A consumable prediction and medical cost management method, device and equipment

Abstract

The present application relates to the technical field of medical consumable management, and particularly relates to a method, device and equipment for consumable prediction and medical cost control, which comprises the following steps: obtaining basic information of a patient, nursing behavior data, consumable flow data and medical insurance project data; determining the consumption driving strength of nursing behavior on consumable consumption, the correlation between consumable consumption and medical insurance project profit and loss, and the dependence strength of the patient on nursing behavior; constructing a full-link causal mapping relationship for different nursing behaviors based on the basic information of the patient, the nursing behavior data, the consumable flow data, the medical insurance project data and the quantified correlation; constructing a consumable demand prediction model based on the endogenous constraint of medical insurance project profit and loss; and predicting the consumable demand of a target patient and the profit and loss of a medical insurance project based on the full-link causal mapping relationship for different nursing behaviors and the consumable demand prediction model, so as to improve the medical service efficiency and the operation efficiency of a hospital.

Description

Technical Field

[0001] This invention relates to the field of medical consumables management technology, and in particular to a method, apparatus and equipment for consumables prediction and medical cost control. Background Technology

[0002] With the continuous advancement of medical technology and the improvement of medical service levels, the types and usage of medical consumables are increasing year by year, especially interventional consumables implanted in the human body. These consumables are divided into low-value interventional consumables and high-value interventional consumables according to their price. Low-value interventional consumables are numerous, frequently used, and consumed quickly, while high-value interventional consumables are expensive, have high storage requirements, and require strict control over their effectiveness and safety.

[0003] Currently, hospitals mostly procure medical consumables through periodic purchasing. However, to achieve automatic replenishment, real-time monitoring and data analysis are needed during the use of medical consumables. This can not only optimize inventory management but also improve the utilization efficiency of medical resources. However, existing medical consumable management systems still have many problems. First, traditional consumable management methods often rely on manual recording and management, which is prone to data omissions and inaccuracies, making it difficult to track consumable usage and potentially leading to inaccurate replenishment timing. Second, existing systems lack emphasis on the real-time nature and accuracy of consumable data, resulting in chaotic inventory management and resource waste.

[0004] Furthermore, controlling the cost of surgical medications is a crucial component of hospital cost control, significantly contributing to alleviating the burden on patients and improving the utilization rate of medical resources. However, the review and decision-making processes still require substantial manpower, and the inherent subjectivity in physicians' prescribing behavior makes it difficult to achieve optimal control over surgical medication costs. Therefore, there is an urgent need for a method that can accurately predict and control the demand for medical consumables by combining factors such as nursing data, disease type, surgery, nursing level, complications, season, and bed days, in order to improve the efficiency of medical services and hospital operations. Summary of the Invention

[0005] In view of the above problems, the present invention provides a method, apparatus and equipment for consumables prediction and medical cost control that overcomes or at least partially solves the above problems.

[0006] In a first aspect, the present invention provides a method for consumables prediction and medical cost control, comprising: Acquire basic patient information, dynamic data on nursing behavior over time, data on the entire lifecycle of consumables, and data on medical insurance programs; Based on the time-series dynamic data of nursing behaviors and the full life cycle circulation data of consumables, the consumption driving intensity of nursing behaviors on consumable consumption is determined. Based on the full life cycle circulation data of the consumables and the medical insurance program data, the correlation between consumable consumption and the profit and loss of medical insurance programs is determined. Based on the patient's basic information and the time-series dynamic data of the nursing behaviors, the patient's dependence on the nursing behaviors is determined; Based on the patient's basic information, time-series dynamic data of nursing behaviors, data on the entire life cycle of consumables, data on medical insurance projects, the intensity of consumption drive, correlation and dependence, a full-link causal mapping relationship is constructed for different nursing behaviors; Construct a consumables demand forecasting model based on the endogenous constraints of profit and loss of medical insurance programs; Based on the full-link causal mapping relationship for different nursing behaviors and the consumable demand prediction model based on the endogenous constraint of medical insurance project profit and loss, the consumable demand for target patients and the profit and loss situation of medical insurance projects are predicted, and the prediction results are obtained.

[0007] Preferably, based on the time-series dynamic data of nursing behaviors and the full life-cycle circulation data of consumables, the consumption driving intensity of nursing behaviors on consumable consumption is determined, including: Based on the time-series dynamic data of the nursing behaviors and the full life-cycle circulation data of the consumables, the ratio of the number of cases using any consumable when performing any nursing behavior to the total number of cases performing the corresponding nursing behavior, the specificity of any consumable to any nursing behavior, and the Granger causality strength of any nursing behavior on the consumption of any consumable are determined. The consumption drive strength of nursing actions on consumable consumption is determined by the ratio of the number of cases using any consumable during any nursing action to the total number of cases performing the corresponding nursing action, the specificity of any consumable to any nursing action, and the Granger causal strength of any nursing action on the consumption of any consumable. This is calculated using the following formula: in, To perform any nursing action Use any consumable The number of cases and the implementation of corresponding nursing care The ratio of the total number of cases, For any consumable For any nursing action Exclusivity For any nursing action For any consumable The Granger causality strength consumed, For L2 normalization term, For target consumables, The intensity of the consumption drive for consumables by nursing behaviors.

[0008] Preferably, based on the entire lifecycle circulation data of the consumables and the medical insurance program data, the correlation between consumable consumption and the profit and loss of medical insurance programs is determined, including: Based on the full life cycle circulation data of the consumables and the medical insurance project data, the average cost of consumables per case for any disease group in the medical insurance project data, the benchmark average cost of consumables per case for any disease group in the medical insurance project data, the utilization rate of any consumables in any disease group in the medical insurance project data, the overspending rate of any case for any consumables in any disease group in the medical insurance project data, and the standard deviation of the average cost of consumables per case for any disease group in the medical insurance project data are determined. Based on the average cost per case of any consumable used in any disease group within the medical insurance program data, the benchmark average cost per case of consumables for any disease group within the medical insurance program data, the utilization rate of any consumable in any disease group within the medical insurance program data, the overspending rate of any consumable used in any disease group within the medical insurance program data, and the standard deviation of the average cost per case of consumables for any disease group within the medical insurance program data, the correlation between consumable consumption and the profit and loss of the medical insurance program is determined using the following formula: in, For any disease group in the medical insurance program data Use any consumable The average cost of consumables per case, For any disease group in the medical insurance program data The corresponding medical insurance benchmark cost per consumable item For any disease group in the medical insurance program data Any consumable in the corresponding category Utilization rate For any disease group in the medical insurance program data Use any consumable The case overspending rate This represents the standard deviation of the average cost of consumables per case in any disease group within the medical insurance program data. The correlation between consumable consumption and the profitability of medical insurance programs.

