Medical data monitoring and early warning method, device, equipment and medium

Abstract

The application discloses a medical data monitoring and early warning method and device, equipment and medium, relates to the technical field of medical information management, and comprises the following steps: collecting medical data composed of multiple source data of outpatient volume, number of patients to be treated and material inventory; comprehensively scoring the medical data by time sequence feature extraction and adaptive weight matrix calculation; judging abnormalities in real time and triggering early warning; improving early warning accuracy and response efficiency; assisting hospitals in timely intervention of resource scheduling; and guaranteeing medical service continuity and patient safety.

Description

Technical Field

[0001] The present invention relates to the field of medical information management technology, and in particular to a medical data monitoring and early warning method, device, equipment and medium. Background Art

[0002] The frequent occurrence of public health emergencies places higher demands on hospitals' emergency response capabilities and resource allocation efficiency. Traditional medical management methods, which rely primarily on manual statistics and empirical judgment, struggle to cope with complex data changes and dynamic resource allocation needs. Furthermore, medical resource allocation involves multiple factors (such as personnel, equipment, and drug inventory). Existing systems often process single-dimensional data in isolation and lack comprehensive evaluation mechanisms, resulting in incomplete and inaccurate early warning information. Therefore, a medical data monitoring and early warning method is urgently needed to improve the efficiency and accuracy of early warnings for key medical indicators. Summary of the Invention

[0003] The main purpose of this application is to provide a medical data monitoring and early warning method, device, equipment and medium, aiming to solve the technical problem of how to improve the accuracy of early warning of key medical indicators.

[0004] To achieve the above objectives, this application proposes a medical data monitoring and early warning method, comprising:

[0005] Obtaining medical data, including outpatient volume, number of patients waiting to be seen, and medical supplies inventory data;

[0006] Extracting time series features from the medical data using a preset neural network model to obtain medical data features, wherein the preset neural network model is a long short-term memory network model;

[0007] Calculating the medical data features in combination with a weight coefficient matrix to obtain a comprehensive score value, wherein the weight coefficient matrix is ​​obtained based on the preset neural network model;

[0008] Comparing the comprehensive score with the preset abnormality threshold score to obtain a judgment result;

[0009] When the judgment result is that an abnormal situation exists, generating early warning information;

[0010] The warning information is sent to the target terminal so that the target terminal performs a corresponding alarm operation.

[0011] In one embodiment, the step of extracting time series features from the medical data using a preset neural network model to obtain medical data features, wherein the preset neural network model is a long short-term memory network model, includes:

[0012] Passing the medical data through a forget gate of the long short-term memory network model to generate a first intermediate state;

[0013] Passing the medical data through the input gate of the long short-term memory network model to generate a second intermediate state;

[0014] updating and combining the first intermediate state and the second intermediate state to obtain a time-dependent feature;

[0015] Passing the temporal dependency feature through the output gate of the long short-term memory network model to generate a hidden state feature;

[0016] The hidden state features are used as medical data features.

[0017] In one embodiment, before the step of calculating the medical data features in combination with the weight coefficient matrix to obtain a comprehensive score value, the step includes:

[0018] Acquire historical medical data and construct a training dataset;

[0019] Inputting the training data set into a preset neural network model to obtain a query vector and a key vector;

[0020] Calculating based on the query vector and the key vector to obtain a similarity score;

[0021] Performing softmax normalization processing on the similarity score to obtain a normalized result;

[0022] The normalized results are used as coefficient values ​​of the weight coefficient matrix to construct the weight coefficient matrix.

[0023] In one embodiment, the step of calculating the medical data features in combination with a weight coefficient matrix to obtain a comprehensive score value includes:

[0024] Converting the medical data features to obtain medical feature data;

[0025] Performing outlier filtering on the medical characteristic data to obtain filtered characteristic data;

[0026] Perform weighted fusion based on the filtered feature data in combination with the weight coefficient matrix to obtain fused feature data;

[0027] The fused feature data is transformed through dynamic normalization to obtain a comprehensive score value.

[0028] In one embodiment, the step of comparing the comprehensive score with a preset abnormality threshold score to obtain a judgment result includes:

[0029] According to the comprehensive score, an outpatient volume score, a waiting list score and an inventory score are obtained;

[0030] Obtaining a preset outpatient volume threshold score, a preset number of patients to be seen threshold score, and a preset inventory threshold score based on a preset abnormality threshold score, wherein the preset abnormality threshold score is dynamically adjusted based on historical medical data and environmental data;

[0031] Comparing the outpatient volume score with the preset outpatient volume threshold score to obtain a first comparison result;

[0032] Comparing the score of the number of patients to be seen with the preset outpatient volume threshold score to obtain a second comparison result;

[0033] Comparing the inventory score with the preset inventory threshold score to obtain a third comparison result;

[0034] When any one of the following conditions is met: the outpatient volume score is lower than the preset outpatient volume threshold score according to the first comparison result, the number of patients waiting to be seen score is lower than the preset number of patients waiting to be seen threshold score according to the second comparison result, and the inventory score is lower than the preset inventory threshold score according to the third comparison result, a judgment result is obtained that an abnormality exists;

[0035] When the first comparison result is that the outpatient volume score is lower than the preset outpatient volume threshold score, the second comparison result is that the number of patients to be seen score is lower than the preset number of patients to be seen threshold score, and the third comparison result is that the inventory score is lower than the preset inventory threshold score, none of them are met, the judgment result is normal.

[0036] In one embodiment, when the judgment result indicates that an abnormal situation exists, the step of generating warning information includes:

[0037] The abnormal situation is judged to obtain the abnormal type, which includes outpatient volume exceeding the standard, number of patients to be seen exceeding the standard, and insufficient inventory;

[0038] Generate a corresponding warning level according to the abnormality type;

[0039] generating warning information according to the warning level;

[0040] The step of generating a corresponding warning level according to the abnormality type includes:

[0041] Match the abnormality type to obtain the corresponding warning level rules;

[0042] Generate a first warning level, a second warning level, and a third warning level according to the warning level rule;

[0043] The first warning level, the second warning level and the third warning level are used as warning levels.

[0044] In one embodiment, before the step of sending the warning information to the target terminal so that the target terminal performs a corresponding alarm operation, the step includes:

[0045] Obtain a table of mapping relationships between warning levels and terminals and warning levels. The terminals include an operations command terminal, a physician mobile terminal, and a drug warehouse management terminal. The operations command terminal corresponds to an abnormal situation where the outpatient volume exceeds the standard, the physician mobile terminal corresponds to an abnormal situation where the number of patients waiting to be seen exceeds the standard, and the drug warehouse management terminal corresponds to an abnormal situation where the inventory is insufficient.

