimodal data.

[0031] Metadata tags are added to the initial multimodal data to obtain multimodal data. The metadata tags include data type, generation time, and weight allocation information.

[0032] In one embodiment, the step of generating risk marker data when the determination result is that the inventory consumption data in the multimodal data exceeds a preset threshold for the corresponding early warning level includes:

[0033] Based on the warning level, a level identifier and a corresponding threshold table are obtained, wherein the threshold table includes a first threshold, a second threshold and a third threshold, the first threshold being less than the second threshold and the second threshold being less than the third threshold;

[0034] The current consumption ratio is obtained based on the inventory consumption data in the multimodal data;

[0035] The current warning level is determined based on the current consumption ratio.

[0036] Based on the current warning level, the target threshold is retrieved from the threshold table to obtain the matching threshold;

[0037] When the current consumption ratio is greater than the matching threshold, an over-limit flag is recorded;

[0038] An event number is generated based on the over-limit flag, and the event number, the current consumption ratio, the matching threshold, and the level identifier are combined into risk labeling data.

[0039] In one embodiment, the step of generating an initial report based on the risk marker data and the multimodal data includes:

[0040] The multimodal data is sorted according to time granularity to obtain a time-series data sequence;

[0041] The risk-labeled data is time-aligned with the time-series data sequence to obtain a labeled data set;

[0042] A first chart is generated based on the data set, and the first chart is used to show the trend of the number of registrations over time;

[0043] A second chart is generated based on the data set, and the second chart is used to display the cumulative distribution of revenue data;

[0044] A third chart is generated based on the data set. The third chart is used to display inventory consumption data and highlight the time period with the risk-marked data.

[0045] The first chart, the second chart, and the third chart are combined according to a preset layout rule to obtain an initial report.

[0046] In one embodiment, the step of adjusting the initial report according to the user interaction instruction to obtain the target report includes:

[0047] The user interaction commands are parsed to extract the adjustment type and adjustment parameters. The adjustment type includes chart style adjustment, time range adjustment, and dimension increase / decrease adjustment.

[0048] If the adjustment type is a chart style adjustment, then the chart in the initial report is replaced according to the adjustment parameters to obtain a style-updated chart;

[0049] If the adjustment type is a time range adjustment, then the multimodal data in the initial report is refiltered according to the adjustment parameters to obtain refiltered data;

[0050] If the adjustment type is a dimension increase / decrease adjustment, then the specified dimension is added or removed from the multimodal data in the initial report according to the adjustment parameters to obtain the dimension adjustment data;

[0051] The style-updated chart, the re-filtered data, or the dimension-adjusted data are remapped to the report layout of the initial report to obtain an intermediate report;

[0052] The intermediate report is rendered and cached to obtain the target report.

[0053] Furthermore, to achieve the above objectives, this application also proposes a dynamic report generation device based on medical data, the dynamic report generation device based on medical data comprising:

[0054] The data acquisition module is used to acquire user interaction commands and report configuration information. The report configuration information includes time granularity, data dimension set and warning level. The time granularity includes hourly granularity, daily granularity, weekly granularity and monthly granularity. The data dimensions include registration data dimension, billing data dimension and drug inventory dimension.

[0055] The allocation module is used to allocate weights to the registration data dimension, fee data dimension and drug inventory dimension in the data dimension according to the time granularity based on the dynamic weighted aggregation algorithm, so as to obtain the corresponding dynamic weights.

[0056] The data generation module is used to combine the report configuration information with the dynamic weight input to the cross-dimensional association engine to generate multimodal data, wherein the multimodal data includes registration data, revenue data and inventory consumption data, and the cross-dimensional association engine is built based on the registration data dimension, the fee data dimension and the drug inventory dimension.

[0057] The judgment module is used to judge the multimodal data according to the warning level and obtain the judgment result;

[0058] The execution module is used to generate risk labeling data when the judgment result is that the inventory consumption data in the multimodal data exceeds the preset threshold of the corresponding early warning level;

[0059] A construction module is used to generate an initial report based on the risk labeling data and the multimodal data;

[0060] The results module is used to adjust the initial report according to the user interaction instructions to obtain the target report.

[0061] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the dynamic report generation method based on medical data as described above.

[0062] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the dynamic report generation method based on medical data as described above.

[0063] This application acquires user interaction commands and report configuration information, and uses a dynamic weighted aggregation algorithm to assign weights to the registration data, fee data, and drug inventory dimensions based on time granularity. This results in corresponding dynamic weights. The report configuration information, combined with these dynamic weights, is input into a cross-dimensional correlation engine to generate multimodal data. The multimodal data is then judged according to the warning level. When the judgment result indicates that the inventory consumption data in the multimodal data exceeds the preset threshold for the corresponding warning level, risk-marked data is generated. An initial report is generated based on the risk-marked data and the multimodal data. The initial report is then adjusted according to user interaction commands to obtain the target report. By using user configuration commands, dynamic weights, and a cross-dimensional correlation engine to aggregate registration, fee, and inventory data in real time, and instantly marking inventory risks according to warning levels and embedding them into reports to dynamically obtain report information, this application achieves automatic generation and synchronous warning of multi-dimensional linked reports, improving the efficiency of hospital operational decision-making and reducing inventory management risks. Attached Figure Description

[0064] To more clearly illustrate the technical solutions in the embodiments of 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, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0065] Figure 1 This is a flowchart illustrating the first embodiment of the dynamic report generation method based on medical data in this application;

[0066] Figure 2 This is a flowchart illustrating the second embodiment of the dynamic report generation method based on medical data in this application;

[0067] Figure 3 This is a flowchart illustrating the third embodiment of the dynamic report generation method based on medical data in this application;

[0068] Figure 4 This is a schematic diagram of the module structure of the dynamic report generation device based on medical data according to the first embodiment of the dynamic report generation method based on medical data of this application;

[0069] Figure 5 This is a schematic diagram of the hardware operating environment involved in the dynamic report generation method based on medical data in this application embodiment.

[0070] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0071] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0072] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0073] With the continuous advancement of healthcare informatization, hospitals have an increasing demand for real-time monitoring of outpatient operations. Traditional medical data reporting and analysis methods often employ static reports or fixed-dimensional statistics, relying heavily on fixed analytical dimensions and weights. This fails to dynamically adjust the importance of each data dimension based on different time granularities (e.g., hours, days, weeks, months), leading to delayed or distorted analysis results and making it difficult to adapt to the dynamic analysis needs of different time periods, departments, and business scenarios. Furthermore, given the complex and ever-changing healthcare service processes, single-dimensional data analysis can no longer meet the needs for refined management and cross-dimensional insights. If managers continue to rely on traditional reporting systems with scheduled batch processing, static weights, and fragmented dimensions, they will be unable to gain real-time insight into the cascading risks of "surging registrations—increased fees—inventory shortages."