[0009] Preferably, based on the patient's basic information and the time-series dynamic data of the nursing behaviors, determining the patient's dependence on the nursing behaviors includes: Based on the patient's basic information and the time-series dynamic data of the nursing behaviors, the first degree of fit of the patient to perform any nursing behavior in the disease group of the medical insurance program is determined, the second degree of fit of the patient's surgical level, nursing level and nursing behavior, and the third degree of fit of the patient's complications and nursing behavior. Based on the first fit, the second fit, and the third fit, the patient's dependence on the nursing care behavior is determined using the following calculation formula: in, For the first fit, For the second degree of fit, For the third degree of fit, The weight corresponding to the first fit. The weight corresponding to the second fit. The weight corresponding to the third fitness. , as well as Determined using the entropy weight method, The degree of dependence of the patient on the aforementioned nursing care.

[0010] Preferably, based on the patient's basic information, dynamic data of nursing behavior over time, data on the entire lifecycle of consumables, data on medical insurance items, the intensity of the consumption drive, the degree of correlation, and the intensity of dependence, a full-link causal mapping relationship for different nursing behaviors is constructed, including: Based on the patient's basic information, the time-series dynamic data of nursing behavior, the whole life cycle circulation data of consumables, and the medical insurance project data, the entities, entity attributes, and time-series relationships at each level are determined. Based on the entities, entity attributes, temporal relationships, consumption driving strength, correlation degree, and dependency strength at each level, the causal relationships between entities are determined. Based on the causal relationships between the entities, a full-link causal mapping relationship is constructed for different nursing behaviors.

[0011] Preferably, a consumable demand forecasting model based on the endogenous constraint of profit and loss of medical insurance programs is constructed, including: By embedding the profit and loss constraints of medical insurance programs into the consumables demand forecasting model, the total loss function of the consumables demand forecasting model takes the following form: in, To smooth out L1 loss, , This represents the actual amount of consumables used. To predict the usage of consumables, For the types and quantities of consumables, The loss is constrained by business rules. Specifically, when the profit and loss constraints of medical insurance programs are embedded in the consumable demand forecasting model, a strong penalty is generated when the predicted cost of consumables exceeds the disease group cost ceiling of the medical insurance program. , The total number of cases in the training sample. For cases Predicted consumable costs, , The unit price of any consumable. For cases any consumable Predicted usage of consumables For cases The upper limit of consumable costs for the disease group under the relevant medical insurance program. , The payment standard for disease groups in the medical insurance program, The percentage of the baseline consumable cost for the patient group. For the safety margin coefficient, For L2 regularization, , For all weights of the consumables demand forecasting model, For any weight in the consumables demand forecasting model, To smooth the weights corresponding to L1 loss, To predict the weight of the strong penalty term that occurs when the cost of consumables exceeds the disease group cost limit of the medical insurance program, These are the weights corresponding to L2 regularization; A consumable demand forecasting model based on the endogenous constraints of profit and loss of medical insurance projects was constructed.

[0012] Preferably, after predicting the consumable demand for the target patient's individual characteristics and the profit and loss situation of the medical insurance program based on the full-link causal mapping relationship for different nursing behaviors and the consumable demand prediction model based on the endogenous constraint of the medical insurance program's profit and loss, and obtaining the prediction results, the method further includes: Based on the prediction results, a dynamic safety stock optimization model is constructed to dynamically adjust the safety stock level, where the safety stock level is the inventory level of consumables.

[0013] Preferably, after predicting the demand for consumables based on the individual characteristics of the target patient and the profit and loss situation of the medical insurance program, a dynamic safety stock optimization model is constructed to dynamically adjust the safety stock level. The safety stock level is the inventory level of consumables, and the model further includes: Based on the aforementioned safety stock level, alternative consumables are recommended.

[0014] Secondly, the present invention also provides an apparatus for consumable prediction and medical cost control, comprising: The acquisition module is used to acquire patients' basic information, dynamic data of nursing behavior over time, data on the entire life cycle of consumables, and data on medical insurance items. The first determining module is used to determine the intensity of the consumption drive of nursing behavior on consumable consumption based on the time-series dynamic data of nursing behavior and the full life cycle circulation data of consumables. The second determining module is used to determine the correlation between consumable consumption and the profit and loss of medical insurance projects based on the full life cycle circulation data of the consumables and the medical insurance project data. The third determining module is used to determine the patient's dependence on the nursing behavior based on the patient's basic information and the time-series dynamic data of the nursing behavior. The first construction module is used to construct a full-link causal mapping relationship for different nursing behaviors based on the patient's basic information, dynamic data of nursing behavior time sequence, data on the entire life cycle of consumables, data on medical insurance projects, the consumption driving strength, correlation and dependence strength. The second construction module is used to construct a consumable demand prediction model based on the endogenous constraint of the profit and loss of medical insurance projects; the prediction module is used to predict the consumable demand of the target patient's individual characteristics and the profit and loss of medical insurance projects based on the full-link causal mapping relationship for different nursing behaviors and the consumable demand prediction model based on the endogenous constraint of the profit and loss of medical insurance projects, and obtain the prediction results.

[0015] Thirdly, the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described in the first aspect.

[0016] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect.

[0017] One or more technical solutions in the embodiments of the present invention have at least the following technical effects or advantages: This invention provides a method for consumables prediction and medical cost control, comprising: acquiring basic patient information, time-series dynamic data of nursing behaviors, full life-cycle circulation data of consumables, and medical insurance project data; determining the consumption driving intensity of consumables consumption during nursing behaviors based on the time-series dynamic data of nursing behaviors and the full life-cycle circulation data of consumables; determining the correlation between consumables consumption and the profit and loss of medical insurance projects based on the full life-cycle circulation data of consumables and the medical insurance project data; determining the patient's dependence on nursing behaviors based on the patient's basic information and time-series dynamic data of nursing behaviors; constructing a full-link causal mapping relationship for different nursing behaviors based on the patient's basic information, time-series dynamic data of nursing behaviors, full life-cycle circulation data of consumables, medical insurance project data, consumption driving intensity, correlation, and dependence intensity; constructing a consumables demand prediction model based on the endogenous constraint of the profit and loss of medical insurance projects; and predicting the consumables demand and the profit and loss of medical insurance projects for the individual characteristics of target patients based on the full-link causal mapping relationship for different nursing behaviors and the consumables demand prediction model based on the endogenous constraint of the profit and loss of medical insurance projects, so as to improve the efficiency of medical services and the operational efficiency of hospitals. Attached Figure Description

[0018] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This diagram illustrates the steps of the method for consumable prediction and medical cost control in an embodiment of the present invention. Figure 2 A schematic diagram of the device for consumable prediction and medical cost control in an embodiment of the present invention is shown; Figure 3 A schematic diagram of the structure of a computer device for implementing a method for consumables prediction and medical cost control in an embodiment of the present invention is shown. Detailed Implementation

[0019] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the invention and to fully convey the scope of the invention to those skilled in the art.