[0046] Based on the warning level and terminal mapping relationship table, query according to the warning level to obtain the target terminal identifier;

[0047] determining a target terminal address according to the target terminal identifier;

[0048] Taking the terminal corresponding to the target terminal address as the target terminal;

[0049] The step of sending the warning information to the target terminal so that the target terminal performs a corresponding alarm operation includes:

[0050] Obtaining the degree of urgency and priority level according to the warning information;

[0051] Sending a flashing light signal to the target terminal according to the warning information, so that the target terminal displays a flashing warning message on the screen when receiving the flashing light signal;

[0052] sending a sound signal to the target terminal according to the warning information, so that the target terminal performs a sound alarm when receiving the sound signal, and the volume of the sound signal is dynamically adjusted according to the urgency;

[0053] A vibration signal is sent to the target terminal according to the early warning information, so that the target terminal performs a vibration alarm when receiving the vibration signal, and the frequency and intensity of the vibration signal are set according to the priority processing level.

[0054] In addition, to achieve the above-mentioned purpose, the present application also proposes a medical data monitoring and early warning device, which includes:

[0055] An acquisition module is used to acquire medical data, including outpatient volume, number of patients waiting to be seen, and medical supplies inventory data;

[0056] a feature extraction module, configured to extract time series features from the medical data using a preset neural network model to obtain medical data features, wherein the preset neural network model is a long short-term memory network model;

[0057] a calculation module, configured to calculate the medical data features in combination with a weight coefficient matrix to obtain a comprehensive score value, wherein the weight coefficient matrix is ​​obtained based on the preset neural network model;

[0058] A judgment module is used to compare the comprehensive score with a preset abnormality threshold score to obtain a judgment result;

[0059] A result module is used to generate warning information when the judgment result shows that an abnormal situation exists;

[0060] The early warning module is used to send the early warning information to the target terminal so that the target terminal performs a corresponding alarm operation.

[0061] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the medical data monitoring and early warning method as described above are implemented.

[0062] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the medical data monitoring and early warning method as described above.

[0063] This application collects medical data consisting of multi-source data such as outpatient volume, number of patients waiting to be seen, and material inventory, and calculates a comprehensive score for the medical data through time series feature extraction and adaptive weight matrix calculation. It judges anomalies in real time and triggers early warnings, thereby improving early warning accuracy and response efficiency, assisting hospitals in timely intervention in resource scheduling, and ensuring the continuity of medical services and patient safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0065] Figure 1 This is a flowchart of the first embodiment of the medical data monitoring and early warning method of this application;

[0066] Figure 2 This is a system architecture diagram of the first embodiment of the medical data monitoring and early warning method of this application;

[0067] Figure 3 This is a flow chart of the second embodiment of the medical data monitoring and early warning method of this application;

[0068] Figure 4 This is a flowchart of the third embodiment of the medical data monitoring and early warning method of this application;

[0069] Figure 5 This is a schematic diagram of the module structure of the medical data monitoring and early warning device of the first embodiment of the medical data monitoring and early warning method of this application;

[0070] Figure 6 This is a schematic diagram of the device structure of the hardware operating environment involved in the medical data monitoring and early warning method in the embodiment of this application.

[0071] 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 DESCRIPTION

[0072] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0073] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0074] Traditional medical management methods rely primarily on manual statistics and empirical judgment, making them incapable of responding to dynamic changes in data such as outpatient volume, waiting lists, and medical supply inventory. Currently, some hospitals use alarm mechanisms based on fixed thresholds or simple statistical analysis methods to monitor anomalies, but these methods suffer from delayed response times, high false alarm rates, and an inability to adapt to complex time series changes. Furthermore, existing systems often rely on single-dimensional data analysis and lack the comprehensive assessment and intelligent early warning capabilities for multi-source medical data, making them unable to meet the practical needs of modern medical institutions for efficient and precise management.

[0075] Therefore, this application proposes a medical data monitoring and early warning method to solve the above-mentioned problems. The main solution of the embodiment of this application is: obtaining medical data; extracting time series features of the medical data through a preset neural network model to obtain medical data features, where the preset neural network model is a long short-term memory network model; calculating the medical data features in combination with a weight coefficient matrix to obtain a comprehensive score value, where the weight coefficient matrix is ​​obtained based on the preset neural network model; comparing the comprehensive score value with a preset abnormality threshold score to obtain a judgment result; when the judgment result is that an abnormal situation exists, generating an early warning message; sending the early warning message to a target terminal so that the target terminal performs a corresponding alarm operation.

[0076] Based on the above, the present application embodiment also provides a medical data monitoring and early warning method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the medical data monitoring and early warning method of this application. In this embodiment, the method is applied to Figure 2 The system architecture diagram shown in the figure shows the main layers of the medical data early warning system. The first is the data acquisition layer, which is responsible for collecting raw data from medical devices and electronic medical record systems. Next is the data processing layer, which cleans and extracts features from the collected data for subsequent analysis. The third layer is the anomaly detection layer, which uses neural network models and rule engines to identify anomalies in the data. Finally, there is the early warning generation layer, which generates corresponding early warning information based on the detected anomalies. This information is then sent to the notification and response layer, which is responsible for transmitting the early warning information to the target terminal and triggering the corresponding alarm operation. The medical data monitoring and early warning method includes steps S10 to S60:

[0077] Step S10: Acquire medical data.

[0078] It should be noted that medical data includes outpatient volume, number of patients waiting to be seen, and medical supplies inventory data. Outpatient volume data reflects the load of the hospital's daily diagnosis and treatment work, usually including daily or hourly patient registrations, number of patients visiting, and their distribution departments; the number of patients waiting to be seen reflects the current backlog of patients waiting for treatment, and is an important indicator for measuring medical service efficiency and patient waiting experience; medical supplies inventory data covers the real-time inventory and consumption rate of key resources such as medicines, consumables, and protective supplies, which are directly related to the continuity and safety of clinical treatment. The above data usually comes from various business platforms such as hospital information systems, electronic medical record systems, and material management systems, and needs to be automatically collected and integrated through data interfaces or Internet of Things devices.