[0074] Therefore, this application proposes a dynamic report generation method based on medical data to solve the above-mentioned problems. The main solution of this application embodiment is as follows: by obtaining user interaction instructions and report configuration information, a dynamic weighted aggregation algorithm is used to assign weights to the registration data dimension, cost data dimension, and drug inventory dimension in the data dimension according to the time granularity, obtaining the corresponding dynamic weights. The report configuration information is combined with the dynamic weights and input into a cross-dimensional association engine to generate multimodal data. The multimodal data is judged according to the warning level to obtain the judgment result. When the judgment result is that the inventory consumption data in the multimodal data exceeds the preset threshold of the corresponding warning level, risk labeling data is generated. An initial report is generated based on the risk labeling data and the multimodal data. The initial report is adjusted according to the user interaction instructions to obtain the target report.

[0075] Based on the above, this application also provides a method for generating dynamic reports based on medical data, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the dynamic report generation method based on medical data according to this application. In this embodiment, the dynamic report generation method based on medical data includes steps S10 to S70:

[0076] Step S10: Obtain user interaction commands and report configuration information.

[0077] It should be noted that user operation commands are received through a graphical interface or API, including but not limited to querying time ranges, focusing on business modules, and specifying visualization style preferences, while also retrieving preset report configuration information. Report configuration information includes time granularity, data dimension sets, and alert levels. Time granularity defines the time unit for data aggregation, including hourly, daily, weekly, and monthly granularity to adapt to the trend analysis and periodic judgment needs of different business scenarios. Data dimensions include registration data, billing data, and drug inventory, corresponding to hospital outpatient operations, financial revenue and expenditure, and material management, respectively. These three together constitute the core indicator system for hospital operations. Alert levels are used to set the sensitivity level for risk identification, typically divided into low, medium, and high levels, corresponding to different threshold standards and response strategies.

[0078] Step S20: Based on the dynamic weighted aggregation algorithm, weights are assigned to the registration data dimension, cost data dimension and drug inventory dimension in the data dimension according to the time granularity to obtain the corresponding dynamic weights.

[0079] It's important to note that this process doesn't use fixed weights. Instead, it dynamically adjusts the contribution of each data dimension to the overall analysis based on different time granularities (e.g., hourly, daily, weekly, or monthly) and the changing importance of each data dimension in specific business scenarios. Specifically, at high-frequency time granularities (e.g., hourly), the focus is on real-time operational status, so the weight of registration data is appropriately increased to reflect fluctuations in outpatient traffic. At low-frequency granularities (e.g., monthly), cost data and drug inventory data may play a more important analytical role in assessing the hospital's overall revenue and material reserves. Furthermore, the dynamic weighted aggregation algorithm adaptively adjusts the initial weights by incorporating auxiliary information such as departmental business priorities, fee types, and drug expiration dates. For example, emergency department registration data has a higher priority than general outpatient data, cost data should be given greater weight in financial analysis, and near-expiration drug inventory data needs higher sensitivity in the early warning module.

[0080] Furthermore, step S20 also includes: obtaining departmental business priorities based on registration data, obtaining chargeable item types based on cost data, and obtaining drug expiration date data based on drug inventory. Specifically, the registration data dimension covers information such as the registration department, doctor, registration type (e.g., general appointment, specialist appointment), and registration time. By analyzing this data, the system can calculate the workload of each department, thereby obtaining departmental business priorities. For example, a surge in registrations in the emergency department within a short period indicates a higher business priority; similarly, a significant increase in registrations for certain specialties during specific time periods (e.g., holidays) will also correspondingly increase their priority. The cost data dimension includes chargeable item names, chargeable amounts, and chargeable times. By classifying and analyzing chargeable items, the system can identify different types of chargeable items, such as examination fees, treatment fees, and drug fees. These item types reflect the hospital's revenue structure and business priorities. In addition, the drug inventory dimension involves information such as drug name, inventory quantity, expiration date, and warehousing time. By analyzing this data, drug expiration date data can be obtained, i.e., the expiration date and inventory turnover of drugs. For example, the rate of inventory depletion of near-expiry drugs needs to be closely monitored to avoid waste due to expiration. Next, based on the data dimensions and time granularity, initial first, second, and third weighting coefficients are determined. The initial first weighting coefficient corresponds to the registration data dimension, the initial second weighting coefficient to the cost data dimension, and the initial third weighting coefficient to the drug inventory dimension. Specifically, the choice of time granularity (e.g., hour, day, week, month) is crucial for weight allocation. The system assigns initial weighting coefficients to the registration data, cost data, and drug inventory dimensions based on the coarseness of the time granularity. For example, for hourly granularity, registration data has stronger real-time characteristics, thus receiving a higher initial weighting coefficient; while for monthly granularity, the cumulative effect of cost data is more significant, and the initial weights are adjusted accordingly. Then, the initial first weighting coefficient is dynamically adjusted based on departmental business priorities to obtain the first weighting coefficient. Specifically, combining the current time granularity (e.g., hourly or daily) and business fluctuations, a sliding window algorithm is used to dynamically adjust the initial first weighting coefficient to obtain the first weighting coefficient, with the specific formula as follows:

[0081]

[0082] in, This represents the initial first weight coefficient. As a regulating factor, This represents the time series fluctuation function. The initial second weighting coefficient is adjusted in stages according to the type of fee item to obtain the second weighting coefficient. Specifically, fee items are divided into basic services (such as registration fees), high-value consumables (such as surgical materials), and medical insurance reimbursement items, etc. Different adjustment coefficient functions are set for different fee item types, and the initial second weighting coefficient is adjusted accordingly, as shown below:

[0083]

[0084] in This represents the initial second weighting coefficient. This represents the correction coefficient function for the fee item type. The initial third weight coefficient is compensated for timeliness based on drug expiration date data to obtain the third weight coefficient. Specifically, based on the remaining shelf life of the drug, an exponential decay function is used to compensate for the timeliness of the initial third weight coefficient, ensuring that drugs nearing their expiration date receive higher sensitivity in the early warning module. This is specifically expressed as follows:

[0085]

[0086] in, This refers to the expiration date of the medication. For the current time, The first, second, and third weighting coefficients are used as attenuation coefficients. Finally, these coefficients are used as dynamic weights for subsequent cross-dimensional correlation modeling and multimodal data fusion analysis, improving the accuracy and practicality of the report results.

[0087] Step S30: Combine the report configuration information with dynamic weights and input them into the cross-dimensional association engine to generate multimodal data.

