A nursing resource allocation management method and system based on emergency triage information

Abstract

This invention discloses a nursing resource allocation and management method and system based on emergency triage information. It introduces a dynamic split weight mechanism, first calculating the global correlation density of each triage feature to quantify its statistical value in the overall population, and simultaneously calculating local mutation significance factors to capture its weak but crucial discriminative information in critically ill samples. Then, the two are combined to generate dynamic split weights, which are used to optimize the split gain of features in the prediction model. This allows the gradient boosting decision tree algorithm to overcome the limitations of traditional global gain bias towards strong features during feature selection, effectively identifying those latent critically ill key features with low global importance but significant local mutations, significantly improving the prediction accuracy for latently critically ill patients. Based on accurate risk prediction results, nursing resources are dynamically allocated, realizing a shift from passive response to proactive intervention, and solving the technical problem of existing technologies failing to intervene in a timely manner due to missed identification of high-risk patients.

Description

Technical Field

[0001] This invention belongs to the field of healthcare information technology and relates to a method and system for managing the allocation of nursing resources based on emergency triage information. Background Technology

[0002] As the front line of a hospital, the rational allocation of nursing resources in the emergency department directly affects patient safety and medical efficiency. With the accelerating aging of the population and the frequent occurrence of public health emergencies, the pressure on emergency departments is increasing daily. In the traditional emergency nursing management model, the triage system mainly relies on experienced nurses to classify patients based on their chief complaints and visible vital signs (such as instantaneous measurements of blood pressure and heart rate), referring to standards such as the Emergency Severity Index (ESI). This model is essentially a rule-matching process based on static thresholds. However, in clinical practice, there exists a special type of patient with latent critical illness, such as trauma patients in the compensatory shock phase, whose vital signs may still be within the normal reference range, or only exhibit subtle fluctuations that are extremely difficult to detect. Conventional triage systems often fail to capture these weak signals and classify them as non-urgent, resulting in the inability of nursing resources to intervene in a timely manner, which can easily lead to delays in rescue efforts.

[0003] While there is an existing method for predicting patient risk using the Gradient Boosting Decision Tree (GBDT) algorithm, this algorithm generates weak classifiers through multiple iterations and combines them linearly, exhibiting strong nonlinear fitting capabilities and assisting medical staff in identifying high-risk patients. However, when using the GBDT algorithm for risk prediction and resource allocation in emergency patients, the algorithm typically bases its calculation of the gain of each feature during the process of constructing the decision tree and finding the optimal split point on the global information gain or the decrease in the Gini index. However, emergency triage data typically exhibits high-dimensional sparsity, and key indicators of latent severe illness are often submerged among a large number of routine features. Specifically, latent severe patients constitute a very small minority in the overall emergency patient population, and their key risk indicators are often weak signals. For example, a patient may have normal blood pressure, but a slight decrease in the entropy value of their heart rate sequence; this feature contributes very little to the global data gain compared to strong features such as overt hypertension. Under the feature selection logic of traditional algorithms, the algorithm tends to prioritize strong features with high global gain and wide sample coverage for splitting, causing weak features that, although insignificant, are crucial for differentiating specific severe cases to be ignored. This bias in feature selection can lead to the model missing the diagnosis of patients with latent severe conditions, thus preventing the pre-allocation of nursing resources based on the actual risk of the patient's condition. Summary of the Invention

[0004] The purpose of this invention is to solve the problem that existing technologies, when processing high-dimensional data from emergency triage, fail to capture weak but crucial latent severe illness signals due to feature selection bias towards strong features, resulting in the model missing high-risk patients and preventing the pre-allocation of nursing resources. This invention provides a nursing resource allocation and management method and system based on emergency triage information.

[0005] To achieve the above objectives, the present invention employs the following technical solution: a nursing resource allocation and management method based on emergency triage information, comprising the following steps: acquiring patient triage data, standardizing the patient triage data to obtain triage features; calculating the global association density and local mutation significance factor of each triage feature, and calculating the dynamic split weight of each triage feature based on the global association density and local mutation significance factor; acquiring a prediction model, optimizing the split gain of features in the prediction model based on the dynamic split weight of each triage feature to obtain an optimized prediction model, classifying and predicting patient triage data based on the optimized prediction model, and managing nursing resource allocation based on the prediction results.