[0079] In addition, to ensure the accuracy and timeliness of subsequent analysis, the initially collected medical data must be cleaned, normalized, and time-series aligned to form data in a unified format so that it can be directly input into the preset neural network model. Specifically, data cleaning is first performed to remove outliers, duplicate records, and missing data, and interpolation or deletion strategies are used to ensure data integrity. Secondly, normalization is implemented for data with different dimensions and value ranges, and indicators such as outpatient volume, number of patients waiting to be seen, and material inventory are uniformly mapped to the same numerical range to improve the stability and convergence speed of model training. In addition, since the frequency and timestamps of multi-source data collection may be inconsistent, time series alignment is also required to synchronize the data in the time dimension through resampling or interpolation methods. Finally, the processed data is organized into a structured time window sequence according to the model input format requirements to ensure that it can be directly input into the preset neural network model, laying the foundation for efficient and accurate time series feature extraction and anomaly warning.

[0080] Step S20: extracting time series features from the outpatient volume, the number of patients waiting to be seen, and the medical supplies inventory data using a preset neural network model to obtain medical data features.

[0081] It should be noted that in this embodiment, the default neural network model is a long short-term memory (LSTM) model because it excels at capturing long-term dependencies and other complex patterns in time series data. Specifically, this LSTM model employs a three-layer LSTM architecture, with each layer containing 50 neurons, providing sufficient capacity to learn complex energy consumption patterns. The Adam optimizer was used during training, with a learning rate of 0.005, a batch size of 32, and mean squared error (MSE) as the loss function. Hyperparameters were adjusted using grid search or random search to ensure optimal model performance.

[0082] Furthermore, step S20 also includes: passing the medical data through the forget gate of the long short-term memory network model to generate a first intermediate state, specifically the formula is:

[0083]

[0084] in, is the Sigmoid activation function, and are the weight matrix and bias term of the forget gate respectively, is the input data state, is the state of the previous time step. The medical data is passed through the input gate of the long short-term memory network model to generate the second intermediate state. The specific formula is:

[0085]

[0086] in, represents the activation state of the input gate, and are the weight matrix and bias term of the input gate respectively, Indicates candidate status, and are the weight matrix and bias term of the candidate state respectively, and ⊙ represents element-wise multiplication.

[0087] The first intermediate state and the second intermediate state are updated and combined to obtain the timing dependency feature, which is specifically expressed as follows:

[0088]

[0089] in, Represents the temporal dependency feature of the previous step. The temporal dependency feature is passed through the output gate of the long short-term memory network model to generate the hidden state feature. The specific formula is:

[0090]

[0091] in, Indicates the activation state of the output gate, and The weight matrix and bias term of the output gate are used as the hidden state features as medical data features. Through the above structure, LSTM can effectively capture the long-term dependencies of medical data such as outpatient volume, number of patients waiting for consultation, and material inventory, thereby improving the accuracy and timeliness of anomaly warnings.

[0092] Step S30: Calculate the medical data features in combination with the weight coefficient matrix to obtain a comprehensive score value.

[0093] It's important to note that this process aims to weight and fuse multi-dimensional, heterogeneous medical data features (such as trends in outpatient volume, fluctuations in the number of patients waiting to be seen, and inventory levels of key supplies) to form quantitative indicators of actual content. The weight coefficient matrix is ​​automatically learned through model training on historical medical data samples and dynamically reflects the importance of different features in different scenarios.

[0094] Furthermore, before step S30, the method further includes: obtaining historical medical data and constructing a training data set; inputting the training data set into a preset neural network model to obtain a query vector and a key vector; performing calculations based on the query vector and the key vector to obtain a similarity score; performing softmax normalization on the similarity score to obtain a normalized result; and using the normalized result as the coefficient value of the weight coefficient matrix to construct the weight coefficient matrix. Specifically, historical medical data is first obtained, including indicators such as outpatient volume, number of patients waiting to be seen, and medical supply inventory, and organized according to time series to construct a structured training data set. This data set is used to simulate medical scenarios under different operating conditions and provide a basis for subsequent model learning. The training data set is input into the long short-term memory network model, and the model will generate a query vector and a key vector corresponding to each time step respectively. By calculating the dot product between the query vector and the key vector, the similarity score between each feature dimension is obtained, reflecting its relevance and importance in the current context. Subsequently, the similarity score is normalized by applying the Softmax function to convert it into a set of probability distribution values ​​with a sum of 1, specifically expressed as:

[0095]

[0096] in, Represents the total number of elements, Represents the query vector Middle elements, Represents the key vector Middle elements, is the activation function, represents the sum of all possible key vectors after exponential operation. Ultimately, this normalized result is used as the element value in the weight coefficient matrix to construct a dynamic and learnable weight matrix. This matrix can automatically adjust the contribution weight of each feature based on the actual data distribution, making the comprehensive score more business-interpretable and predictive, thereby improving the system's early warning performance in complex medical scenarios.

[0097] Step S40: Compare the comprehensive score with the preset abnormality threshold score to obtain a judgment result.

[0098] It should be noted that by setting a reasonable threshold range, the comprehensive score values ​​are compared to obtain a positive or abnormal judgment result, thereby guiding the execution of subsequent warning actions.

[0099] Furthermore, step S40 further includes: obtaining an outpatient volume score, a patient waiting list score, and an inventory score based on the comprehensive score. Specifically, the comprehensive score is broken down into multiple key indicator sub-scores, including an outpatient volume score, a patient waiting list score, and a medical supply inventory score. These sub-scores reflect the hospital's operating pressure and risk level in different dimensions. Then, based on the preset abnormality threshold scores, a preset outpatient volume threshold score, a preset patient waiting list threshold score, and a preset inventory threshold score are obtained. Specifically, these preset abnormality threshold scores are not fixed and are dynamically adjusted based on historical medical data (such as seasonal peak visit data) and environmental data (such as weather, holidays, and emergencies) to ensure that the thresholds are time-varying and scenario-adaptive. For example, during peak influenza season, the outpatient volume threshold can be automatically raised to avoid false alarms; while during periods of material shortages, the inventory score threshold can be lowered accordingly to increase warning sensitivity.