[0088] It should be noted that the cross-dimensional correlation engine is built upon registration data, billing data, and drug inventory dimensions, possessing multi-source data fusion, dynamic weighted modeling, and cross-modal reasoning capabilities. Specifically, a time series analysis model is constructed based on the registration data dimension. This model extracts information such as the number of patients and appointment ratios for each department and time period from historical registration records. Classic time series modeling methods such as ARIMA, LSTM, or Prophet are used to establish outpatient traffic prediction models at different time granularities (hourly, daily, and weekly). This model can not only generate registration volume data for the current period but also help identify peak periods and abnormal fluctuations, providing a basis for subsequent resource scheduling. Secondly, a department-billing item correlation matrix is ​​constructed based on the billing data dimension. Specifically, the relationship between each hospital department and its corresponding billing item (such as examination fees, treatment fees, and drug fees) is represented in matrix form. Rows represent departments, columns represent billing items, and element values ​​represent the business share or monetary contribution of a particular department to a certain type of billing item. By employing methods such as hierarchical clustering or principal component analysis (PCA), the financial similarities and differences between departments can be further explored for the classification, statistics, and trend analysis of revenue data. Next, combining the dimensions of drug inventory and registration data, a graph neural network is used to calculate the departmental consumption correlation of each drug, constructing an inventory consumption network graph. In this graph, network nodes represent drug inventory status (e.g., inventory quantity, expiration date), edges represent the flow of drugs between different departments, and edge weights represent consumption rates. Through graph convolutional networks (GCNs), the usage patterns of drugs in different departments can be learned, identifying frequently consumed drugs and potential shortage risk points. Finally, the time series analysis model, the department-fee item correlation matrix, and the inventory consumption network graph are combined to obtain a cross-dimensional correlation engine. This engine receives configuration parameters and dynamic weights through a unified data interface layer, calls various sub-models, and performs collaborative calculations to achieve spatiotemporal alignment and semantic fusion of registration, fee, and inventory data. The multimodal data output by the engine includes both single-dimensional trend information and cross-dimensional correlation features, thus providing high-quality data support for intelligent report generation, risk warning, and decision support.

[0089] Through the above mechanism, the cross-dimensional correlation engine can output structured multimodal data, including registration data, revenue data, and inventory consumption data. The three are aligned in the time and business dimensions and are accompanied by metadata tags for subsequent visualization, risk warning, and report generation.

[0090] Step S40: Determine the multimodal data according to the warning level and obtain the judgment result.

[0091] It should be noted that this process typically involves assessing data from different dimensions based on predefined warning levels to identify potential risks or opportunities and take appropriate measures accordingly. A series of warning levels are defined based on business needs and historical data analysis. These levels can be simple binary classifications (e.g., normal, abnormal) or more granular grading (e.g., low risk, medium risk, high risk, emergency). Each level corresponds to a specific threshold range, which can be determined based on key indicators such as the number of registrations, fluctuations in fees, and inventory consumption data. In this embodiment, inventory consumption data is used as the indicator for assessment. If the inventory of certain drugs exceeds the safety stock level and the consumption rate remains consistently low, a "low risk" warning may be triggered, indicating a risk of inventory backlog.

[0092] Secondly, by collecting and processing the latest data from time series analysis models, department-fee item correlation matrices, and inventory consumption network diagrams, this data is mapped to pre-set warning levels. Through machine learning algorithms, the system can not only identify which warning level the current state belongs to, but also predict the possible trend of state changes in the future.

[0093] Step S50: When the judgment result is that the inventory consumption data in the multimodal data exceeds the preset threshold of the corresponding early warning level, risk labeling data is generated.

[0094] It should be noted that a threshold table is obtained based on the user-configured alert levels (e.g., low risk, medium risk, high risk), where each level corresponds to one or more key indicator thresholds. Subsequently, the current drug inventory status is compared with each alert threshold in real time. If the actual inventory of a certain type of drug is detected to be lower than the set safety threshold, or its consumption rate is abnormally high and exceeds the normal fluctuation range, it is determined to be an out-of-limit event and risk labeling data is generated.

[0095] Further, step S50 also includes: obtaining a level identifier and corresponding threshold table based on the warning level, wherein the threshold table includes a first threshold, a second threshold, and a third threshold, the first threshold being less than the second threshold, and the second threshold being less than the third threshold. Specifically, a level identifier and corresponding threshold table is established according to the warning level configured by the user to define the risk standards for different levels. For example, three warning levels are set: low risk, medium risk, and high risk, and the corresponding thresholds are the first threshold, the second threshold, and the third threshold, respectively. Second threshold and the third threshold Satisfying the relationship These thresholds are typically calculated based on parameters such as the drug's safety stock level, historical consumption rate, and expiration date. Then, the current consumption ratio is obtained from the inventory consumption data in the multimodal data. This is the ratio between the current consumption and the baseline consumption, used to measure whether drug use is within the normal range. Next, the current warning level is determined based on the current consumption ratio. Specifically, the target threshold is retrieved from the threshold table based on the current warning level to obtain the matching threshold. When the current consumption ratio exceeds the matching threshold, an exceedance flag is recorded. In detail, the current consumption ratio... By comparing it with the threshold values ​​for each level, the current warning level is determined, as shown below:

[0096]

[0097] After determining the current warning level, the system looks up the matching threshold for that level in the threshold table. and the ratio of actual consumption When comparing, When an abnormal consumption of a drug is detected, it is marked as exceeding the limit, indicating abnormal consumption behavior at the current time granularity. An event number is then generated based on the exceeding limit flag, and the event number, current consumption ratio, matching threshold, and level identifier are combined to form risk labeling data. This data can be used not only to highlight abnormal situations in reports using colors and icons, but also as a log record for subsequent analysis, auditing, and automatic replenishment strategy invocation.

[0098] Step S60: Generate an initial report based on risk labeling data and multimodal data.

[0099] It's important to note that the multimodal data encompasses multiple dimensions, including registration volume, revenue, and inventory consumption. This data, generated through a cross-dimensional correlation engine, comprehensively reflects the hospital's operational status. Risk-labeled data, on the other hand, is generated based on the multimodal data and according to warning levels. It identifies abnormal data conditions, such as inventory consumption exceeding preset thresholds. This process aims to integrate structured business data with risk warning information and present it to users in a visual manner, helping them quickly understand the hospital's operational status and make informed decisions.

[0100] Further, step S60 also includes: sorting the multimodal data by time granularity to obtain a time-series data sequence, and then aligning the risk-labeled data with the time-series data sequence to obtain a labeled data set. Specifically, the multimodal data (including registration volume, revenue, inventory consumption, etc.) obtained from the cross-dimensional correlation engine is sorted by time granularity to form a time-series data sequence; simultaneously, the previously generated risk-labeled data is aligned with these time-series data to ensure that each risk event can accurately correspond to the corresponding time period or data point, thereby constructing a labeled data set containing normal data and abnormal indicators. Next, a first chart is generated based on the data set. The first chart is used to display the trend of registration volume data over time. Specifically, this chart is usually displayed in the form of a line chart, where the X-axis represents the time dimension (such as date, week, or month), and the Y-axis represents the number of registrations. Through this intuitive visualization method, medical institutions can quickly identify peak and off-peak periods for medical visits, providing a basis for resource allocation. For example, when the number of registrations increases significantly during the flu season, hospitals can arrange more medical staff in advance. Secondly, a second chart is generated based on the dataset. This second chart displays the cumulative distribution of revenue data and can use stacked area charts or bar charts to show the revenue contribution of different departments or projects. This chart design not only helps in understanding overall financial performance but also provides in-depth insights into the specific contribution ratio of each department to total revenue. For example, by analyzing this data, management can identify which services are most popular and which aspects may need improvement or promotion, thereby optimizing resource allocation to improve profitability. Furthermore, a third chart is generated based on the dataset. This third chart displays inventory consumption data and highlights periods with risk indicators. Considering the importance and special nature of medical supplies, the third chart typically uses heatmaps or color-coded bar charts. Under normal circumstances, inventory levels remain within a safe range. However, once the inventory of certain medicines or supplies approaches the warning line, the corresponding time period is highlighted in red or otherwise prominently displayed. This allows supply chain managers to take timely measures to prevent service disruptions due to shortages. In addition, by comparing historical data, future demand trends can be predicted, further enhancing the effectiveness and foresight of inventory management.