[0006] A further improvement of the present invention is that: the calculation of the global correlation density of each triage feature includes: in, Indicates the first Global association density of triage features; Indicates the first Triage characteristics Disease outcome label Mutual information values ​​between them; Indicates the first Information entropy of each triage feature; To prevent tiny constants with a denominator of zero; Indicates the first The standard deviation of each triage feature across all samples; Indicates the first The mean of each triage feature across all samples.

[0007] The calculation of the local mutation significance factor for each triage feature includes: assessing the distribution difference of each triage feature between severe and general samples based on KL divergence, applying redundancy penalty based on the correlation between features, and calculating the local mutation significance factor for each triage feature. In the formula, Indicates the first Significant local mutation factors in triage characteristics; Indicates the first Each triage feature is in the positive sample probability distribution With negative sample probability distribution The relative entropy between them; As a normalized mapping function, the KL divergence is mapped to the interval between 0 and 1 using the saturation property of the exponential function; Indicates the total number of features; Indicates the first The triage characteristics and the first Pearson correlation coefficients among triage characteristics.

[0008] The calculation of the dynamic splitting weight of each triage feature based on the global association density and local mutation significance factor of each triage feature includes: In the formula, Indicates the first Dynamic splitting weights for each triage feature; Indicates the first Global association density of triage features; Indicates the first Significant factors of local mutations in triage characteristics.

[0009] The acquisition of the prediction model, which optimizes the split gain of features in the prediction model based on the dynamic split weight of each triage feature, to obtain the optimized prediction model, includes: introducing a gradient boosting decision tree model, setting initial parameters, and starting training the model; in each iteration of building the decision tree, calculating the weighted split gain of all candidate split features of the current node based on the dynamic split weight of each triage feature. in, This represents the split gain calculated by the original gradient boosting decision tree model; Represents the dynamic split weight of each triage feature; Indicates the first The weighted split gain of each feature is calculated; the feature corresponding to the maximum value among all candidate split features is selected as the optimal split feature of the current node, and the optimal split point is determined based on the optimal split feature; the sample data in the current node is divided into left child nodes and right child nodes based on the optimal split point, and the above process is recursively executed based on the left child nodes and right child nodes until the stopping condition is met, and the optimized prediction model is obtained.

[0010] The process of classifying and predicting patient triage data based on the optimized prediction model, and allocating and managing nursing resources according to the prediction results, includes: inputting the collected real-time patient triage data into the optimized prediction model, and outputting the probability value of the patient belonging to the latent severe illness category. Set risk thresholds; when At that time, the patient was determined to be a high-risk group for latent severe illness, and the current patient's nursing level was adjusted to Level I standard; when At that time, the nursing level of the current patient is assigned according to the original allocation rules.

[0011] A nursing resource allocation and management system based on emergency triage information includes: a data acquisition module for acquiring patient triage data, standardizing the patient triage data to obtain triage features; a dynamic split weight calculation module for calculating the global association density and local mutation significance factor of each triage feature, and calculating the dynamic split weight of each triage feature based on the global association density and local mutation significance factor; and a prediction module for acquiring a prediction model, optimizing the split gain of features in the prediction model based on the dynamic split weight of each triage feature to obtain an optimized prediction model, classifying and predicting patient triage data based on the optimized prediction model, and managing nursing resource allocation based on the prediction results.

[0012] A computer program product includes a computer program that, when executed by a processor, implements any one of the methods described.

[0013] A terminal device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of any of the methods described above.

[0014] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of any of the methods described herein.

[0015] Compared with existing technologies, this invention has the following beneficial effects: This invention discloses a nursing resource allocation and management method based on emergency triage information. It introduces a dynamic split weight mechanism. First, it calculates the global association density of each triage feature to quantify its statistical value in the overall population. At the same time, it calculates the local mutation significance factor to capture its weak but key discriminative information in critically ill samples. Then, it combines the two to generate dynamic split weights, which are used to optimize the split gain of features in the prediction model. This allows the gradient boosting decision tree algorithm to break through the limitations of traditional global gain bias towards strong features in the feature selection process, effectively identifying those hidden critically ill key features with low global importance but significant local mutations, and significantly improving the prediction accuracy of hidden critically ill patients. Based on accurate risk prediction results, it dynamically allocates nursing resources, realizing the transformation from passive response to proactive intervention. It solves the technical problem of existing technologies where nursing resources cannot be intervened in a timely manner due to missed identification of high-risk patients, and ultimately improves the efficiency of emergency department resource allocation. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart disclosed in an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0019] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0020] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0021] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper," "lower," "horizontal," or "inner" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, terms such as "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0022] Furthermore, the use of the term "horizontal" does not imply that the component must be absolutely horizontal, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0023] In the description of the embodiments of the present invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.