[0100] Furthermore, the outpatient volume score is compared with the preset outpatient volume threshold score to obtain a first comparison result; the waiting list score is compared with the preset outpatient volume threshold score to obtain a second comparison result; and the inventory score is compared with the preset inventory threshold score to obtain a third comparison result. Specifically, by comparing the outpatient volume score, the waiting list score, and the inventory score with their respective preset thresholds, a multi-dimensional assessment of the hospital's operating conditions can be achieved. The outpatient volume score reflects the patient load per unit time. If it is lower than the preset outpatient volume threshold score, it indicates that the current outpatient pressure is too high, and there may be emergencies or insufficient resource allocation. The waiting list score reflects the degree of match between doctors' reception capacity and patient waiting time. When it is lower than the preset threshold, it indicates a large backlog of patients and the need for strengthened human resource scheduling. The inventory score is used to measure the reserve level of key medical supplies (such as medicines and consumables). If it is lower than the preset inventory threshold score, it indicates that inventory is tight, which may affect subsequent diagnosis and treatment services. Therefore, when any one of the following conditions is met: the outpatient volume score is lower than the preset outpatient volume threshold score, the second comparison result is that the number of patients to be seen is lower than the preset number of patients to be seen threshold score, and the third comparison result is that the inventory score is lower than the preset inventory threshold score, the judgment result is that an abnormal situation exists, that is, as long as any indicator score is lower than the corresponding threshold, it is judged that an abnormal situation exists, the early warning mechanism is triggered, and processing suggestions are pushed. When the first comparison result is that the outpatient volume score is lower than the preset outpatient volume threshold score, the second comparison result is that the number of patients to be seen is lower than the preset number of patients to be seen threshold score, and the third comparison result is that the inventory score is lower than the preset inventory threshold score, the judgment result is normal. Only when all scores are not lower than their respective thresholds can it be judged as a normal state. This "or" logical judgment method ensures that the system can respond to various potential risks in a timely manner and improve the intelligence and foresight level of hospital management.

[0101] Step S50: When the judgment result shows that an abnormal situation exists, a warning message is generated.

[0102] It's important to note that when an abnormality is identified, the type of abnormality is further identified and its specific cause clarified, such as excessive outpatient volume, excessive waiting list, or insufficient inventory. Different types of abnormalities have varying degrees of impact on medical services, necessitating the generation of corresponding warning levels based on their urgency and scope. Warning information is then generated based on the warning level and notified to relevant management or business personnel through a multi-channel push mechanism, enabling them to take timely countermeasures. Warning information typically includes the abnormality type, time of occurrence, the indicators involved, the current score, the threshold reference value, and a recommended response strategy, ensuring a rapid response. For example, an outpatient volume score below the threshold may indicate a surge in patients in the short term, necessitating the temporary redeployment of medical staff; an abnormal waiting list score may indicate decreased physician efficiency or inappropriate scheduling; and an excessively low inventory score suggests prompt replenishment to avoid the risk of supply shortages.

[0103] Furthermore, the system can combine historical data with similar cases to intelligently recommend emergency response plans, such as resource scheduling suggestions, process optimization prompts, or emergency plan invocations, to assist managers in making informed decisions. Warning information can also be simultaneously pushed to the hospital management platform, mobile app, or SMS / email system, enabling multi-terminal linkage reminders.

[0104] Step S60: sending warning information to the target terminal so that the target terminal performs a corresponding alarm operation.

[0105] It should be noted that when sending early warning information to the terminal, it is necessary to confirm the location of the target terminal. Specifically, obtain the early warning level and terminal mapping relationship table and the early warning level. The above-mentioned terminals include the operation command terminal, the physician mobile terminal and the drug warehouse management terminal. The operation command terminal corresponds to the abnormal situation of outpatient volume exceeding the limit, the physician mobile terminal corresponds to the abnormal situation of the number of patients to be seen exceeding the limit, and the drug warehouse management terminal corresponds to the abnormal situation of insufficient inventory. The above-mentioned early warning level and terminal mapping relationship table corresponds different types of abnormal situations to the corresponding management terminals according to the hospital's organizational structure and division of responsibilities, ensuring that the early warning information can be accurately delivered to the most appropriate processing personnel. Specifically, the operation command terminal is mainly responsible for the scheduling and supervision of the overall outpatient operation, and therefore corresponds to the abnormal situation of outpatient volume exceeding the limit; the physician mobile terminal is aimed at the front-line doctor group, and is used to receive reminders of the number of patients to be seen exceeding the limit, so as to facilitate timely adjustment of the schedule or optimization of the diagnosis and treatment process; the drug warehouse management terminal focuses on monitoring the status of medical supplies. Once the inventory score is lower than the threshold, the drug warehouse management personnel will be notified immediately to replenish or allocate. Then, based on the warning level and terminal mapping relationship table, a query is performed according to the warning level to obtain the target terminal identifier. The target terminal address is then determined based on the target terminal identifier. Finally, the terminal corresponding to the target terminal address is used as the target terminal. Specifically, after identifying the anomaly and determining the warning level, the warning level and terminal mapping relationship table is used to query the target terminal identifier to be notified, such as the IP address, device ID, or communication port number. The communication address of the target terminal is then parsed based on this identifier to complete the information routing configuration. Ultimately, the warning information will be sent to the corresponding target terminal and trigger the corresponding alarm operation, such as a pop-up prompt, sound alarm, SMS notification, etc.

[0106] Furthermore, step S60 includes: obtaining the degree of urgency and the priority level according to the early warning information. Specifically, the alarm mode is intelligently adapted and graded for response control according to the degree of urgency and the priority level in the early warning information. Specifically, the urgency of the early warning is divided into three levels: low, medium and high, and the priority level is set according to the type of abnormality and its potential impact range. Based on these parameters, a flashing light signal, a sound signal and a vibration signal are generated and sent to the target terminal to achieve a multimodal alarm prompt. Then, a flashing light signal is sent to the target terminal according to the early warning information, so that the target terminal displays a flashing warning message on the screen when receiving the flashing light signal; a sound signal is sent to the target terminal according to the early warning information, so that the target terminal performs a sound alarm when receiving the sound signal, and the volume of the sound signal is dynamically adjusted according to the degree of urgency; a vibration signal is sent to the target terminal according to the early warning information, so that the target terminal performs a vibration alarm when receiving the vibration signal, and the frequency and intensity of the vibration signal are set according to the priority level. Specifically, the flashing light signal is used to trigger a flashing warning message on the target terminal screen, such as a red or yellow highlighted pop-up window, to alert the user to the current pending matters. The sound signal dynamically adjusts the alarm volume according to the degree of urgency. For example, general warnings only play a low-volume prompt tone, while high-level warnings play a high-volume, repetitive alarm sound to quickly attract attention. The vibration signal then sets the vibration frequency and intensity according to the priority level. Low-priority warnings only trigger slight, short-term vibrations, while high-priority warnings use a high-frequency, strong vibration mode to ensure that they can be perceived even in noisy environments. Through multi-channel collaborative alarm methods, the efficiency of warning information transmission and response speed are improved, effectively supporting hospitals in making rapid decisions and handling emergencies in complex environments.

[0107] This embodiment collects medical data consisting of multi-source data such as outpatient volume, number of patients waiting to be seen, and material inventory, and performs comprehensive scoring on the medical data through time series feature extraction and adaptive weight matrix calculation. It then identifies anomalies in real time and triggers early warnings, thereby improving early warning accuracy and response efficiency, assisting hospitals in timely intervention in resource scheduling, and ensuring the continuity of medical services and patient safety.