[0101] Finally, the first, second, and third charts are stitched together according to preset layout rules to obtain the initial report. Specifically, during the stitching process, the size and position of each chart are automatically adjusted according to the preset layout rules. For example, the registration volume trend chart may be set to occupy a large area in the upper left corner of the report to highlight its importance; the cumulative revenue distribution chart is placed in the upper right corner to contrast with the registration volume trend chart, making it easy for users to view the changing trends of two key indicators simultaneously. The inventory consumption chart is placed at the bottom of the report, occupying a wide horizontal area to show the dynamic changes in inventory in detail. In addition to stitching the charts, the information in the report is also integrated and optimized. For example, a title bar is added to the top of the report to display basic information such as the report name, generation time, and time granularity; a footer is added to the bottom of the report to display auxiliary information such as data source and report version. In addition, highlights or special annotations are added to the charts based on risk-marked data to remind users to pay attention to potential risk points. The final generated initial report not only contains rich data information, but also highlights key data and risk points through reasonable layout and highlighting, providing hospital managers with comprehensive and intuitive decision support.

[0102] Step S70: Adjust the initial report according to the user interaction instructions to obtain the target report.

[0103] It's important to note that user interaction commands are parsed to extract the adjustment type and parameters. Adjustment types may include changes to chart styles (e.g., switching from a bar chart to a line chart), adjustments to the time range (e.g., switching from monthly to daily granularity), and additions or subtractions of data dimensions (e.g., adding or removing registration data for a specific department). Adjustment parameters specify the details of the adjustment, such as the new time range or newly added dimension fields. Then, based on the parsed adjustment type, corresponding operations are executed. For example, if the user wants to change the chart style, the corresponding chart in the initial report will be replaced according to the adjustment parameters, generating a new chart style. If the user adjusts the time range, the multimodal data will be re-filtered, updated data will be generated based on the new time range, and the relevant charts will be re-rendered. If the user needs to add or remove data dimensions, the specified dimensions will be added or removed from the multimodal data according to the adjustment parameters, thereby updating the data display in the report. Through this series of steps, the initial report can be flexibly adjusted according to user interaction commands to generate a target report that better meets the user's needs.

[0104] This embodiment acquires user interaction commands and report configuration information, and assigns weights to the registration data, fee data, and drug inventory dimensions based on time granularity using a dynamic weighted aggregation algorithm. This yields corresponding dynamic weights. The report configuration information, combined with these dynamic weights, is input into a cross-dimensional correlation engine to generate multimodal data. The multimodal data is then judged according to the warning level. When the judgment result indicates that the inventory consumption data in the multimodal data exceeds the preset threshold for the corresponding warning level, risk-marked data is generated. An initial report is generated based on the risk-marked data and the multimodal data. The initial report is then adjusted according to user interaction commands to obtain the target report. By using user configuration commands, dynamic weights, and a cross-dimensional correlation engine to aggregate registration, fee, and inventory data in real time, and instantly marking inventory risks according to warning levels and embedding them into reports to dynamically acquire report information, this achieves automatic generation and synchronous warning of multi-dimensional linked reports, improving hospital operational decision-making efficiency and reducing inventory management risks.

[0105] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in Embodiment 1 above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 The dynamic report generation method based on medical data, step S30, further includes steps S201 to S205:

[0106] Step S201: Input the report configuration information into the time series analysis model of the cross-dimensional correlation engine to generate registration volume data.

[0107] It's important to note that the time series analysis model is a core component of the cross-dimensional correlation engine, specifically designed for processing time-related data. Based on the input report configuration information, it can extract data related to registration volume from medical data sources and aggregate and analyze it according to a specified time granularity. For example, if the time granularity is set to "daily granularity," the model will extract daily registration volume data and calculate key indicators such as the total daily registration volume and the distribution of registration volume across departments.

[0108] Furthermore, in generating registration volume data, the model considers various factors, such as holiday effects, seasonal variations, and hospital scheduling. For example, on holidays or weekends, the number of registrations in some departments may increase significantly, while others may see relatively fewer. By incorporating these dynamic factors, the model can more accurately predict and reflect actual changes in registration volume. In addition, the model smooths the data to reduce the impact of short-term fluctuations, thus more clearly displaying long-term trends. For example, through techniques such as moving averages or exponential smoothing, the model can generate a smoother registration volume curve, helping users better understand the overall trend of the data. Finally, the registration volume data generated by the time series analysis model will be output in a structured form, including information such as total daily registrations, registrations by department, and registration volume trends. This data can not only be directly used for report generation but also serve as input for other analytical modules (such as revenue analysis and inventory consumption analysis) to achieve multi-dimensional data linkage analysis.

[0109] Step S202: Input the report configuration information into the department-fee item association matrix of the cross-dimensional association engine and calculate the revenue data through hierarchical clustering algorithm.

[0110] It's important to note that the Department-Billing Item Association Matrix is ​​a core component of the cross-dimensional association engine, specifically designed for handling revenue-related data. It constructs a matrix structure with departments as rows and billing items as columns. Each element in the matrix represents the amount charged by a department for a specific billing item within a specified time granularity. For example, if the time granularity is set to "daily granularity," the matrix will record the daily billing information for different billing items charged by each department.

[0111] Specifically, during the revenue data generation process, a hierarchical clustering algorithm is invoked to analyze the association matrix. This algorithm calculates the similarity between departments in terms of their billing items, gradually merging similar departments to form a cluster tree. For example, by calculating cosine similarity, the system can identify which departments have similar billing patterns. During clustering, a preset similarity threshold (e.g., 0.8) is used to determine whether to merge two departments. When the similarity exceeds the threshold, the system merges the two departments into one cluster and calculates the total billing amount for that cluster. In this way, the system can group departments with similar billing patterns together, simplifying the display of revenue data. Ultimately, the revenue data generated by the hierarchical clustering algorithm is output in a structured format, including the total billing amount for each cluster, a list of departments within the cluster, a list of billing items, and timestamps. This data can not only be directly used for report generation but also serve as input for other analytical modules (such as cost analysis and profit analysis), enabling multi-dimensional data analysis.

[0112] Step S203: Input the report configuration information into the inventory consumption network diagram of the cross-dimensional association engine for calculation to obtain inventory consumption data.

[0113] It's important to note that the inventory consumption network graph is a core component of the cross-dimensional association engine, specifically designed for processing inventory-related data. It constructs a dynamic graph structure, using medicines as nodes and departmental consumption records as edges. The edge weights represent the department's consumption of that medicine within a specified time granularity. For example, if the time granularity is set to "daily granularity," the network graph will record the daily consumption of different medicines by each department.