[0024] The present invention will now be described in further detail with reference to the accompanying drawings: See also Figure 1 This invention discloses a nursing resource allocation management method and system based on emergency triage information. By analyzing the global correlation and local mutation significance of triage features, the dynamic splitting weight of the features is calculated, thereby forcibly improving the expression ability of key weak signals during the training process, and realizing the accurate identification of patients with hidden critical illnesses and the pre-allocation of nursing resources.

[0025] Example 1: This embodiment of the invention discloses a nursing resource allocation and management method based on emergency triage information, including the following steps: Step 1: Collect triage vital sign data and chief complaint information of emergency patients; Step 2: Analyze the global correlation and local mutation significance of features, calculate the dynamic split weight of features, and optimize the feature selection process of the algorithm; Step 3: Use the optimized algorithm to predict the hidden critical illness risk of patients, and dynamically allocate nursing resources accordingly.

[0026] The solution disclosed in this invention will be further described below with reference to Embodiment 2.

[0027] Example 2: This embodiment of the invention discloses a nursing resource allocation and management method based on emergency triage information, including the following steps: Step 1: Collect triage vital signs data and chief complaint information of emergency patients.

[0028] The specific steps include: This step collects patient triage data through the monitoring equipment and electronic medical record system at the emergency triage station. The collected data includes basic vital sign sequences, collected once per minute, for a total of 10 minutes prior to triage, as well as unstructured text information. For unstructured text, natural language processing techniques are used to extract keywords and perform one-hot encoding. Specifically, for patient complaints and medical history, word segmentation and TF-IDF algorithms are applied to filter out core symptom words such as "chest pain" and "hypertension." These keywords are then mapped into high-dimensional sparse vectors based on a feature dictionary, with a value of 1 assigned to the keyword index position and 0 assigned to the remaining positions, thus converting natural language into discrete numerical features that the model can compute. Simultaneously, the patient's subsequent emergency outcome labels are obtained for model training. For continuous vital sign data, a minimum-maximum normalization method is used to map all feature values ​​to a closed interval between 0 and 1, obtaining a standardized feature set to ensure mathematical validity and dimensional consistency in subsequent calculations.

[0029] Furthermore, in this embodiment, the basic vital signs sequence includes heart rate, systolic blood pressure, diastolic blood pressure, blood oxygen saturation, and respiratory rate, and the unstructured text information includes the chief complaint and past medical history.

[0030] Furthermore, in this embodiment, the emergency outcome label is used to indicate whether a severe outcome has occurred.

[0031] Step 2: Analyze the global correlation and local mutation significance of features, calculate the dynamic split weight of features, and optimize the feature selection process of the algorithm.

[0032] The core objective of this step is to accurately identify patients with latent severe illnesses using the gradient boosting decision tree algorithm in order to allocate resources. During the process, the feature splitting gain when the algorithm constructs the decision tree will be weighted and optimized. This is mainly based on the analysis of the global correlation between features and outcome labels and the local performance of features in severe cases.

[0033] Specifically, this step optimizes the feature splitting process by including the following steps: Step 2.1: Analyze the statistical dependency between each triage feature and the patient's disease outcome label to obtain the global association density of each triage feature; Step 2.2: Analyze the distribution differences of each triage feature between positive and negative samples and the redundancy relationship between features, correct the global association density, and obtain the local mutation significance factor of each triage feature; Step 2.3: Combine the global association density and the local mutation significance factor to calculate the dynamic splitting weight of each triage feature.

[0034] The specific steps are as follows: Step 2.1: Analyze the statistical dependency between each triage feature and the patient's condition outcome label to obtain the global association density of each triage feature.