[0020] Example 1: Embodiments of the present invention provide a method for consumables prediction and medical cost control, such as... Figure 1 As shown, it includes: S101, acquire basic patient information, dynamic data of nursing behavior over time, data on the entire life cycle of consumables, and data on medical insurance items; S102, Based on the time-series dynamic data of nursing behavior and the full life cycle circulation data of consumables, determine the intensity of the consumption drive of nursing behavior on consumable consumption; S103, Based on the entire life cycle circulation data of consumables and medical insurance project data, determine the correlation between consumable consumption and the profit and loss of medical insurance projects; S104, Based on the patient's basic information and the time-series dynamic data of nursing behaviors, determine the patient's dependence on the nursing behaviors; S105, based on the patient's basic information, the time-series dynamic data of nursing behavior, the whole life cycle circulation data of consumables, medical insurance project data, consumption driving intensity, correlation and dependence intensity, construct the full-link causal mapping relationship for different nursing behaviors; S106, Construct a consumable demand forecasting model based on the endogenous constraints of profit and loss of medical insurance projects; S107, based on the full-link causal mapping relationship for different nursing behaviors and the consumable demand prediction model based on the endogenous constraint of medical insurance program profit and loss, predicts the consumable demand for individual characteristics of target patients and the profit and loss of medical insurance program, and obtains the prediction results.

[0021] First, execute S101 to obtain the patient's basic information, dynamic data on nursing behavior over time, data on the entire life cycle of consumables, and data on medical insurance items.

[0022] To avoid the limitations of conventional methods that only collect static patient data, this invention, in addition to acquiring basic patient information, also obtains dynamic data on the time sequence of patient care behaviors, data on the entire lifecycle of consumables, and data on medical insurance programs. This achieves data coverage across the entire patient treatment cycle, the entire lifecycle of consumables, and medical insurance programs, providing reliable data support for subsequent analysis and processing.

[0023] The following is a detailed introduction to each type of data: The patient's basic information includes: age, gender, disease group classification, surgical code, surgical complexity level, nursing level, complication code, admission bed days, treatment effect level, medical insurance items corresponding to the disease group classification, disease group weight, basic consumable cost ratio, historical profit and loss data, etc.

[0024] The dynamic data of nursing behavior time sequence includes: nursing operation data of the entire process of patients from admission to discharge, including nursing operation code, operation name, time sequence node, operation frequency and corresponding duration, qualification of the performing nurse, nursing stage, and real-time reimbursement record of corresponding consumables.

[0025] The data on the entire lifecycle of medical consumables includes: the collection of basic attributes of medical consumables (type, specifications, unit price, classification, medical insurance charging attributes), inventory data, and the entire process of consumables application, issuance, receipt, use, return, and billing, as well as alternative consumable catalogs, procurement lead time, historical price fluctuation data, etc.

[0026] The medical insurance program data includes: the historical average cost of consumables per case, the average total medical cost per case, the medical insurance payment standard, the profit and loss ratio, the proportion of cases with excessive consumable costs, the cost structure details, etc. for each medical insurance program disease group.

[0027] After acquiring this data, it is cleaned. Specifically, an automated cleaning scheme is adopted, such as using the Isolation Forest algorithm to identify and process outliers, using multiple imputation to fill missing values, and using a combination of rule matching and semantic deduplication to remove duplicate records. For unstructured nursing records and surgical records, a BERT pre-trained model is used for semantic recognition and structured information extraction to ensure the integrity and accuracy of the data.

[0028] Next, temporal alignment and standardization processing are required, including feature classification based on disease type, surgical level, nursing level, and nursing operation type. A combination of one-hot encoding and label encoding is used to map the feature vectors into low-dimensional dense feature vectors.

[0029] Time alignment refers to aligning all nursing procedures, consumable usage, treatment behaviors, and cost data to the same timeline, using the patient's admission time as the baseline zero point. This constructs a time sequence with the previous hospitalization day as the unit and nursing behavior as the smallest unit, avoiding the problems of time sequence chaos and inability to capture dynamic changes in existing solutions.

[0030] To address numerical characteristics such as consumable costs, usage, and bed days, the DRG (Diagnosis Related Groups) adaptive Z-score normalization method was adopted to eliminate differences in characteristic distributions between different disease groups.

[0031] Next, for time series features, multi-scale sliding windows of 3 days, 7 days, and 14 days can be used to extract time series statistical features such as the frequency of nursing behaviors, the total amount of consumables consumed, and the probability of complications, so as to provide a basis for subsequent time series prediction.

[0032] Next, a full-link causal mapping relationship for different nursing behaviors is constructed. Specifically, by establishing correlation factors among basic information, time-series dynamic data of nursing behaviors, data on the entire life cycle of consumables, and data on medical insurance programs, a full-link causal mapping relationship based on different nursing behaviors is constructed.

[0033] The following steps, S102, S103, and S104, quantify the data and correlation degree of related factors: S102, based on the time-series dynamic data of nursing behavior and the full life cycle circulation data of consumables, determine the consumption driving intensity of consumables consumption during nursing behavior.

[0034] The intensity of a single nursing behavior's driving force on the consumption of specific consumables is quantified by integrating three dimensions: frequency, specificity, and causality, thereby eliminating spurious interference.

[0035] Specifically, based on the time-series dynamic data of nursing behaviors and the full life-cycle circulation data of consumables, the ratio of the number of cases using any consumable when performing any nursing behavior to the total number of cases performing the corresponding nursing behavior, the specificity of any consumable to any nursing behavior, and the Granger causal strength of any nursing behavior on the consumption of any consumable are determined. The consumption drive strength of nursing behaviors on consumable consumption is determined by the ratio of the number of cases using any consumable when performing any nursing behavior to the total number of cases performing the corresponding nursing behavior, the specificity of any consumable to any nursing behavior, and the Granger causal strength of any nursing behavior on the consumption of any consumable.

[0036] The following formula is used to obtain the result: in, To perform any nursing action Use any consumable The number of cases and the implementation of corresponding nursing care The ratio of the total number of cases, For any consumable For any nursing action Exclusivity For any nursing action For any consumable The Granger causality strength consumed, For L2 normalization term, For target consumables, The intensity of the consumption drive for consumables by nursing behaviors.

[0037] The value was obtained through Granger causality test and ranges from [0,1]. The specific calculation process is as follows: 1) Constructing a time sequence of nursing behaviors Consumable consumption time series Determine the optimal lag order .

[0038] 2) Establish an unconstrained regression model with lags and a constrained regression model without lags, and calculate the first sum of squared residuals for the unconstrained regression model. And the second residual sum of squares of the constrained regression model .

[0039] 3) Calculate the F-statistic and obtain the p-value for the Granger causality test.

[0040] 4) Through Normalization makes The value ranges from [0,1], and the larger the value, the stronger the Granger causality.

[0041] Among them, the nursing behavior time sequence Specifically, this refers to the frequency and intensity of nursing actions performed over time. Consumable consumption time sequence. Specifically, it refers to the amount of consumables used over time.