[0114] Furthermore, the process of generating inventory consumption data considers various factors, such as the expiration date of medications, inventory quantity, and departmental consumption habits. For example, for medications nearing their expiration date, the system prioritizes their consumption to avoid waste due to expiration. By introducing these dynamic factors, the system can more accurately predict and reflect actual inventory consumption. In addition, the data is smoothed to reduce the impact of short-term fluctuations, thus more clearly displaying long-term trends. For example, techniques such as moving averages or exponential smoothing can generate smoother inventory consumption curves, helping users better understand the overall trend of the data. Finally, the inventory consumption data generated by the inventory consumption network diagram will be output in a structured format, including daily total consumption, consumption by each department, and consumption trends. This data can not only be directly used for report generation but also serve as input for other analytical modules (such as registration volume analysis and revenue analysis) to achieve multi-dimensional data linkage analysis.

[0115] Step S204: Based on dynamic weights, the registration data, revenue data, and inventory consumption data are spatiotemporally aligned and normalized to obtain initial multimodal data.

[0116] It's important to note that spatiotemporal alignment is the process of aligning registration volume data, revenue data, and inventory consumption data according to the same time granularity. For example, if the time granularity is set to "daily granularity," the system will ensure that all data is summarized and displayed on a daily basis. This process is achieved through timestamp standardization and data resampling, ensuring consistency in time across different dimensions of data. For instance, some data may require interpolation or smoothing to fill in missing time points, thus ensuring data integrity. Normalization is the process of transforming data from different dimensions to the same scale. Registration volume, revenue, and inventory consumption data typically have different dimensions and orders of magnitude, and direct comparison can lead to data imbalance. Normalization transforms these data to the range [0, 1], allowing data from different dimensions to be compared on the same scale. For example, the max-min normalization method can be used to convert each data point into a relative value relative to its maximum and minimum values. This method not only eliminates the influence of dimensions but also makes the data more intuitive and readable.

[0117] Furthermore, after spatiotemporal alignment and normalization, the registration data, revenue data, and inventory consumption data are merged into initial multimodal data. This data not only contains information from the original data but also reflects the relative importance of different data dimensions at a specific time granularity through dynamic weighting. The initial multimodal data provides a unified data foundation for subsequent chart generation and comprehensive analysis, enabling reports to comprehensively and accurately display the hospital's operational status. Through spatiotemporal alignment and normalization based on dynamic weights, accurate and unified initial multimodal data can be generated, providing strong support for medical management decisions. This process not only improves the accuracy and consistency of the data but also enhances the flexibility and adaptability of the reporting system, allowing reports to better meet the needs of different users.

[0118] Step S205: Add metadata tags to the initial multimodal data to obtain multimodal data.

[0119] It should be noted that the metadata tags include data type, generation time, and weight allocation information. After data fusion is completed, three layers of metadata tags are added to the initial multimodal data: the "Data Type" field distinguishes between three categories—registration volume, revenue, and inventory consumption—using enumerated values ​​to facilitate subsequent routing to the corresponding visualization components; the "Generation Time" is accurate to the millisecond level, recording the moment the calculation of this data entry is completed, along with time zone information to ensure cross-hospital synchronization; and the "Weight Allocation Information" is written in JSON format, containing a dynamic weight array for registration, revenue, and inventory at this time granularity, an adjustment reason code (such as "Emergency Peak" or "Holidays"), and a summary of the parameters used in the weight calculation. All tags are Base64 encoded and appended to the data packet header. This tagging mechanism not only improves data organization efficiency but also provides a unified data semantic foundation for subsequent risk labeling, user interaction adjustments, and report export.

[0120] This embodiment generates registration volume, revenue, and inventory consumption data by inputting report configuration information into different sub-modules of the cross-dimensional association engine. Based on dynamic weights, it performs spatiotemporal alignment and normalization processing, and finally merges them into multimodal data with metadata tags, realizing multidimensional fusion and semantic enhancement of medical data.

[0121] Based on the first embodiment of this application, in the third embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 3 The dynamic report generation step S70 based on medical data further includes steps S301 to S306:

[0122] Step S301: Parse the user interaction command and extract the adjustment type and adjustment parameters.

[0123] It's important to note that user interaction commands are parsed to extract the adjustment type and adjustment parameters. Adjustment types may include changes to chart styles (e.g., switching from a bar chart to a line chart), adjustments to the time range (e.g., switching from monthly to daily granularity), and additions or subtractions of data dimensions (e.g., adding or removing registration data for a specific department). Adjustment parameters specify the details of the adjustment, such as the new time range or newly added dimension fields.

[0124] Step S302: If the adjustment type is chart style adjustment, then replace the chart in the initial report according to the adjustment parameters to obtain the style-updated chart.

[0125] It should be noted that when the user interaction command includes an adjustment type of chart style, the charts in the initial report will be replaced or optimized according to the adjustment parameters specified by the user, generating updated charts that meet the user's preferences and display needs. This feature enhances the interactivity and visualization capabilities of the reporting system, allowing users with different roles (such as hospital administrators, finance personnel, and pharmacy staff) to flexibly customize the data presentation according to their own needs.

[0126] Specifically, the system first analyzes the user's adjustment parameters to identify the required chart type (e.g., changing a line chart to a bar chart or stacked area chart), color theme, axis settings, and data aggregation method. Then, the system calls the built-in visualization component library to re-render the corresponding chart objects based on the new configuration, ensuring compatibility with the original report layout and maintaining the overall information structure. For example, for a registration volume trend chart, the user might want to switch from a "daily granular line chart" to a "weekly granular bar chart" for a clearer comparison of overall weekly visit volume changes; while for a revenue distribution chart, the user might choose to replace the two-dimensional bar chart with a pie chart or donut chart to highlight the revenue share of each department.

[0127] Furthermore, it supports saving and reusing custom chart styles, allowing users to save frequently used chart styles as templates for quick use in subsequent reports. This flexible style adjustment mechanism not only enhances the user experience but also strengthens the adaptability and expressiveness of dynamic reports across multiple scenarios, thereby better serving the hospital's intelligent decision-making needs.

[0128] Step S303: If the adjustment type is time range adjustment, then the multimodal data in the initial report is re-filtered according to the adjustment parameters to obtain the re-filtered data.

[0129] It should be noted that when the adjustment type is time range adjustment, the system will re-filter the multimodal data in the initial report based on the adjustment parameters provided by the user. This means that whether it is hospital registration records, patient information, drug inventory, or financial data, they will all be filtered and reorganized according to the new time range requirements to generate a data view that conforms to the specific time period.

[0130] Specifically, the system first parses the user-specified time range parameters. This may involve selecting start and end dates, or querying specific years, quarters, months, or even more precisely, hourly segments. Next, the system iterates through relevant forms and fields in the original database or data warehouse, using efficient query algorithms to extract all records falling within the specified time interval. For datasets requiring complex calculations (such as average daily patient visits or monthly revenue growth rates), the system dynamically adjusts its statistical logic to ensure the results accurately reflect the actual situation under the new timeframe. For example, when analyzing a department's annual drug consumption, if the user decides to shorten the time range from the whole year to the third quarter, the system will automatically recalculate and display the actual consumption of various drugs during that period, helping managers accurately grasp seasonal drug usage trends. Furthermore, to assess patient satisfaction within a specific time period, the system can retrieve relevant questionnaire results and feedback information based on the updated time parameters, thereby generating more targeted reports.