[0035] This step aims to screen for features valuable for assessing the condition from a macro-statistical perspective. In information theory, mutual information is a commonly used indicator to measure the interdependence between two random variables. The greater the mutual information between a triage feature and the disease outcome label, the more information the feature contains about the condition, and the higher its global importance. However, emergency room data contains a lot of noise, and simple mutual information calculation may be dominated by highly variable noisy features. Therefore, it is necessary to introduce the statistical stability of features as a constraint. If the coefficient of variation (standard deviation divided by mean) of a feature in the overall population is smaller, it indicates that the measurement of the feature is more stable, the mutual information it contains is more reliable, and the corresponding global association density should be greater. Based on the above logic, the global association density of each triage feature is calculated: In the formula, Indicates the first Global association density of triage features; Indicates the first Triage characteristics Disease outcome label Mutual information values ​​between them; Indicates the first Information entropy of each triage feature (calculated based on the probability distribution after histogram discretization, ensuring the value is non-negative); To prevent the use of tiny constants with a denominator of zero, a value of 0.001 is used; Indicates the first The standard deviation of each triage feature across all samples Indicates the first The mean of each triage feature across all samples.

[0036] Furthermore, in the expression disclosed in this step, The larger the value, the higher the relevance of the information. The higher the statistical dependence of a triage feature on the disease outcome, the more helpful the feature is in judging the disease, and the greater its corresponding global correlation density. The purpose of the denominator is to punish features whose mutual information is artificially inflated due to excessively dispersed values. The coefficient of variation, representing a characteristic, reflects the relative dispersion of the data. The larger the coefficient of variation (i.e., the smaller the coefficient of variation), the more stable the feature is in the overall population from the perspective of data stability, the higher the reliability of its statistical regularity, and the greater the corresponding global association density.

[0037] Step 2.2: Analyze the distribution differences of each triage feature between positive and negative samples and the redundancy relationship between features, correct the global correlation density, and obtain the local mutation significance factor of each triage feature.

[0038] While the global association density calculated through the above steps reflects the general importance of features, it risks failure in the specific scenario of identifying latent severe cases. Latent severe cases account for a very small percentage of total emergency room visits, and their pathological characteristics are often subtle and specific. Under this asymmetric distribution, the global mutual information of this feature is diluted by a massive number of normal samples, leading to a severe underestimation of the calculated global association density. If subsequent calculations use multiplicative logic, this feature will be ignored due to its small cardinality. Therefore, a local mutation significance factor is introduced. The greater the difference in probability distribution (KL divergence) between positive samples (severe cases) and negative samples (normal cases) for a feature, the stronger its local discriminative power, and it should receive a high weight. Simultaneously, to avoid redundancy caused by multiple highly correlated features simultaneously occupying high weights, a correlation penalty is introduced. Based on the above logic, the local mutation significance factor for each triage feature is calculated: In the formula, Indicates the first Significant local mutation factors in triage characteristics; Indicates the first Each triage feature is in the positive sample probability distribution With negative sample probability distribution The relative entropy (KL divergence) between them. As a normalized mapping function, the KL divergence is mapped to the interval between 0 and 1 using the saturation property of the exponential function; Indicates the total number of features; Indicates the first The triage characteristics and the first Pearson correlation coefficients among triage characteristics.

[0039] Furthermore, in the expression disclosed in this step The larger the value, the more significant the distributional differences. The triage characteristics exhibit distinctly different behaviors in critically ill patients compared to the general patient population. The stronger the characteristic's ability to detect latent risks, the larger its corresponding local mutation significance factor. (The denominator contains...) The larger the value, the higher the information redundancy level. The higher the overlap between a triage feature and other features, the less new information it provides. In order to force the model to focus on unique features, the corresponding local mutation significance factor should be smaller. The setting is to ensure that the baseline value of the factor is greater than 1, so that it plays an amplifying rather than suppressing role in the exponentiation operation in subsequent steps.

[0040] Step 2.3: Combine the global association density and the local mutation significance factor to calculate the dynamic split weight of each triage feature.

[0041] Through the above analysis, the global association density, reflecting the universal value of a feature, and the local mutation significance factor, reflecting the unique value of a feature in critical care identification, were obtained. To address the issue that the global association density calculated in step 2.1 might be too small due to data sparsity, leading to the suppression of weak signals, this step employs a weighted compensation strategy to construct dynamic split weights. When the global association density of a feature is low, if its local mutation significance factor is extremely high, the compensation effect of the additive term can still ensure that the feature obtains a high final weight. Based on the above logic, the dynamic split weights for each triage feature are calculated: In the formula, Indicates the first Dynamic splitting weights for each triage feature; Indicates the first Global association density of triage features; Indicates the first Significant factors of local mutations in triage characteristics.

[0042] Furthermore, the expression disclosed in this step can take into account both the global applicability of common pathological features and the local specificity of latent severe features when calculating the split gain, thus solving the technical problem of high-dimensional features drowning out weak signals.