[0042] The formula for the unconstrained regression (including lagged nursing behaviors) is as follows: in, For the intercept term, These are the autoregressive coefficients. The core coefficient for Granger causality test. For the first random error term, For the current number of the previous ones Consumable usage during the period For the current number of the previous ones The frequency and intensity of nursing care during the period.

[0043] Next, the first sum of squared residuals of the non-regression model can be calculated. .

[0044] The formula for the constrained regression (excluding delayed nursing behaviors) is as follows: in, This is the second random error term.

[0045] Next, the second sum of squared residuals of the regression model can be calculated. .

[0046] Based on the first residual sum of squares and the second residual sum of squares Calculate the F-statistic: in, This represents the sample size of the time series.

[0047] Next, based on the F-statistic and degrees of freedom (k, T-2k-1), the p-value of the Granger causality test is obtained.

[0048] Finally, normalization yields , specifically: exist The smaller the size, the stronger the causal relationship. The closer to 1; The larger the value, the weaker the causal relationship. The closer it is to 0.

[0049] denominator For L2 normalization term, For the target consumable, the consumption driving intensity of nursing behavior on consumable consumption can be obtained by following the above calculations.

[0050] S103, based on the entire life cycle circulation data of consumables and medical insurance project data, determine the correlation between consumable consumption and the profit and loss of medical insurance projects.

[0051] Specifically, quantifying the correlation between the use of a single consumable and the profit and loss of a specific medical insurance program directly corresponds to the payment rules of the medical insurance program.

[0052] This includes: based on the entire lifecycle circulation data of consumables and medical insurance project data, determining the average cost of consumables per case for any consumable used in any disease group in the medical insurance project data, the benchmark average cost of consumables per case for any disease group in the medical insurance project data, the utilization rate of any consumable in any disease group in the medical insurance project data, the overspending rate of any consumable used in any disease group in the medical insurance project data, and the standard deviation of the average cost of consumables per case for any disease group in the medical insurance project data; Based on the average cost of consumables per case for any disease group using any consumable in the medical insurance program data, the benchmark average cost of consumables per case for any disease group in the medical insurance program data, the utilization rate of any consumable in any disease group in the medical insurance program data, the overspending rate of any case using any consumable in any disease group in the medical insurance program data, and the standard deviation of the average cost of consumables per case for any disease group in the medical insurance program, the correlation between consumable consumption and the profit and loss of the medical insurance program is determined.

[0053] Specifically, it is obtained using the following formula: in, For any disease group in the medical insurance program data Use any consumable The average cost of consumables per case, For any disease group in the medical insurance program The corresponding medical insurance benchmark cost per consumable item, For any disease group in medical insurance data Any consumable in the corresponding category Utilization rate For any disease group in the medical insurance program Use any consumable The case overspending rate, The standard deviation of the average cost of consumables per case in any disease group within the medical insurance program. The correlation between consumable consumption and the profitability of medical insurance programs.

[0054] Next, S104 is executed to determine the patient's dependence on nursing behaviors based on the patient's basic information and the time-series dynamic data of nursing behaviors.

[0055] Specifically, based on the patient's basic information and the time-series dynamic data of nursing behaviors, the first degree of fit of the patient to perform any nursing behavior in the disease group of the medical insurance program is determined, the second degree of fit of the patient's surgical level, nursing level and nursing behavior is determined, and the third degree of fit of the patient's complications and nursing behavior is determined. Based on the first fit, second fit, and third fit, the intensity of the patient's dependence on the nursing behavior is determined.

[0056] Specifically, it is obtained using the following formula: in, For first-degree fit, For the second fit, For the third degree of fit, The weight corresponding to the first fitness score. The weight corresponding to the second fitness. The weight corresponding to the third fitness. , as well as Determined using the entropy weight method, The degree of dependence of patients on nursing care.

[0057] The first fit is the percentage of cases in which the patient receives any nursing care within the disease group covered by medical insurance. The second fit is obtained as follows: Table 1: Surgical Grade (G): Quantified into 4 levels.

[0058]

[0059] Table 2: Nursing Level (Q): Quantified into 4 levels.

[0060]

[0061] Table 3: Nursing behaviors (B): Assigned according to level.

[0062] First, nursing behaviors are divided into four main categories: basic nursing, specialized nursing, patient monitoring, and safety management. Each category is further subdivided into specific items, and a standard weight W is assigned according to the nursing level, with higher weights for special-grade and first-grade nursing. Table 3 is as follows:

[0063] The formula for calculating the second fitness degree is as follows: The actual execution weight of any nursing action, The importance coefficient for any nursing action. The standard weight corresponding to any nursing level, Surgical level With nursing level The matching coefficient.

[0064] The formula for calculating the third fitness degree is as follows: Calculate using the entropy weight method , as well as This avoids the subjectivity of manual assignment. The patient's dependence on nursing care is calculated using the above formula.

[0065] Next, S105 is executed, which constructs a full-link causal mapping relationship for different nursing behaviors based on the patient's basic information, the time-series dynamic data of nursing behaviors, the full life cycle circulation data of consumables, medical insurance project data, consumption driving intensity, correlation and dependence intensity.

[0066] Specifically, based on the patient's basic information, the time-series dynamic data of nursing behavior, the whole life cycle circulation data of consumables, and the medical insurance project data, the entities, entity attributes, and time-series relationships at each level are determined; Based on entities at each level, entity attributes, temporal relationships, consumption driving strength, correlation degree, and dependency strength, the causal relationships between entities are determined. Based on the causal relationships between entities, a full-link causal mapping relationship is constructed for different nursing behaviors.

[0067] First, Table 4 shows the entities and their attributes at each level as follows:

[0068] After identifying the entities and their attributes at each level, temporal relationships are established. This involves creating temporal relationships between entities based on their time frame, such as the order of nursing actions and the timing of consumable usage. Simultaneously, causal relationships are determined based on the intensity of consumption drive, correlation, and dependence between entities. These include the three main categories of causal relationships mentioned above: the intensity of nursing action's consumption drive on consumables, the correlation between consumable consumption and the profitability of medical insurance programs, and the patient's dependence on nursing actions. The quantitative results of these causal relationships are then used to quantify the relationships between entities, thus forming a full-chain causal mapping relationship for different nursing actions.

[0069] Next, execute S106 to construct a consumable demand forecasting model based on the endogenous constraints of the medical insurance program's profit and loss.

[0070] A multi-scale attention spatiotemporal prediction network is constructed, specifically by embedding the endogenous constraints of medical insurance project profit and loss into the model loss function, so as to achieve the simultaneous completion of consumable demand prediction and cost control, and solve the cold start problem of small sample disease groups.

[0071] First, an end-to-end multi-branch feature fusion learning network is constructed, which mainly includes four modules: input layer, feature encoding layer, multi-scale fusion attention layer, and dual-output head output layer.

[0072] The input layer has three parallel feature input branches: static feature branch, temporal feature branch, and graph feature branch. The static feature branch is used to input the basic attribute features of patients, disease groups, and consumables. The temporal feature branch is used to input the multi-scale temporal sequence features of nursing behavior and consumable consumption. The graph feature branch is used to input the correlation features and graph embedding vectors extracted from the temporal causal knowledge graph.