[0108] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above embodiment 1 can be referred to the above introduction and will not be described in detail later. Figure 3 The medical data monitoring and early warning method step S30 further includes steps S201 to S204:

[0109] Step S201: convert the medical data features to obtain medical feature data.

[0110] It should be noted that the medical data features extracted by LSTM were normalized and the influence of outliers was eliminated using an improved RobustScaler method, which uses median and quartile scaling. Feature decoupling techniques were then used to decompose the feature space into trend, periodic, and residual subspaces through orthogonal transformation. Finally, an adaptive feature fusion strategy was employed to obtain the final medical feature data.

[0111] Step S202: filtering the medical feature data for outliers to obtain filtered feature data.

[0112] It should be noted that a multidimensional anomaly detection mechanism based on the fusion of statistics and machine learning is used, combined with statistical methods such as Z-score and boxplots to initially identify data points that significantly deviate from the normal range. At the same time, unsupervised learning algorithms such as Isolation Forest and Local Outlier Factor (LOF) are introduced to further explore potential hidden abnormal patterns, especially in complex scenarios such as sudden increases in outpatient volume and abnormal fluctuations in drug consumption. Subsequently, threshold boundaries are dynamically set based on the distribution of historical data, and a sliding window mechanism is used to continuously monitor real-time feature data. For data points determined to be abnormal, interpolation, KNN filling, or GAN-based generative models are used for reasonable replacement rather than simple elimination to ensure data integrity and temporal consistency, resulting in filtered feature data.

[0113] Step S203 , performing weighted fusion based on the filtered feature data in combination with a weight coefficient matrix to obtain fused feature data.

[0114] It is important to note that the input is feature data that has been filtered for outliers and standardized, such as outpatient volume, number of patients waiting to be seen, and drug inventory data. These features typically have different dimensions and importance, so a weight coefficient matrix is ​​introduced to quantify the contribution of each feature to the task at hand. The weight matrix is ​​automatically learned and optimized through model learning methods. During the fusion process, each feature is multiplied by its corresponding weight and then summed or normalized to form fused feature data with comprehensive expressive power. To further enhance model adaptability, an attention mechanism can be introduced to achieve dynamic weight adjustment, allowing the model to automatically identify the most relevant key features in different time windows or business scenarios. The final fused feature data output not only retains the business meaning of the original data, but also enhances its discriminability and robustness through weighted fusion, providing high-quality data support for subsequent intelligent early warning, trend forecasting, and resource scheduling.

[0115] Step S204: convert the fused feature data through dynamic normalization to obtain a comprehensive score value.

[0116] It should be noted that this process aims to uniformly map heterogeneous feature data after weighted fusion onto a standardized numerical range (e.g., [0, 100]), thereby generating a comparable and interpretable comprehensive score for subsequent anomaly identification and decision support. The entire conversion process first partitions the fused feature data into time series windows. This involves extracting feature vectors within the current time period at fixed time intervals (e.g., hourly or daily), ensuring the timeliness and continuity of the scoring results. Subsequently, the system employs a dynamic normalization algorithm, adjusting the normalization parameters in real time based on the mean and standard deviation of historical data within a sliding window. This effectively addresses seasonal fluctuations and unexpected trends in medical data, improving the stability and adaptability of the score. Furthermore, nonlinear compression functions (such as Sigmoid or Tanh) can be introduced to further process the normalized values, making the score distribution smoother and more consistent with actual business logic.

[0117] The final output comprehensive score value can intuitively reflect the health of the hospital's current operating status, and provide accurate data support for subsequent threshold comparison, early warning generation and resource scheduling.

[0118] This embodiment constructs a comprehensive medical scoring system through feature conversion, outlier filtering, weighted fusion and dynamic normalization processing, improves data quality and model robustness, enhances the accuracy of anomaly identification, and improves medical management efficiency and scientific decision-making.

[0119] Based on the first embodiment of the present application, in the third embodiment of the present application, the same or similar contents as those in the first embodiment can be referred to the above introduction and will not be described in detail later. Figure 4 The medical data monitoring and early warning step S50 further includes steps S301 to S303:

[0120] Step S301: determine the abnormal situation and obtain the abnormal type.

[0121] It should be noted that by analyzing key indicators in medical operations, the above-mentioned abnormality types include three types: outpatient volume exceeded, number of patients waiting to be seen exceeded, and insufficient inventory, each corresponding to different business risks and response strategies. Excessive outpatient volume means that the number of patients visiting per unit time far exceeds the historical average or set threshold, which may indicate a public health emergency or a high incidence of seasonal diseases; excessive number of patients waiting to be seen reflects insufficient doctor's ability to see patients or unreasonable scheduling, which affects the efficiency of patients' medical treatment; insufficient inventory indicates that the reserves of drugs or consumables are below the safety line, and there is a risk of supply interruption. By identifying specific abnormality types, the system can match the corresponding emergency response mechanism to improve the intelligence and refinement of hospital management.

[0122] Step S302: Generate a corresponding warning level according to the abnormality type.

[0123] It should be noted that different types of anomalies have different degrees of impact on medical services. Therefore, it is necessary to set differentiated warning levels based on historical data, business rules and scope of impact.

[0124] Furthermore, step S302 also includes: matching according to the abnormality type to obtain the corresponding warning level rules; generating the first warning level, the second warning level and the third warning level according to the warning level rules; and using the first warning level, the second warning level and the third warning level as warning levels. Specifically, when the abnormality type is "outpatient volume exceeds the standard", if the current outpatient volume exceeds the historical average value for the same period by less than 30%, it can be determined as the first warning level, indicating that there is a public health emergency and the emergency plan needs to be activated immediately. However, there are different risk levels for the same warning level. The corresponding warning risk level will be matched based on factors such as the degree of deviation between the current comprehensive score and the threshold and the duration. For example, when the score exceeds the threshold by 10%-30%, it is a low risk, prompting attention to trend changes; when it exceeds 30%-50%, it is upgraded to a medium risk; when it exceeds 50% or the duration exceeds the set window period, it is defined as a high risk.