[0131] In addition, considering the diversity of multimodal data, including various formats such as text, numbers, and images, the system also needs to ensure that different types of data can be properly processed and integrated during the time range adjustment process.

[0132] Step S304: If the adjustment type is dimension increase / decrease adjustment, then add or remove the specified dimension based on the multimodal data in the initial report according to the adjustment parameters to obtain dimension adjustment data.

[0133] It should be noted that when the adjustment type is a dimension addition / removal adjustment, the system dynamically adds or removes specified data dimensions from the multimodal data of the initial report based on the adjustment parameters in the user interaction command, generating dimension adjustment data. For example, if a user wants to add a "medical insurance payment ratio" dimension to the current report to analyze the cost structure, the corresponding field will be extracted from the multimodal data and integrated into the revenue chart; conversely, if a user considers the "registration source channel" dimension redundant, it can also be removed from the report via command to reduce information interference. The system supports dynamic switching of registration data dimensions, fee data dimensions, drug inventory dimensions, etc., and ensures that the adjusted data remains consistent in the timeline and visualization layout.

[0134] Step S305: Remap the style-updated charts, re-filtered data, or dimension-adjusted data to the report layout of the initial report to obtain the intermediate report.

[0135] It should be noted that this process aims to reintegrate the chart styles, time ranges, or data dimensions modified by users through interactive commands into a unified report framework, generating a structurally complete and visually consistent intermediate report, laying the foundation for the final output of high-quality visualization results.

[0136] Specifically, the system identifies the current update content based on the user's operation type: if it's a chart style adjustment, the original chart component is replaced while maintaining its relative position in the layout; if it's a time range adjustment, the corresponding dataset is refreshed and dynamically rendered in the original chart; if it's a dimension addition or reduction adjustment, the report structure is partially restructured, adding or hiding relevant dimension information. Throughout the mapping process, the system uses a responsive layout engine to automatically adapt to different chart sizes and data densities, ensuring the overall page aesthetics and logical clarity.

[0137] In addition, to enhance user experience, the system supports historical version retrospective and multi-view parallel display, allowing users to compare report effects under different configurations on the same interface. For example, it can simultaneously display registration trend charts calculated by "daily granularity" and "monthly granularity," or display revenue analysis results including and excluding a certain dimension side by side.

[0138] Step S306: Render and cache the intermediate report to obtain the target report.

[0139] It's worth noting that during the rendering phase, the system uses a front-end visualization engine or a server-side rendering framework to uniformly draw all chart objects, data labels, layout rules, and other elements in the intermediate reports. The system supports multiple output formats, such as HTML, PDF, PNG, or Excel, ensuring that reports maintain good display quality across different terminals and application scenarios. Simultaneously, the system automatically optimizes the rendering strategy based on parameters such as the user's device resolution and screen ratio to enhance the visualization experience.

[0140] Regarding the caching mechanism, the system employs a combination of in-memory caching and persistent storage to temporarily or permanently store generated target reports and their metadata (such as report configuration, time range, dimension selection, style settings, etc.). For frequently accessed report templates, the system can automatically cache their rendering results to speed up subsequent calls and reduce the consumption of resources from repeated calculations. In addition, the caching mechanism also supports version management and historical rollback, making it easy for users to view the report status at different points in time or restore the configuration before a mistake.

[0141] Through rendering and intelligent caching mechanisms, the system not only improves report generation efficiency and response speed, but also enhances user experience and system stability, enabling dynamic medical reports to truly possess real-time, interactive, and reusable capabilities, meeting the data analysis and decision support needs of hospitals across multiple roles and scenarios.

[0142] This embodiment parses user interaction commands, identifies adjustment types, and performs corresponding chart style replacement, time range filtering, or dimension addition / reduction operations to ultimately generate a complete and adaptable target report. It supports real-time adjustment and visualization of multi-dimensional data, enhancing the efficiency of hospital data analysis and decision support capabilities.

[0143] Based on the first embodiment of this application, this application also provides a dynamic report generation device based on medical data. Please refer to... Figure 4 The device includes:

[0144] The data acquisition module 10 is used to acquire user interaction commands and report configuration information. The report configuration information includes time granularity, data dimension set and warning level. Time granularity includes hourly granularity, daily granularity, weekly granularity and monthly granularity. Data dimensions include registration data dimension, billing data dimension and drug inventory dimension.

[0145] The allocation module 20 is used to allocate weights to the registration data dimension, cost data dimension and drug inventory dimension in the data dimension according to the time granularity based on the dynamic weighted aggregation algorithm, so as to obtain the corresponding dynamic weights.

[0146] The data generation module 30 is used to input the report configuration information and dynamic weights into the cross-dimensional association engine to generate multimodal data, which includes registration data, revenue data and inventory consumption data. The cross-dimensional association engine is built based on the registration data dimension, the fee data dimension and the drug inventory dimension.

[0147] The judgment module 40 is used to judge the multimodal data according to the warning level and obtain the judgment result.

[0148] The execution module 50 is used to generate risk labeling data when the judgment result is that the inventory consumption data in the multimodal data exceeds the preset threshold of the corresponding warning level.

[0149] Module 60 is used to generate initial reports based on risk-labeled data and multimodal data.

[0150] The results module 70 is used to adjust the initial report according to user interaction instructions to obtain the target report.

[0151] The dynamic report generation device based on medical data provided in this application, employing the dynamic report generation method based on medical data in the above embodiments, can solve the technical problem of how to dynamically generate multi-dimensional linked reports and simultaneously complete inventory early warning. Compared with the prior art, the beneficial effects of the dynamic report generation device based on medical data provided in this application are the same as those of the dynamic report generation method based on medical data provided in the above embodiments, and other technical features in the dynamic report generation device based on medical data are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0152] In one embodiment, the allocation module 20 is further configured to: obtain departmental business priorities based on registration data; obtain chargeable item types based on cost data; obtain drug expiration date data based on drug inventory; determine initial first weight coefficients, initial second weight coefficients, and initial third weight coefficients based on time granularity according to data dimensions, wherein the registration data dimension corresponds to the initial first weight coefficient, the cost data dimension corresponds to the initial second weight coefficient, and the drug inventory dimension corresponds to the initial third weight coefficient; dynamically adjust the initial first weight coefficient based on departmental business priorities to obtain a first weight coefficient; grade and correct the initial second weight coefficient according to chargeable item types to obtain a second weight coefficient; compensate the initial third weight coefficient for timeliness based on drug expiration date data to obtain a third weight coefficient; and use the first weight coefficient, second weight coefficient, and third weight coefficient as the dynamic weights.

[0153] In one embodiment, the data generation module 30 is further configured to construct a time series analysis model based on the registration data dimension; construct a department-fee item association matrix based on the fee data dimension; combine the drug inventory dimension and the registration data dimension to calculate the department consumption correlation degree of each drug based on a graph neural network, and construct an inventory consumption network graph, wherein the network nodes of the inventory consumption graph represent the drug inventory status, and the edge weights represent the consumption rate; and combine the time series analysis model, the department-fee item association matrix, and the inventory consumption network graph to obtain a cross-dimensional association engine.