[0073] The feature encoding layer is used to perform differentiated encoding for features of different branches. For example, for the static feature branch, it is encoded into a 64-dimensional dense feature vector through two fully connected layers; for the temporal feature branch, it extracts short / medium / long-term nursing behavior temporal features of 3 days / 7 days / 14 days respectively through a multi-scale temporal convolutional network to capture the consumption patterns at different time scales; for the graph feature branch, it is encoded through a graph attention network to output a 128-dimensional graph embedding vector to capture the causal relationship features between entities.

[0074] This multi-scale fusion attention layer employs a multi-head attention mechanism to automatically learn the contribution weights of different features to consumable demand prediction, focusing on nursing operation features with high causal correlation and consumable features with high profit and loss impact, thereby achieving adaptive fusion of multi-dimensional features.

[0075] The dual-output head output layer has two parallel output heads: a consumable demand forecast head and a cost-profit forecast head. The consumable demand forecast head outputs the predicted demand for various consumables throughout the patient's surgical cycle / the next 7 days / the next 14 days, the usage time points, and the predicted costs. The cost-profit forecast head outputs the predicted profit-loss rate and the probability of overspending for the corresponding medical insurance project group under the consumable plan.

[0076] The network model architecture described above still needs to be improved through continuous training with historical data.

[0077] Most importantly, the profit and loss constraints of medical insurance programs need to be embedded into the consumable demand forecasting model, so that the total loss function of the consumable demand forecasting model is in the following form: in, To smooth out L1 loss, , This represents the actual amount of consumables used. To predict the usage of consumables, For the types and quantities of consumables, The loss is constrained by business rules. Specifically, when the profit and loss constraints of medical insurance programs are embedded in the consumable demand forecasting model, a strong penalty is generated when the predicted cost of consumables exceeds the disease group cost ceiling of the medical insurance program. , The total number of cases in the training sample. For cases Predicted consumable costs, , The unit price of any consumable. For cases any consumable Predicted usage of consumables For cases The upper limit of consumable costs for the disease group under the relevant medical insurance program. , The payment standard for disease groups in the medical insurance program. The percentage of the baseline consumable cost for the patient group. For the safety margin coefficient, For L2 regularization, , For all weights of the consumables demand forecasting model, For any weight in the consumables demand forecasting model, To smooth the weights corresponding to L1 loss, To predict the weight of the strong penalty term that occurs when the cost of consumables exceeds the disease group cost limit of the medical insurance program, These are the weights corresponding to L2 regularization; A consumable demand forecasting model based on the endogenous constraints of profit and loss of medical insurance projects was constructed.

[0078] In a specific implementation, a network search method is used to determine the optimal values ​​of each weight, with the default weight value being... , , The adjustments will be made dynamically based on the hospital's cost control needs.

[0079] Next, after obtaining the consumable demand prediction model, it was optimized: the AdamW optimization algorithm was adopted, with an initial learning rate of 0.0001, a cosine annealing learning rate strategy, a maximum training epoch of 2000 epochs, and an early stopping strategy. Training was stopped when the validation set loss did not decrease for 50 consecutive epochs to avoid overfitting. For small-sample scenarios involving newly added disease groups, new surgical procedures, and rare disease groups, a transfer learning scheme was adopted. The parameters of the feature extraction layer trained on common disease groups were transferred to the small-sample model, and only the parameters of the fully connected layer and the output layer were fine-tuned to solve the problem of poor prediction accuracy in small-sample scenarios.

[0080] Finally, evaluation indicators are set, including prediction accuracy and cost control indicators. For example, the average prediction error of the test set should be ≤2%, and the coverage rate of consumable demand prediction should be ≥99%. The disease group cost control compliance rate of medical insurance projects should be ≥95%, and the overspending early warning accuracy rate should be ≥90%.

[0081] Next is the application. Execute S107, which uses the full-link causal mapping relationship for different nursing behaviors and the consumable demand prediction model based on the profit and loss constraints of medical insurance projects to predict the consumable demand for individual characteristics of the target patient and the profit and loss situation of medical insurance projects, and obtain the prediction results.

[0082] The entire causal mapping relationship of different nursing behaviors corresponding to the target patient is input into the consumable demand prediction model based on the profit and loss constraints of medical insurance projects, thereby obtaining the consumable demand of the target individual characteristics and the profit and loss situation of medical insurance projects.

[0083] After obtaining the prediction results, it also includes: intelligent management and risk warning of the entire process of consumables.

[0084] First, when the deviation rate between the predicted result and the actual usage of consumables exceeds the preset value, the source of the deviation is traced to locate the source of the deviation, and the input features are incrementally updated to re-predict.

[0085] The formula for calculating the deviation rate is as follows: in, This represents the actual amount of consumables used. To predict the usage of consumables, To predict the deviation rate between the usage of consumables and the actual usage of consumables.

[0086] Second, when the deviation rate between the predicted results and the actual usage of consumables does not exceed a preset value, a dynamic safety stock optimization model is constructed, allowing the safety stock level to be dynamically adjusted according to predicted demand, surgical plans, and supply cycles. The formula for calculating the optimal safety stock level for a single consumable is as follows: in, For consumables at all times The optimal safety stock level, For consumables The average daily forecast demand at any given time. To allow for the lead time of consumable procurement, For safety factors, the level of importance of consumables is determined. The standard deviation of the daily demand for consumables. The formula for calculating the coverage of replacement consumables is as follows: in, It is a collection of alternative consumables for consumables. To replace consumables With original consumables Functional similarity, To replace the inventory of consumables, This refers to the inventory of original consumables.

[0087] The DRG surgical plan fluctuation factor, i.e. , For period DRG surgical plan volume within the scope, This represents the average number of surgeries performed during the same period in history.

[0088] Next, based on this optimal safety stock level, a tiered replenishment strategy is set up: Emergency replenishment: Based on existing inventory levels In such cases, emergency replenishment is required, with the replenishment quantity sufficient to meet the projected demand for the next 7 days.

[0089] Regular replenishment, based on existing inventory levels Regular restocking is required at this time.

[0090] Overstock warning, based on existing inventory levels In such cases, it is necessary to issue an inventory backlog warning, suspend the procurement of consumables, and prioritize clinical use.

[0091] Next, it can intelligently recommend alternative consumables, optimizing alternative consumables through multi-objective optimization, as calculated in the following formula: For recommended alternative consumables, in response to cases raw materials Recommended rating, To replace consumables With original consumables Functional similarity, To achieve cost savings, To replace the unit price of consumables, This refers to the unit price of any consumable, specifically the original consumable. To replace consumables With Case Clinical fit, which is determined by case studies. Alternative consumables used in the DRG group The weighted value of the percentage of cases and the rate of treatment effectiveness achievement. The weighting of the functional similarity between the replacement consumable and the original consumable is used to determine the weighting. As a weight for cost savings rate, To replace consumables With Case The weight of clinical fit.