[0125] For "excessive waiting list", if the number of waiting people is too large and affects the efficiency of diagnosis and treatment, it will be set to the second warning level, and the specific risk level is set as when the number of waiting patients exceeds the daily average by 10%-30%, it is a low risk, but still within the controllable range, suggesting that attention should be paid to the changing trend of diagnosis and treatment efficiency. At this time, it is recommended to make appropriate adjustments within the department, such as optimizing the schedule or slightly increasing temporary support staff. If the number of waiting patients exceeds 30%-50% of the historical average, it means that the existing resources can no longer meet the current demand and begin to affect the efficiency of diagnosis and treatment, which is a medium-risk level. At this time, it is recommended to activate the department-level emergency plan, such as increasing the number of doctors on duty, extending working hours and other measures. Once the number of waiting patients exceeds the historical average by more than 50%, or the duration exceeds the preset time window, it means that medical services are facing serious challenges and are at high risk. Emergency response measures at the hospital level must be taken immediately, including but not limited to cross-departmental deployment of doctors, suspension of non-emergency surgeries, etc.

[0126] As for the "insufficient inventory" type of anomalies, if the reserves of key drugs or consumables are lower than the safety line, it is set to the third warning level. The specific risk level is set as follows: when the reserves of key materials are lower than the safety line but higher than the minimum limit, it is a low-risk state. The main purpose is to remind management personnel to make preparations for replenishment in advance to ensure that there will be no supply interruption. If the inventory level drops to near the minimum limit, but there are still a few days of buffer period for allocation, it is defined as medium risk. It is necessary to speed up the procurement process and consider temporary allocation plans to prevent service interruptions due to shortages. When the inventory level drops below the standard that cannot support normal operations, or has entered a stock-out state, it is a high risk. At this time, the highest level of emergency response mechanism should be triggered immediately, including emergency procurement, requesting assistance from other medical institutions, etc., to restore supply stability as soon as possible. Step S303 generates warning information according to the warning level.

[0127] Step S303: Generate warning information according to the warning level.

[0128] It should be noted that after identifying the anomaly type and determining its corresponding warning level, structured warning information will be automatically generated based on preset rules and business logic. This warning information typically includes several key elements: warning level, anomaly type, occurrence time, involved indicators, current score, threshold reference value, scope of impact, and recommended handling strategies. For example, when the outpatient volume score far exceeds the set threshold and triggers a red alert, the warning information will clearly indicate "Outpatient volume exceeds the standard," "June 11, 2025, 15:30," "Comprehensive score of 9.8 / 10," "Threshold reference value of 7.5," and other content. Combined with historical trends and real-time data, it will recommend response measures, such as "Immediately activate the emergency reception plan," "Add additional medical staff to support outpatient clinics," and "Direct patients to nearby medical institutions."

[0129] Furthermore, early warning information is formatted and prioritized based on the display needs of different terminals (such as operation control screens, physician mobile terminals, or drug storage management platforms) to ensure efficient and accurate information dissemination. Through this mechanism, hospitals can immediately identify operational anomalies and respond quickly based on scientific judgment, significantly improving their overall emergency management capabilities and medical service guarantee level.

[0130] This embodiment realizes intelligent monitoring and graded warning of medical operation status by identifying abnormal types and generating corresponding warning levels and warning information, improving abnormal response efficiency, optimizing resource allocation, enhancing the scientific and intelligent level of hospital management, and ensuring the safety and continuity of medical services.

[0131] Based on the first embodiment of the present application, the present application also provides a medical data monitoring and early warning device, please refer to Figure 5 , the device comprises:

[0132] An acquisition module 10 is used to acquire medical data, including outpatient volume, number of patients waiting to be seen, and medical supplies inventory data;

[0133] A feature extraction module 20 is used to extract time series features from medical data using a preset neural network model to obtain medical data features. The preset neural network model is a long short-term memory network model.

[0134] A calculation module 30 is used to calculate the medical data features in combination with a weight coefficient matrix to obtain a comprehensive score value, where the weight coefficient matrix is ​​obtained based on the preset neural network model;

[0135] The judgment module 40 is used to compare the comprehensive score with the preset abnormality threshold score to obtain a judgment result;

[0136] The result module 50 is used to generate warning information when the judgment result shows that an abnormal situation exists;

[0137] The warning module 60 is used to send warning information to the target terminal so that the target terminal performs a corresponding alarm operation.

[0138] The medical data monitoring and early warning device provided in this application, employing the medical data monitoring and early warning method described in the aforementioned embodiments, can address the technical problem of improving the accuracy of early warnings for key medical indicators. Compared to the prior art, the beneficial effects of the medical data monitoring and early warning device provided in this application are the same as those of the medical data monitoring and early warning method described in the aforementioned embodiments. Other technical features of the device are the same as those disclosed in the aforementioned embodiments and are not further elaborated here.

[0139] In one embodiment, the feature extraction module 20 is also used to pass the medical data through the forget gate of the long short-term memory network model to generate a first intermediate state; pass the medical data through the input gate of the long short-term memory network model to generate a second intermediate state; update and combine the first intermediate state and the second intermediate state to obtain a time-dependent feature; pass the time-dependent feature through the output gate of the long short-term memory network model to generate a hidden state feature; and use the hidden state feature as a medical data feature.

[0140] In one embodiment, the computing module 30 is further used to obtain historical medical data and construct a training data set; input the training data set into a preset neural network model to obtain a query vector and a key vector; perform calculations based on the query vector and the key vector to obtain a similarity score; perform softmax normalization processing on the similarity score to obtain a normalized result; and use the normalized result as a coefficient value of a weight coefficient matrix to construct a weight coefficient matrix.

[0141] In one embodiment, the calculation module 30 is also used to convert medical data features to obtain medical feature data; filter outliers on the medical feature data to obtain filtered feature data; perform weighted fusion based on the filtered feature data combined with the weight coefficient matrix to obtain fused feature data; and convert the fused feature data through dynamic normalization to obtain a comprehensive score value.

[0142] In one embodiment, the judgment module 40 is further used to obtain an outpatient volume score, a waiting number score and an inventory score based on the comprehensive score value; obtain a preset outpatient volume threshold score, a preset waiting number threshold score and a preset inventory threshold score based on the preset abnormal threshold score, and the preset abnormal threshold score is dynamically adjusted according to historical medical data and environmental data; compare the outpatient volume score with the preset outpatient volume threshold score to obtain a first comparison result; compare the waiting number score with the preset outpatient volume threshold score to obtain a second comparison result; compare the inventory score with the preset inventory threshold score to obtain The third comparison result; when the first comparison result is that the outpatient volume score is lower than the preset outpatient volume threshold score, the second comparison result is that the number of patients to be received score is lower than the preset number of patients to be received threshold score, and the third comparison result is that the inventory score is lower than the preset inventory threshold score, any one of the following is met, the judgment result is that there is an abnormal situation; when the first comparison result is that the outpatient volume score is lower than the preset outpatient volume threshold score, the second comparison result is that the number of patients to be received score is lower than the preset number of patients to be received threshold score, and the third comparison result is that the inventory score is lower than the preset inventory threshold score, none of them are met, the judgment result is that the situation is normal.