[0154] In one embodiment, the data generation module 30 is further configured to input report configuration information into the time series analysis model of the cross-dimensional association engine to generate registration volume data; input the report configuration information into the department-fee item association matrix of the cross-dimensional association engine and calculate it through a hierarchical clustering algorithm to obtain revenue data; input the report configuration information into the inventory consumption network diagram of the cross-dimensional association engine for calculation to obtain inventory consumption data; perform spatiotemporal alignment and normalization processing on the registration volume data, revenue data and inventory consumption data based on dynamic weights to obtain initial multimodal data; add metadata tags to the initial multimodal data to obtain multimodal data, wherein the metadata tags include data type, generation time and weight allocation information.

[0155] In one embodiment, the execution module 50 is further configured to obtain a level identifier and a corresponding threshold table based on the warning level, wherein the threshold table includes a first threshold, a second threshold, and a third threshold, the first threshold being less than the second threshold, and the second threshold being less than the third threshold; obtain the current consumption ratio based on inventory consumption data in multimodal data; determine the current warning level based on the current consumption ratio; retrieve a target threshold from the threshold table based on the current warning level to obtain a matching threshold; record an over-limit flag when the current consumption ratio is greater than the matching threshold; generate an event number based on the over-limit flag, and combine the event number, the current consumption ratio, the matching threshold, and the level identifier into risk labeling data.

[0156] In one embodiment, the construction module 60 is further configured to sort the multimodal data by time granularity to obtain a time-series data sequence; align the risk-marked data with the time-series data sequence to obtain a marked data set; generate a first chart based on the data set, which is used to display the trend of registration volume data over time; generate a second chart based on the data set, which is used to display the cumulative distribution of revenue data; generate a third chart based on the data set, which is used to display inventory consumption data and highlight the time period with risk-marked data; and stitch the first chart, the second chart, and the third chart together according to a preset layout rule to obtain an initial report.

[0157] In one embodiment, the result module 70 is further configured to parse user interaction commands, extract adjustment types and adjustment parameters, whereby adjustment types include chart style adjustment, time range adjustment, and dimension increase / decrease adjustment; if the adjustment type is chart style adjustment, the chart in the initial report is replaced according to the adjustment parameters to obtain a style-updated chart; if the adjustment type is time range adjustment, the multimodal data in the initial report is re-filtered according to the adjustment parameters to obtain re-filtered data; if the adjustment type is dimension increase / decrease adjustment, a specified dimension is added or removed from the multimodal data in the initial report according to the adjustment parameters to obtain dimension-adjusted data; the style-updated chart, re-filtered data, or dimension-adjusted data is remapped to the report layout of the initial report to obtain an intermediate report; the intermediate report is rendered and cached to obtain the target report.

[0158] This application provides a dynamic report generation device based on medical data. The dynamic report generation device based on medical data includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the dynamic report generation method based on medical data in the first embodiment described above.

[0159] The following is for reference. Figure 5The diagram illustrates a structural schematic of a dynamic report generation device based on medical data suitable for implementing embodiments of this application. The dynamic report generation device based on medical data in these embodiments may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 5 The illustrated dynamic report generation device based on medical data is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0160] like Figure 5 As shown, the dynamic report generation device based on medical data may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 (ROM) or a program loaded from a storage device 1003 into a random access memory 1004 (RAM). The RAM 1004 also stores various programs and data required for the operation of the dynamic report generation device based on medical data. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the dynamic report generation device based on medical data to communicate wirelessly or wiredly with other devices to exchange data. Although various dynamic report generation devices based on medical data are shown in the figures, it should be understood that implementation or possession of all of them is not required. More or fewer may be implemented alternatively.

[0161] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable storage medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0162] The dynamic report generation device based on medical data provided in this application, employing the dynamic report generation method based on medical data in the above embodiments, can solve the technical problem of how to dynamically generate multi-dimensional linked reports and simultaneously complete inventory early warning. Compared with the prior art, the beneficial effects of the dynamic report generation device based on medical data provided in this application are the same as those of the dynamic report generation method based on medical data provided in the above embodiments, and other technical features in this dynamic report generation device based on medical data are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0163] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0164] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0165] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the dynamic report generation method based on medical data in the above embodiments.

[0166] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible storage medium containing or storing a program that can be executed by instructions, used by a device, or used in conjunction with it. The program code contained on the computer-readable storage medium may be transmitted using any suitable storage medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0167] The aforementioned computer-readable storage medium may be included in a dynamic report generation device based on medical data; or it may exist independently and not be assembled into a dynamic report generation device based on medical data.

[0168] The aforementioned computer-readable storage medium carries one or more programs that, when executed by a dynamic report generation device based on medical data, enable the device to write computer program code for performing the operations of this application in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0169] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using dedicated hardware-based implementations that perform the specified functions or operations, or can be implemented using a combination of dedicated hardware and computer instructions.

[0170] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0171] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described dynamic report generation method based on medical data. This solves the technical problem of how to dynamically generate multi-dimensional linked reports and simultaneously complete inventory early warning. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the dynamic report generation method based on medical data provided in the above embodiments, and will not be repeated here.

[0172] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for generating dynamic reports based on medical data.

[0173] The computer program product provided in this application can solve the technical problem of how to dynamically generate multi-dimensional linked reports and simultaneously complete inventory early warning. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the dynamic report generation method based on medical data provided in the above embodiments, and will not be repeated here.

[0174] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for generating dynamic reports based on medical data, characterized in that, The method includes: Obtain user interaction commands and report configuration information, wherein the report configuration information includes time granularity, data dimension set and warning level, the time granularity includes hourly granularity, daily granularity, weekly granularity and monthly granularity, and the data dimensions include registration data dimension, billing data dimension and drug inventory dimension; Based on the dynamic weighted aggregation algorithm, weights are assigned to the registration data dimension, fee data dimension and drug inventory dimension in the data dimension according to the time granularity to obtain the corresponding dynamic weights; The report configuration information is combined with the dynamic weights and input into the cross-dimensional association engine to generate multimodal data, wherein the multimodal data includes registration data, revenue data and inventory consumption data, and the cross-dimensional association engine is built based on the registration data dimension, the fee data dimension and the drug inventory dimension. The multimodal data is judged based on the warning level to obtain the judgment result; When the judgment result indicates that the inventory consumption data in the multimodal data exceeds the preset threshold of the corresponding early warning level, risk marker data is generated; An initial report is generated based on the risk labeling data and the multimodal data; The initial report is adjusted according to the user interaction instructions to obtain the target report; The step of combining the report configuration information with the dynamic weights and inputting them into the cross-dimensional association engine to generate multimodal data, wherein the multimodal data includes registration data, revenue data, and inventory consumption data, and prior to the step of the cross-dimensional association engine being built based on the registration data dimension, the fee data dimension, and the drug inventory dimension, includes: A time series analysis model is constructed based on the aforementioned registration data dimensions; Construct a department-fee item association matrix based on the aforementioned fee data dimensions; Combining the drug inventory dimension and the registration data dimension, the departmental consumption correlation of each drug is calculated based on a graph neural network to construct an inventory consumption network graph, wherein the network nodes of the inventory consumption graph represent the drug inventory status and the edge weights represent the consumption rate. By combining the time series analysis model, the department-fee item association matrix, and the inventory consumption network diagram, a cross-dimensional association engine is obtained. The step of combining the report configuration information with the dynamic weight input into the cross-dimensional association engine to generate multimodal data includes: Input the report configuration information into the time series analysis model of the cross-dimensional correlation engine to generate registration volume data; The report configuration information is input into the department-fee item association matrix of the cross-dimensional association engine and the revenue data is obtained by hierarchical clustering algorithm. The report configuration information is input into the inventory consumption network diagram of the cross-dimensional association engine for calculation to obtain inventory consumption data; Based on the dynamic weights, the registration data, revenue data and inventory consumption data are spatiotemporally aligned and normalized to obtain initial multimodal data. Metadata tags are added to the initial multimodal data to obtain multimodal data. The metadata tags include data type, generation time, and weight allocation information.