[0092] It can improve in high-overspending risk scenarios. Prioritize cost control; improve efficiency in low-risk scenarios. , Prioritize clinical safety and suitability.

[0093] Based on the above recommendation scores, alternative consumable replacement plans can be generated and simultaneously pushed to the clinical nursing and procurement ends, realizing the linkage between clinical use and supply chain management.

[0094] Next, a dynamic overspending risk scoring model will be constructed to achieve accurate graded early warning.

[0095] The formula for the dynamic overspending risk scoring is as follows: For cases At any moment Overspending risk score, For this case At the deadline The cumulative cost of consumables already incurred, For this case The upper limit of consumable costs for the DRG disease group For this case The number of hospitalized beds has reached [number] days. This represents the average length of hospital stay for this disease group. For this case The probability of final overspending, For this case At the deadline The cumulative cost of consumables already incurred and the case The weight of the ratio of the upper limit of consumable costs to the DRG disease group. For this case The weight of the ratio of the number of hospitalized bed days to the average length of stay for that disease group. For this case The weight of the final overspending probability.

[0096] The resulting overspending risk score establishes a three-level early warning mechanism: Low risk: Normal monitoring, no early warning notifications.

[0097] Medium risk: It provides departmental-level early warnings and pushes detailed information on consumable usage, cost progress, and optimization to clinical departments.

[0098] High risk: It provides dual early warnings at the hospital and department levels, and simultaneously pushes out information on the reasons for overspending, source analysis, recommended alternative consumables, and cost control pathways.

[0099] One or more technical solutions in the embodiments of the present invention have at least the following technical effects or advantages: This invention provides a method for consumables prediction and medical cost control, comprising: acquiring basic patient information, time-series dynamic data of nursing behaviors, full life-cycle circulation data of consumables, and medical insurance project data; determining the consumption driving intensity of consumables consumption during nursing behaviors based on the time-series dynamic data of nursing behaviors and the full life-cycle circulation data of consumables; determining the correlation between consumables consumption and the profit and loss of medical insurance projects based on the full life-cycle circulation data of consumables and the medical insurance project data; determining the patient's dependence on nursing behaviors based on the patient's basic information and time-series dynamic data of nursing behaviors; constructing a full-link causal mapping relationship for different nursing behaviors based on the patient's basic information, time-series dynamic data of nursing behaviors, full life-cycle circulation data of consumables, medical insurance project data, consumption driving intensity, correlation, and dependence intensity; constructing a consumables demand prediction model based on the endogenous constraint of the profit and loss of medical insurance projects; and predicting the consumables demand and the profit and loss of medical insurance projects for the individual characteristics of target patients based on the full-link causal mapping relationship for different nursing behaviors and the consumables demand prediction model based on the endogenous constraint of the profit and loss of medical insurance projects, so as to improve the efficiency of medical services and the operational efficiency of hospitals.

[0100] Example 2: Based on the same inventive concept, embodiments of the present invention provide a device for consumable prediction and medical cost control, such as... Figure 2 As shown, it includes: The acquisition module 201 is used to acquire patients' basic information, dynamic data of nursing behavior over time, data on the entire life cycle of consumables, and data on medical insurance items. The first determining module 202 is used to determine the consumption driving intensity of nursing behavior on consumable consumption based on the time-series dynamic data of nursing behavior and the full life cycle circulation data of consumables. The second determining module 203 is used to determine the correlation between consumable consumption and the profit and loss of medical insurance projects based on the consumable's full life cycle circulation data and the medical insurance project data. The third determining module 204 is used to determine the patient's dependence on the nursing behavior based on the patient's basic information and the time-series dynamic data of the nursing behavior; The first construction module 205 is used to construct a full-link causal mapping relationship for different nursing behaviors based on the patient's basic information, dynamic data of nursing behavior time sequence, data of the entire life cycle of consumables, data of medical insurance projects, the consumption driving strength, correlation and dependence strength. The second construction module 206 is used to construct a consumable demand forecasting model based on the endogenous constraints of profit and loss of medical insurance projects. The prediction module 207 is used to predict the consumable demand of the target patient and the profit and loss of the medical insurance program based on the full-link causal mapping relationship for different nursing behaviors and the consumable demand prediction model based on the endogenous constraint of the profit and loss of the medical insurance program, and obtain the prediction results.

[0101] In one optional implementation, the first determining module 202 is configured to: Based on the time-series dynamic data of the nursing behaviors and the full life-cycle circulation data of the consumables, the ratio of the number of cases using any consumable when performing any nursing behavior to the total number of cases performing the corresponding nursing behavior, the specificity of any consumable to any nursing behavior, and the Granger causality strength of any nursing behavior on the consumption of any consumable are determined. The consumption drive strength of nursing actions on consumable consumption is determined by the ratio of the number of cases using any consumable during any nursing action to the total number of cases performing the corresponding nursing action, the specificity of any consumable to any nursing action, and the Granger causal strength of any nursing action on the consumption of any consumable. This is calculated using the following formula: in, To perform any nursing action Use any consumable The number of cases and the implementation of corresponding nursing care The ratio of the total number of cases, For any consumable For any nursing action Exclusivity For any nursing action For any consumable The Granger causality strength consumed, For L2 normalization term, For target consumables, The intensity of the consumption drive for consumables by nursing behaviors.

[0102] In one alternative implementation, the second determining module 203 is configured to: Based on the full life cycle circulation data of the consumables and the medical insurance project data, the average cost of consumables per case for any disease group in the medical insurance project data, the benchmark average cost of consumables per case for any disease group in the medical insurance project data, the utilization rate of any consumables in any disease group in the medical insurance project data, the overspending rate of any case for any consumables in any disease group in the medical insurance project data, and the standard deviation of the average cost of consumables per case for any disease group in the medical insurance project data are determined. Based on the average cost per case of any consumable used in any disease group within the medical insurance program data, the benchmark average cost per case of consumables for any disease group within the medical insurance program data, the utilization rate of any consumable in any disease group within the medical insurance program data, the overspending rate of any consumable used in any disease group within the medical insurance program data, and the standard deviation of the average cost per case of consumables for any disease group within the medical insurance program data, the correlation between consumable consumption and the profit and loss of the medical insurance program is determined using the following formula: in, For any disease group in the medical insurance program data Use any consumable The average cost of consumables per case, For any disease group in the medical insurance program data The corresponding medical insurance benchmark cost per consumable item, For any disease group in the medical insurance program data Any consumable in the corresponding category Utilization rate For any disease group in the medical insurance program data Use any consumable The case overspending rate, This represents the standard deviation of the average cost of consumables per case in any disease group within the medical insurance program data. The correlation between consumable consumption and the profitability of medical insurance programs.