[0143] In one embodiment, the result module 50 is also used to judge the abnormal situation and obtain the abnormal type, which includes excessive outpatient volume, excessive number of patients to be seen, and insufficient inventory; generate a corresponding warning level according to the abnormal type; and generate warning information according to the warning level.

[0144] In one embodiment, the result module 50 is also used to match according to the abnormality type to obtain the corresponding warning level rules; generate the first warning level, the second warning level and the third warning level according to the warning level rules; and use the first warning level, the second warning level and the third warning level as warning levels.

[0145] In one embodiment, the early warning module 60 is also used to obtain a mapping relationship table between early warning levels and terminals and an early warning level. The terminals include an operation command terminal, a physician mobile terminal, and a drug warehouse management terminal. The operation command terminal corresponds to an abnormal situation where the outpatient volume exceeds the standard, the physician mobile terminal corresponds to an abnormal situation where the number of patients to be seen exceeds the standard, and the drug warehouse management terminal corresponds to an abnormal situation where the inventory is insufficient. Based on the mapping relationship table between early warning levels and terminals, a query is performed according to the early warning level to obtain a target terminal identifier; the target terminal address is determined according to the target terminal identifier; and the terminal corresponding to the target terminal address is used as the target terminal.

[0146] In one embodiment, the early warning module 60 is further used to obtain the degree of urgency and the priority level based on the early warning information; send a flashing light signal to the target terminal based on the early warning information, so that the target terminal displays a flashing warning message on the screen when receiving the flashing light signal; send a sound signal to the target terminal based on the early warning information, so that the target terminal performs a sound alarm when receiving the sound signal, and the volume of the sound signal is dynamically adjusted according to the degree of urgency; send a vibration signal to the target terminal based on the early warning information, so that the target terminal performs a vibration alarm when receiving the vibration signal, and the frequency and intensity of the vibration signal are set according to the priority level.

[0147] The present application provides a medical data monitoring and early warning device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the medical data monitoring and early warning method in the above-mentioned embodiment one.

[0148] Reference below Figure 6 , which shows a schematic diagram of the structure of a medical data monitoring and early warning device suitable for implementing the embodiments of the present application. The medical data monitoring and early warning device in the embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), in-vehicle terminals (such as in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 6 The medical data monitoring and early warning device shown is merely an example and should not impose any limitations on the functions and scope of use of the embodiments of the present application.

[0149] like Figure 6As shown, the medical data monitoring and early warning device may include a processing device 1001 (e.g., a central processing unit, graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in a read-only memory (ROM) 1002 or programs loaded from a storage device 1003 into a random access memory (RAM) 1004. RAM 1004 also stores various programs and data required for the operation of the medical data monitoring and early warning device. Processing device 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 may be connected to I / O interface 1006: input devices 1007, such as a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008, such as a liquid crystal display (LCD), speaker, vibrator, etc.; storage device 1003, such as a magnetic tape or hard disk; and communication device 1009. The communication device 1009 can allow the medical data monitoring and early warning device to communicate wirelessly or wired with other devices to exchange data. Although various medical data monitoring and early warning devices are shown in the figure, it should be understood that it is not required to implement or have all of the shown ones. More or fewer can be implemented or have alternatively.

[0150] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable storage medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0151] The medical data monitoring and early warning device provided in this application, employing the medical data monitoring and early warning method described in the aforementioned embodiment, can address the technical problem of improving the accuracy of early warnings for key medical indicators. Compared to the prior art, the beneficial effects of the medical data monitoring and early warning device provided in this application are the same as those of the medical data monitoring and early warning method described in the aforementioned embodiment. Other technical features of this medical data monitoring and early warning device are the same as those disclosed in the aforementioned embodiment and are not further elaborated here.

[0152] 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 one or more embodiments or examples in a suitable manner.

[0153] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0154] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, a computer program) stored thereon, and the computer-readable program instructions are used to execute the medical data monitoring and early warning method in the above-mentioned embodiment.

[0155] 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: an electrical connection with one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible storage medium that contains or stores 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.

[0156] The computer-readable storage medium may be included in the medical data monitoring and early warning device; or it may exist independently without being assembled into the medical data monitoring and early warning device.

[0157] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the medical data monitoring and early warning device, the medical data monitoring and early warning device can write computer program codes for performing the operations of the present application in one or more programming languages ​​or a combination thereof. The programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, through the Internet using an Internet service provider).

[0158] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-based implementation that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0159] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0160] The computer-readable storage medium provided in this application stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned medical data monitoring and early warning method. This computer-readable storage medium addresses the technical problem of improving the accuracy of early warnings for key medical indicators. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are similar to those of the medical data monitoring and early warning method provided in the aforementioned embodiments, and are not further elaborated here.

[0161] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned medical data monitoring and early warning method when executed by a processor.

[0162] The computer program product provided in this application can solve the technical problem of how to improve the accuracy of early warnings for key medical indicators. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the medical data monitoring and early warning methods provided in the above embodiments, and will not be elaborated here.

[0163] The above description is only part of the embodiments of the present application and does not limit the patent scope of the present application. All equivalent structural transformations made by using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A medical data monitoring and early warning method, characterized in that: The method comprises: Obtaining medical data, including outpatient volume, number of patients waiting to be seen, and medical supplies inventory data; Extracting time series features from the medical data using a preset neural network model to obtain medical data features, wherein the preset neural network model is a long short-term memory network model; Calculating the medical data features in combination with a weight coefficient matrix to obtain a comprehensive score value, wherein the weight coefficient matrix is ​​obtained based on the preset neural network model; Comparing the comprehensive score with the preset abnormality threshold score to obtain a judgment result; When the judgment result is that an abnormal situation exists, an early warning information is generated, which includes the abnormality type, occurrence time, involved indicators, current score value, threshold reference value and recommended processing strategy; Sending the warning information to a target terminal so that the target terminal performs a corresponding alarm operation; Before the step of calculating the medical data features in combination with the weight coefficient matrix to obtain a comprehensive score value, the method includes: Acquire historical medical data and construct a training dataset; Inputting the training data set into a preset neural network model to obtain a query vector and a key vector; Calculating based on the query vector and the key vector to obtain a similarity score; Performing softmax normalization processing on the similarity score to obtain a normalized result; Taking the normalized results as coefficient values ​​of a weight coefficient matrix to construct a weight coefficient matrix; Before the step of sending the warning information to the target terminal so that the target terminal performs a corresponding alarm operation, the method includes: Obtain a table of mapping relationships between warning levels and terminals and the warning levels. The terminals include an operations command terminal, a physician mobile terminal, and a drug warehouse management terminal. The operations command terminal corresponds to an abnormal situation where the outpatient volume exceeds the standard, the physician mobile terminal corresponds to an abnormal situation where the number of patients waiting to be seen exceeds the standard, and the drug warehouse management terminal corresponds to an abnormal situation where the inventory is insufficient. Based on the warning level and terminal mapping relationship table, query according to the warning level to obtain the target terminal identifier; determining a target terminal address according to the target terminal identifier; The terminal corresponding to the target terminal address is used as the target terminal.