[0043] Step 3: Use the optimized gradient boosting decision tree algorithm to predict the patient's hidden risk of serious illness and dynamically allocate nursing resources accordingly.

[0044] This step utilizes the dynamic split weights calculated in step 2 to improve the algorithm's training process and apply it to a real-world scenario. The steps include: Step 3.1: Model Construction and Training; Initialize model parameters, setting the number of iteration trees to 100, the learning rate to 0.1, and the maximum tree depth to 6. In each iteration of building the decision tree, for all candidate split features of the current node, instead of directly using the original Gini index decrease or mean squared error decrease as the splitting basis, a weighted split gain is calculated. Specifically, the... Weighted split gain of each feature The calculation expression is: in, The splitting gain calculated using the original GBDT algorithm. The dynamic splitting weight of the feature calculated in step 2.3.

[0045] Further, select The largest feature is used as the best splitting feature for the current node, and the best splitting point is determined.

[0046] Furthermore, based on the optimal split point, the sample data within the current node is divided into two subsets, left and right, and assigned to the newly generated left and right child nodes respectively. Subsequently, the weighted split gain calculation and sample partitioning process is recursively repeated for the newly generated child nodes, continuously growing the decision tree downwards until the stopping condition is met, i.e., reaching the maximum tree depth limit of 6. Once a decision tree has completed the splitting and growth of all nodes, the output value of each leaf node is calculated to fit the current prediction residual, and the tree is added to the overall model with a preset learning rate. Then, the next iteration begins to build the next tree, until all 100 iterative trees are constructed.

[0047] Step 3.2: Predicting the Risk of Latent Severe Illness; Input the real-time collected triage data of emergency patients into the trained and optimized model. The model outputs the probability value that the patient belongs to the category of latent severe illness. Triage data includes heart rate, systolic blood pressure, diastolic blood pressure, blood oxygen saturation, respiratory rate, chief complaint, and past medical history.

[0048] Step 3.3: Dynamically allocate nursing resources; set risk thresholds. .

[0049] like The system determines that the patient is at high risk of latent severe illness, triggering pre-treatment instructions: even if the patient is classified as level III or IV, immediately upgrade their care level to level I; lock the reserved bed in the resuscitation room; if Nursing resources are allocated according to the standard emergency triage and allocation criteria. The standard emergency triage and allocation criteria disclosed in this embodiment are the original allocation criteria.

[0050] Example 3: This embodiment of the invention discloses a nursing resource allocation and management system based on emergency triage information, comprising the following steps: a data acquisition module, used to acquire patient triage data, standardize the patient triage data to obtain triage features; a dynamic split weight calculation module, used to calculate the global association density and local mutation significance factor of each triage feature, and calculate the dynamic split weight of each triage feature based on the global association density and local mutation significance factor; a prediction module, used to acquire a prediction model, optimize the split gain of features in the prediction model based on the dynamic split weight of each triage feature to obtain an optimized prediction model, classify and predict patient triage data based on the optimized prediction model, and manage nursing resource allocation based on the prediction results.

[0051] This invention significantly improves the gradient boosting decision tree algorithm's ability to capture weak but crucial signals when processing high-dimensional medical data by introducing a dynamic splitting weight mechanism for features. The method first calculates the global correlation density of features to quantify their universal value from a macro-statistical perspective. Then, addressing the failure of global indicators in handling rare but fatal features (i.e., some features only show abnormalities in a few critically ill samples), a local mutation significance factor is introduced for correction. Dynamic splitting weights are constructed by analyzing the differences in the probability distribution of features between critically ill and ordinary samples and their redundancy with other features. This optimization allows the algorithm to focus on weak features with low global gain but life-saving potential in critical moments when constructing the decision tree, avoiding the missed diagnosis of patients with latent critical illnesses. Furthermore, based on accurate risk prediction, the system can shift nursing resources, such as emergency room beds and senior nurse rounds, from traditional post-event remediation to pre-event proactive measures, greatly improving the safety and efficiency of emergency department operations.

[0052] A schematic diagram of a terminal device according to an embodiment of the present invention. The terminal device of this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the various method embodiments described above. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the various device embodiments described above.

[0053] The computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention.

[0054] The terminal device can be a desktop computer, laptop computer, cloud server, or other device with strong computing power. The terminal device may include, but is not limited to, a processor and memory.

[0055] The optimal choice for the processor is a multi-core high-speed central processing unit (CPU).