[0103] In one alternative implementation, the third determining module 204 is configured to: Based on the patient's basic information and the time-series dynamic data of the nursing behaviors, the first degree of fit of the patient to perform any nursing behavior in the disease group of the medical insurance program is determined, the second degree of fit of the patient's surgical level, nursing level and nursing behavior, and the third degree of fit of the patient's complications and nursing behavior. Based on the first fit, the second fit, and the third fit, the patient's dependence on the nursing behavior is determined using the following calculation formula: in, For the first fit, For the second degree of fit, For the third degree of fit, The weight corresponding to the first fit. The weight corresponding to the second fit. The weight corresponding to the third fitness level. , as well as Determined using the entropy weight method, The degree of dependence of the patient on the aforementioned nursing care.

[0104] In one alternative implementation, the first building module 205 is configured to: Based on the patient's basic information, the time-series dynamic data of nursing behavior, the whole life cycle circulation data of consumables, and the medical insurance project data, the entities, entity attributes, and time-series relationships at each level are determined. Based on the entities, entity attributes, temporal relationships, consumption driving strength, correlation degree, and dependency strength at each level, the causal relationships between entities are determined. Based on the causal relationships between the entities, a full-link causal mapping relationship is constructed for different nursing behaviors.

[0105] In one alternative implementation, the second building module 206 is configured to: By embedding the profit and loss constraints of medical insurance programs into the consumables demand forecasting model, the total loss function of the consumables demand forecasting model takes the following form: in, To smooth out L1 loss, , This represents the actual amount of consumables used. To predict the usage of consumables, For the types and quantities of consumables, The loss is constrained by business rules. Specifically, when the profit and loss constraints of medical insurance programs are embedded in the consumable demand forecasting model, a strong penalty is generated when the predicted cost of consumables exceeds the disease group cost ceiling of the medical insurance program. , The total number of cases in the training sample. For cases Predicted consumable costs, , The unit price of any consumable. For cases any consumable Predicted usage of consumables For cases The upper limit of consumable costs for the disease group under the relevant medical insurance program. , The payment standard for disease groups in the medical insurance program, The percentage of the baseline consumable cost for the patient group. For the safety margin coefficient, For L2 regularization, , For all weights of the consumables demand forecasting model, For any weight in the consumables demand forecasting model, To smooth the weights corresponding to L1 loss, To predict the weight of the strong penalty term that occurs when the cost of consumables exceeds the disease group cost limit of the medical insurance program, These are the weights corresponding to L2 regularization; A consumable demand forecasting model based on the endogenous constraints of profit and loss of medical insurance projects was constructed.

[0106] In one alternative implementation, it further includes: a safety stock adjustment module, used for: Based on the prediction results, a dynamic safety stock optimization model is constructed to dynamically adjust the safety stock level, where the safety stock level is the inventory level of consumables.

[0107] In one alternative implementation, it further includes: a recommendation module, used for: Based on the aforementioned safety stock level, alternative consumables are recommended.

[0108] Example 3: Based on the same inventive concept, embodiments of the present invention provide a computer device, such as... Figure 3As shown, it includes a memory 304, a processor 302, and a computer program stored in the memory 304 and executable on the processor 302. When the processor 302 executes the program, it implements the steps of the above-mentioned consumables prediction and medical cost control method.

[0109] Among them, Figure 3 In this document, a bus architecture (represented by bus 300) is used. Bus 300 may include any number of interconnected buses and bridges, linking various circuits including one or more processors represented by processor 302 and memory represented by memory 304. Bus 300 may also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. Bus interface 306 provides an interface between bus 300 and receiver 301 and transmitter 303. Receiver 301 and transmitter 303 may be the same element, i.e., a transceiver, providing a unit for communicating with various other devices over a transmission medium. Processor 302 is responsible for managing bus 300 and general processing, while memory 304 can be used to store data used by processor 302 during operation.

[0110] Example 4: Based on the same inventive concept, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method for consumables prediction and medical cost control.

[0111] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, this invention is not directed to any particular programming language. It should be understood that the contents of the invention described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of the invention.

[0112] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0113] Similarly, it should be understood that, in order to simplify the invention and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof. However, this disclosure should not be construed as reflecting an intention that the claimed invention requires more features than are explicitly recited in each embodiment. Rather, as reflected in each embodiment, inventive aspects lie in fewer than all features of the single foregoing disclosed embodiment. Therefore, the claims, following the detailed description, are hereby expressly incorporated into this detailed description, wherein each claim itself is a separate embodiment of the invention.

[0114] Those skilled in the art will understand that modules in the device of the embodiments can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or device so disclosed. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0115] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the specific implementation, any of the claimed embodiments can be used in any combination.

[0116] The various component embodiments of the present invention can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some or all of the components in the consumable prediction and medical cost control device or computer equipment according to embodiments of the present invention. The present invention can also be implemented as a device or apparatus program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such programs implementing the present invention can be stored on a computer-readable medium or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0117] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

Claims

1. A method for consumables prediction and medical cost control, characterized in that, include: Acquire basic patient information, dynamic data on nursing behavior over time, data on the entire lifecycle of consumables, and data on medical insurance programs; Based on the time-series dynamic data of nursing behaviors and the full life cycle circulation data of consumables, the consumption driving intensity of nursing behaviors on consumable consumption is determined. Based on the full life cycle circulation data of the consumables and the medical insurance program data, the correlation between consumable consumption and the profit and loss of medical insurance programs is determined. Based on the patient's basic information and the time-series dynamic data of the nursing behaviors, the patient's dependence on the nursing behaviors is determined; Based on the patient's basic information, time-series dynamic data of nursing behaviors, data on the entire life cycle of consumables, data on medical insurance projects, the intensity of consumption drive, correlation and dependence, a full-link causal mapping relationship is constructed for different nursing behaviors; Construct a consumables demand forecasting model based on the endogenous constraints of profit and loss of medical insurance programs; Based on the full-link causal mapping relationship for different nursing behaviors and the consumable demand prediction model based on the endogenous constraint of medical insurance project profit and loss, the consumable demand for target patients and the profit and loss situation of medical insurance projects are predicted, and the prediction results are obtained.

2. The method as described in claim 1, characterized in that, Based on the time-series dynamic data of nursing behaviors and the full life-cycle circulation data of consumables, the consumption driving intensity of nursing behaviors on consumable consumption is determined, including: Based on the time-series dynamic data of the nursing behaviors and the full life-cycle circulation data of the consumables, the ratio of the number of cases using any consumable when performing any nursing behavior to the total number of cases performing the corresponding nursing behavior, the specificity of any consumable to any nursing behavior, and the Granger causality strength of any nursing behavior on the consumption of any consumable are determined. The consumption drive strength of nursing actions on consumable consumption is determined by the ratio of the number of cases using any consumable during any nursing action to the total number of cases performing the corresponding nursing action, the specificity of any consumable to any nursing action, and the Granger causal strength of any nursing action on the consumption of any consumable. This is calculated using the following formula: ； in, To perform any nursing action Use any consumable The number of cases and the implementation of corresponding nursing care The ratio of the total number of cases, For any consumable For any nursing action Exclusivity For any nursing action For any consumable The Granger causality strength consumed, For L2 normalization term, For target consumables, The intensity of the consumption drive for consumables by nursing behaviors.