2. The method according to claim 1, wherein The step of extracting time series features from the medical data using a preset neural network model to obtain medical data features, wherein the preset neural network model is a long short-term memory network model, comprises: Passing the medical data through a forget gate of the long short-term memory network model to generate a first intermediate state; Passing the medical data through the input gate of the long short-term memory network model to generate a second intermediate state; updating and combining the first intermediate state and the second intermediate state to obtain a time-dependent feature; Passing the temporal dependency feature through the output gate of the long short-term memory network model to generate a hidden state feature; The hidden state features are used as medical data features.

3. The method according to claim 1, wherein The step of calculating the medical data features in combination with the weight coefficient matrix to obtain a comprehensive score value includes: Converting the medical data features to obtain medical feature data; Performing outlier filtering on the medical characteristic data to obtain filtered characteristic data; Perform weighted fusion based on the filtered feature data in combination with the weight coefficient matrix to obtain fused feature data; The fused feature data is transformed through dynamic normalization to obtain a comprehensive score value.

4. The method according to claim 1, wherein The step of comparing the comprehensive score with a preset abnormality threshold score to obtain a judgment result includes: According to the comprehensive score, an outpatient volume score, a waiting list score and an inventory score are obtained; Obtaining a preset outpatient volume threshold score, a preset number of patients to be seen threshold score, and a preset inventory threshold score based on a preset abnormality threshold score, wherein the preset abnormality threshold score is dynamically adjusted based on historical medical data and environmental data; Comparing the outpatient volume score with the preset outpatient volume threshold score to obtain a first comparison result; Comparing the score of the number of patients to be seen with the preset outpatient volume threshold score to obtain a second comparison result; Comparing the inventory score with the preset inventory threshold score to obtain a third comparison result; When any one of the following conditions is met: the outpatient volume score is lower than the preset outpatient volume threshold score according to the first comparison result, the number of patients waiting to be seen score is lower than the preset number of patients waiting to be seen threshold score according to the second comparison result, and the inventory score is lower than the preset inventory threshold score according to the third comparison result, a judgment result is obtained that an abnormality exists; When the first comparison result is that the outpatient volume score is lower than the preset outpatient volume threshold score, the second comparison result is that the number of patients to be seen score is lower than the preset number of patients to be seen threshold score, and the third comparison result is that the inventory score is lower than the preset inventory threshold score, none of them are met, the judgment result is normal.

5. The method according to claim 1, wherein When the judgment result indicates that an abnormal situation exists, the step of generating early warning information includes: The abnormal situation is judged to obtain the abnormal type, which includes outpatient volume exceeding the standard, number of patients to be seen exceeding the standard, and insufficient inventory; Generate a corresponding warning level according to the abnormality type; generating warning information according to the warning level; The step of generating a corresponding warning level according to the abnormality type includes: Match the abnormality type to obtain the corresponding warning level rules; Generate a first warning level, a second warning level, and a third warning level according to the warning level rule; The first warning level, the second warning level and the third warning level are used as warning levels.

6. The method according to claim 1, wherein The step of sending the warning information to the target terminal so that the target terminal performs a corresponding alarm operation includes: Obtaining the degree of urgency and priority level according to the warning information; Sending a flashing light signal to the target terminal according to the warning information, so that the target terminal displays a flashing warning message on the screen when receiving the flashing light signal; sending a sound signal to the target terminal according to the warning information, so that the target terminal performs a sound alarm when receiving the sound signal, and the volume of the sound signal is dynamically adjusted according to the urgency; A vibration signal is sent to the target terminal according to the early warning information, so that the target terminal performs a vibration alarm when receiving the vibration signal, and the frequency and intensity of the vibration signal are set according to the priority processing level.

7. A medical data monitoring and early warning device, characterized in that: The device comprises: An acquisition module is used to acquire medical data, including outpatient volume, number of patients waiting to be seen, and medical supplies inventory data; a feature extraction module, configured to extract time series features from the medical data using a preset neural network model to obtain medical data features, wherein the preset neural network model is a long short-term memory network model; a calculation module for calculating the medical data features in combination with a weight coefficient matrix to obtain a comprehensive score, wherein the weight coefficient matrix is ​​obtained based on the preset neural network model; further for obtaining historical medical data and constructing a training data set; inputting the training data set into the preset neural network model to obtain a query vector and a key vector; performing calculations based on the query vector and the key vector to obtain a similarity score; performing softmax normalization processing on the similarity score to obtain a normalized result; and using the normalized result as a coefficient value of the weight coefficient matrix to construct the weight coefficient matrix; A judgment module is used to compare the comprehensive score with a preset abnormality threshold score to obtain a judgment result; A result module is used to generate warning information when the judgment result is that an abnormal situation exists. The warning information includes the abnormality type, occurrence time, involved indicators, current score value, threshold reference value and recommended processing strategy; The early warning module is used to send the early warning information to the target terminal so that the target terminal performs the corresponding alarm operation; it is also used to obtain the early warning level and terminal mapping relationship table and the early warning level, the terminal includes an operation command terminal, a physician mobile terminal and a drug warehouse management terminal, the operation command terminal corresponds to the abnormal situation of excessive outpatient volume, the physician mobile terminal corresponds to the abnormal situation of excessive number of patients to be seen, and the drug warehouse management terminal corresponds to the abnormal situation of insufficient inventory; based on the early warning level and terminal mapping relationship table, a query is performed according to the early warning level to obtain the target terminal identification; the target terminal address is determined according to the target terminal identification; and the terminal corresponding to the target terminal address is used as the target terminal.

8. A medical data monitoring and early warning device, characterized in that: The device includes: a memory, a processor, and a medical data monitoring and early warning program stored in the memory and running on the processor, wherein the medical data monitoring and early warning program is configured to implement the steps of the medical data monitoring and early warning method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium stores a medical data monitoring and early warning program, which, when executed by the processor, implements the steps of the medical data monitoring and early warning method according to any one of claims 1 to 6.

Images