[0056] The memory can be used to store the computer program and / or module. The processor implements various functions of the terminal device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory.

[0057] If the modules / units integrated into the terminal device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0058] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for managing the allocation of nursing resources based on emergency triage information, characterized in that, Includes the following steps: Obtain patient triage data, standardize the patient triage data, and obtain triage features; Calculate the global association density and local mutation significance factor for each triage feature, and calculate the dynamic split weight for each triage feature based on the global association density and local mutation significance factor; obtain the prediction model, optimize the split gain of the features in the prediction model based on the dynamic split weight of each triage feature, and obtain the optimized prediction model; classify and predict the triage data of patients based on the optimized prediction model, and manage the allocation of nursing resources based on the prediction results.

2. The nursing resource allocation and management method based on emergency triage information according to claim 1, characterized in that, The calculation of the global association density for each triage feature includes: in, Indicates the first Global association density of triage features; Indicates the first Triage characteristics Disease outcome label Mutual information values ​​between them; Indicates the first Information entropy of each triage feature; To prevent tiny constants with a denominator of zero; Indicates the first The standard deviation of each triage feature across all samples; Indicates the first The mean of each triage feature across all samples.

3. The nursing resource allocation and management method based on emergency triage information according to claim 1, characterized in that, The calculation of the local mutation significance factor for each triage feature includes: assessing the distribution difference of each triage feature between severe and general samples based on KL divergence, applying redundancy penalty based on the correlation between features, and calculating the local mutation significance factor for each triage feature. In the formula, Indicates the first Significant local mutation factors in triage characteristics; Indicates the first Each triage feature is in the positive sample probability distribution With negative sample probability distribution The relative entropy between them; As a normalized mapping function, the KL divergence is mapped to the interval between 0 and 1 using the saturation property of the exponential function; Indicates the total number of features; Indicates the first The triage characteristics and the first Pearson correlation coefficients among triage characteristics.

4. The nursing resource allocation and management method based on emergency triage information according to claim 1, characterized in that, The calculation of the dynamic splitting weight of each triage feature based on the global association density and local mutation significance factor of each triage feature includes: In the formula, Indicates the first Dynamic splitting weights for each triage feature; Indicates the first Global association density of triage features; Indicates the first Significant factors of local mutations in triage characteristics.

5. A nursing resource allocation and management method based on emergency triage information according to claim 1, characterized in that, The acquisition of the prediction model, which optimizes the split gain of features in the prediction model based on the dynamic split weight of each triage feature, to obtain the optimized prediction model, includes: introducing a gradient boosting decision tree model, setting initial parameters, and starting training the model; in each iteration of building the decision tree, calculating the weighted split gain of all candidate split features of the current node based on the dynamic split weight of each triage feature. in, This represents the split gain calculated by the original gradient boosting decision tree model; Represents the dynamic split weight of each triage feature; Indicates the first The weighted split gain of each feature is calculated; the feature corresponding to the maximum value among all candidate split features is selected as the optimal split feature of the current node, and the optimal split point is determined based on the optimal split feature; the sample data in the current node is divided into left child nodes and right child nodes based on the optimal split point, and the above process is recursively executed based on the left child nodes and right child nodes until the stopping condition is met, and the optimized prediction model is obtained.

6. A nursing resource allocation and management method based on emergency triage information according to claim 1, characterized in that, The process of classifying and predicting patient triage data based on the optimized prediction model, and allocating and managing nursing resources according to the prediction results, includes: inputting the collected real-time patient triage data into the optimized prediction model, and outputting the probability value of the patient belonging to the latent severe illness category. Set risk thresholds; when When the patient is identified as a high-risk individual for latent severe illness, the current nursing care level is adjusted to Level I standard; when At that time, the nursing level of the current patient is assigned according to the original allocation rules.

7. A nursing resource allocation and management system based on emergency triage information, characterized in that, include: The data acquisition module is used to acquire patient triage data, standardize the patient triage data, and obtain triage features; The dynamic split weight calculation module is used to calculate the global association density and local mutation significance factor of each triage feature, and calculate the dynamic split weight of each triage feature based on the global association density and local mutation significance factor. The prediction module is used to obtain the prediction model, optimize the split gain of the features in the prediction model based on the dynamic split weight of each triage feature, obtain the optimized prediction model, classify and predict the triage data of patients based on the optimized prediction model, and manage the allocation of nursing resources based on the prediction results.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-6.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-6.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-6.

Images