3. The method as described in claim 1, characterized in that, Based on the entire lifecycle circulation data of the consumables and the medical insurance program data, the correlation between consumable consumption and the profit and loss of medical insurance programs is determined, including: Based on the full life cycle circulation data of the consumables and the medical insurance project data, the average cost of consumables per case for any disease group in the medical insurance project data, the benchmark average cost of consumables per case for any disease group in the medical insurance project data, the utilization rate of any consumables in any disease group in the medical insurance project data, the overspending rate of any case using any consumables in any disease group in the medical insurance project data, and the standard deviation of the average cost of consumables per case for any disease group in the medical insurance project data are determined. Based on the average cost per case of any consumable used in any disease group within the medical insurance program data, the benchmark average cost per case of consumables for any disease group within the medical insurance program data, the utilization rate of any consumable in any disease group within the medical insurance program data, the overspending rate of any consumable used in any disease group within the medical insurance program data, and the standard deviation of the average cost per case of consumables for any disease group within the medical insurance program data, the correlation between consumable consumption and the profit and loss of the medical insurance program is determined using the following formula: ； in, For any disease group in the medical insurance program data Use any consumable The average cost of consumables per case, For any disease group in the medical insurance program data The corresponding medical insurance benchmark cost per consumable item For any disease group in the medical insurance program data Any consumable in the corresponding category Utilization rate For any disease group in the medical insurance program data Use any consumable The case overspending rate This represents the standard deviation of the average cost of consumables per case in any disease group within the medical insurance program data. The correlation between consumable consumption and the profitability of medical insurance programs.

4. The method as described in claim 1, characterized in that, Based on the patient's basic information and the time-series dynamic data of the nursing behaviors, the patient's dependence on the nursing behaviors is determined, including: Based on the patient's basic information and the time-series dynamic data of the nursing behaviors, the first degree of fit of the patient to perform any nursing behavior in the disease group of the medical insurance program is determined, the second degree of fit of the patient's surgical level, nursing level and nursing behavior, and the third degree of fit of the patient's complications and nursing behavior. Based on the first fit, the second fit, and the third fit, the patient's dependence on the nursing care behavior is determined using the following calculation formula: ； in, For the first fit, For the second degree of fit, For the third degree of fit, The weight corresponding to the first fit. The weight corresponding to the second fit. The weight corresponding to the third fitness. , as well as Determined using the entropy weight method, The degree of dependence of the patient on the aforementioned nursing care.

5. The method as described in claim 1, characterized in that, Based on patient basic information, time-series dynamic data of nursing behaviors, full life-cycle circulation data of consumables, medical insurance project data, and the strength of consumption drivers, correlations, and dependencies, a full-link causal mapping relationship is constructed for different nursing behaviors, including: Based on the patient's basic information, the time-series dynamic data of nursing behavior, the whole life cycle circulation data of consumables, and the medical insurance project data, the entities, entity attributes, and time-series relationships at each level are determined. Based on the entities, entity attributes, temporal relationships, consumption driving strength, correlation degree, and dependency strength at each level, the causal relationships between entities are determined. Based on the causal relationships between the entities, a full-link causal mapping relationship is constructed for different nursing behaviors.

6. The method as described in claim 1, characterized in that, Construct a demand forecasting model for consumables based on the endogenous constraints of the profit and loss of medical insurance programs, including: By embedding the profit and loss constraints of medical insurance programs into the consumables demand forecasting model, the total loss function of the consumables demand forecasting model takes the following form: ； in, To smooth out L1 loss, , This represents the actual amount of consumables used. To predict the usage of consumables, For the types and quantities of consumables, The loss is constrained by business rules. Specifically, when the profit and loss constraints of medical insurance programs are embedded in the consumable demand forecasting model, a strong penalty is generated when the predicted cost of consumables exceeds the disease group cost ceiling of the medical insurance program. , The total number of cases in the training sample. For cases Predicted consumable costs, , The unit price of any consumable. For cases any consumable Predicted usage of consumables For cases The upper limit of consumable costs for the disease group under the relevant medical insurance program. , The payment standard for disease groups in the medical insurance program, The percentage of the baseline consumable cost for the patient group. For the safety margin coefficient, For L2 regularization, , For all weights of the consumables demand forecasting model, For any weight in the consumables demand forecasting model, To smooth the weights corresponding to L1 loss, To predict the weight of the strong penalty term that occurs when the cost of consumables exceeds the disease group cost limit of the medical insurance program, These are the weights corresponding to L2 regularization; A consumable demand forecasting model based on the endogenous constraints of profit and loss of medical insurance projects was constructed.

7. The method as described in claim 1, characterized in that, Based on the full-link causal mapping relationship for different nursing behaviors and the consumable demand prediction model based on the endogenous constraint of medical insurance program profitability, the consumable demand for individual target patients and the profitability of medical insurance programs are predicted. After obtaining the prediction results, the method further includes: Based on the prediction results, a dynamic safety stock optimization model is constructed to dynamically adjust the safety stock level, where the safety stock level is the inventory level of consumables.

8. The method as described in claim 7, characterized in that, Based on the prediction of consumable demand and the profitability of medical insurance programs according to the individual characteristics of the target patients, a dynamic safety stock optimization model is constructed to dynamically adjust the safety stock level. After the safety stock level is the inventory level of consumables, the model further includes: Based on the aforementioned safety stock level, alternative consumables are recommended.

9. A device for consumables prediction and medical cost control, characterized in that, include: The acquisition module is used to acquire patients' basic information, dynamic data of nursing behavior over time, data on the entire life cycle of consumables, and data on medical insurance items. The first determining module is used to determine the intensity of the consumption drive of nursing behavior on consumable consumption based on the time-series dynamic data of nursing behavior and the full life cycle circulation data of consumables. The second determining module is used to determine the correlation between consumable consumption and the profit and loss of medical insurance projects based on the full life cycle circulation data of the consumables and the medical insurance project data. The third determining module is used to determine the patient's dependence on the nursing behavior based on the patient's basic information and the time-series dynamic data of the nursing behavior. The first construction module is used to construct a full-link causal mapping relationship for different nursing behaviors based on the patient's basic information, dynamic data of nursing behavior time sequence, data on the entire life cycle of consumables, data on medical insurance projects, the consumption driving strength, correlation and dependence strength. The second building module is used to build a consumable demand forecasting model based on the endogenous constraints of profit and loss of medical insurance projects. The prediction module is used to predict the consumable demand of the target patient and the profit and loss of the medical insurance program based on the full-link causal mapping relationship for different nursing behaviors and the consumable demand prediction model based on the endogenous constraint of the profit and loss of the medical insurance program, and obtain the prediction results.

10. A computer 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 program, it implements the method as described in any one of claims 1 to 8.

Images