2. The method as described in claim 1, characterized in that, The steps for assigning weights to the registration data dimension, fee data dimension, and drug inventory dimension in the data dimension according to the time granularity based on the dynamic weighted aggregation algorithm to obtain the corresponding dynamic weights include: Departmental business priorities are obtained based on the aforementioned registration data dimensions; The types of chargeable items are obtained based on the aforementioned cost data dimensions; Drug efficacy data is obtained based on the aforementioned drug inventory dimension; Based on the data dimensions and the time granularity, an initial first weight coefficient, an initial second weight coefficient, and an initial third weight coefficient are determined, wherein the registration data dimension corresponds to the initial first weight coefficient, the cost data dimension corresponds to the initial second weight coefficient, and the drug inventory dimension corresponds to the initial third weight coefficient. The initial first weight coefficient is dynamically adjusted based on the department's business priority to obtain the first weight coefficient. The initial second weight coefficient is adjusted in stages according to the type of the chargeable item to obtain the second weight coefficient. The initial third weighting coefficient is compensated for by combining the drug expiration date data to obtain the third weighting coefficient. The first weight coefficient, the second weight coefficient, and the third weight coefficient are used as the dynamic weights.

3. The method as described in claim 1, characterized in that, The step of generating risk marker data when the judgment result indicates that the inventory consumption data in the multimodal data exceeds the preset threshold of the corresponding early warning level includes: Based on the warning level, a level identifier and a corresponding threshold table are obtained, wherein the threshold table includes a first threshold, a second threshold and a third threshold, the first threshold being less than the second threshold and the second threshold being less than the third threshold; The current consumption ratio is obtained based on the inventory consumption data in the multimodal data; The current warning level is determined based on the current consumption ratio. Based on the current warning level, the target threshold is retrieved from the threshold table to obtain the matching threshold; When the current consumption ratio is greater than the matching threshold, an over-limit flag is recorded; An event number is generated based on the over-limit flag, and the event number, the current consumption ratio, the matching threshold, and the level identifier are combined into risk labeling data.

4. The method as described in claim 1, characterized in that, The step of generating an initial report based on the risk labeling data and the multimodal data includes: The multimodal data is sorted according to time granularity to obtain a time-series data sequence; The risk-labeled data is time-aligned with the time-series data sequence to obtain a labeled data set; A first chart is generated based on the data set, and the first chart is used to show the trend of the number of registrations over time; A second chart is generated based on the data set, and the second chart is used to display the cumulative distribution of revenue data; A third chart is generated based on the data set. The third chart is used to display inventory consumption data and highlight the time period with the risk-marked data. The first chart, the second chart, and the third chart are combined according to a preset layout rule to obtain an initial report.

5. The method as described in claim 1, characterized in that, The step of adjusting the initial report according to the user interaction instructions to obtain the target report includes: The user interaction commands are parsed to extract the adjustment type and adjustment parameters. The adjustment type includes chart style adjustment, time range adjustment, and dimension increase / decrease adjustment. If the adjustment type is a chart style adjustment, then the chart in the initial report is replaced according to the adjustment parameters to obtain a style-updated chart; If the adjustment type is a time range adjustment, then the multimodal data in the initial report is refiltered according to the adjustment parameters to obtain refiltered data; If the adjustment type is a dimension increase / decrease adjustment, then the specified dimension is added or removed from the multimodal data in the initial report according to the adjustment parameters to obtain the dimension adjustment data; The style-updated chart, the re-filtered data, or the dimension-adjusted data are remapped to the report layout of the initial report to obtain an intermediate report; The intermediate report is rendered and cached to obtain the target report.

6. A dynamic report generation device based on medical data, characterized in that, The device includes: The data acquisition module is used to acquire user interaction commands and report configuration information. The report configuration information includes time granularity, data dimension set and warning level. The time granularity includes hourly granularity, daily granularity, weekly granularity and monthly granularity. The data dimensions include registration data dimension, billing data dimension and drug inventory dimension. The allocation module is used to allocate weights to the registration data dimension, fee data dimension and drug inventory dimension in the data dimension according to the time granularity based on the dynamic weighted aggregation algorithm, so as to obtain the corresponding dynamic weights. The data generation module is used to input the report configuration information and the dynamic weights into a cross-dimensional association engine to generate multimodal data, including registration volume data, revenue data, and inventory consumption data. The cross-dimensional association engine is built based on the registration data dimension, the fee data dimension, and the drug inventory dimension. It is also used to construct a time series analysis model based on the registration data dimension; construct a department-fee item association matrix based on the fee data dimension; and calculate the department consumption association degree of each drug based on a graph neural network, combining the drug inventory dimension and the registration data dimension, to construct an inventory consumption network graph, where the network nodes of the inventory consumption graph represent the drug inventory status, and the edge weights represent the consumption rate. The module integrates the time series analysis model and the department-fee item association matrix... The system combines the array with the inventory consumption network diagram to obtain a cross-dimensional association engine. It also inputs the report configuration information into the time series analysis model of the cross-dimensional association engine to generate registration volume data; inputs the report configuration information into the department-fee item association matrix of the cross-dimensional association engine and calculates revenue data using a hierarchical clustering algorithm; inputs the report configuration information into the inventory consumption network diagram of the cross-dimensional association engine for calculation to obtain inventory consumption data; performs spatiotemporal alignment and normalization processing on the registration volume data, revenue data, and inventory consumption data based on the dynamic weights to obtain initial multimodal data; adds metadata tags to the initial multimodal data to obtain multimodal data, where the metadata tags include data type, generation time, and weight allocation information. The judgment module is used to judge the multimodal data according to the warning level and obtain the judgment result; The execution module is used to generate risk labeling data when the judgment result is that the inventory consumption data in the multimodal data exceeds the preset threshold of the corresponding early warning level; A construction module is used to generate an initial report based on the risk labeling data and the multimodal data; The results module is used to adjust the initial report according to the user interaction instructions to obtain the target report.

7. A dynamic report generation device based on medical data, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the dynamic report generation method based on medical data as described in any one of claims 1 to 5.

8. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the dynamic report generation method based on medical data as described in any one of claims 1 to 5.

Images