Hospital comprehensive budget lean management system and method based on artificial intelligence

Abstract

The invention discloses a hospital comprehensive budget lean management system and method based on artificial intelligence. The system comprises a data acquisition module, a budgeting module, an execution monitoring module, a dynamic adjustment module, a risk early warning module, a resource allocation module and a visual interface. According to the hospital comprehensive budget lean management system and method based on artificial intelligence, the scientificity, the accuracy, the real-time performance and the effectiveness of hospital budget management are comprehensively improved through an advanced technical means and a scientific management method, resource allocation optimization, operation cost reduction and service quality improvement of a hospital are facilitated, and the hospital budget management system and method are suitable for popularization and application. The core competitiveness of hospitals is enhanced, and remarkable economic benefits and social benefits are achieved.

Description

Technical Field

[0001] This invention relates to the field of hospital economic management technology, and in particular to a lean management system and method for comprehensive hospital budgeting based on artificial intelligence. Background Technology

[0002] Currently, many hospitals in China still use traditional manual or semi-automated management models for economic management, which presents the following problems:

[0003] The budget preparation lacks scientific rigor and fails to meet the hospital's internal operational and management needs;

[0004] Data sources are scattered, and information silos are serious, leading to a disconnect between budget execution and actual operations;

[0005] The lack of intelligent analytical tools makes it impossible to monitor and dynamically adjust budget execution in real time.

[0006] Traditional management models are ill-suited to policy requirements such as "full coverage of DRG / DIP payment methods" and "new medical reform policies". Summary of the Invention

[0007] In view of the above-mentioned shortcomings in the prior art, the present invention provides a lean management system and method for comprehensive hospital budget based on artificial intelligence. The purpose is to provide a lean management system and method for comprehensive hospital budget based on artificial intelligence that realizes intelligent, refined and efficient hospital budget management.

[0008] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0009] An AI-based lean budget management system for hospitals includes:

[0010] The data acquisition module is used to integrate data from the hospital's HIS system, financial system, and external market data.

[0011] The budget preparation module uses machine learning algorithms to analyze historical budget execution data and generate dynamic budget plans.

[0012] The execution monitoring module tracks the budget execution status in real time and calculates the deviation rate between actual expenditures and the budget.

[0013] The dynamic adjustment module dynamically adjusts the budget allocation based on the bias rate and external environment parameters using a reinforcement learning model.

[0014] The risk warning module allows setting multiple budget thresholds and triggering warning signals.

[0015] The process optimization module identifies redundant steps in the budget process through value stream mapping and generates optimization suggestions.

[0016] The resource allocation module combines queuing theory and game theory algorithms to optimize the allocation of medical resources.

[0017] The visual interface provides multi-dimensional data dashboards and interactive decision support.

[0018] Furthermore, the data acquisition module includes:

[0019] Structured data units that interface with the hospital's ERP system for financial and operational data;

[0020] Unstructured data units are used to parse medical equipment maintenance logs and clinical records using NLP technology;

[0021] External data units acquire medical insurance policies, drug price indices, and epidemiological data in real time.

[0022] Furthermore, the dynamic adjustment module implements budget redistribution using the following steps:

[0023] Step S1: Calculate the fluctuation of resource demand in the next 3 months based on the time series forecasting model;

[0024] Step S2: Evaluate the risk values ​​of different allocation strategies through Monte Carlo simulation;

[0025] Step S3: Generate the Pareto optimal solution set using a multi-objective optimization algorithm;

[0026] Step S4: Select the final allocation scheme based on the decision-maker's preference weight.

[0027] Furthermore, the risk warning module includes:

[0028] Threshold setting unit, set the tolerance for differences according to department / project;

[0029] The tiered early warning unit triggers a yellow alert if the deviation rate is ≥10%, an orange alert if the deviation rate is ≥15%, and a red alert if the deviation rate is ≥20%.

[0030] The response strategy library stores three types of response strategies: budget freeze, resource allocation, and emergency approval.

[0031] Furthermore, the resource allocation module employs an improved genetic algorithm, including:

[0032] Chromosome coding: Encoding hospital beds, operating rooms, and medical staff into binary gene chains;

[0033] Fitness function: An evaluation system is constructed by combining bed turnover rate, patient waiting time, and cost-effectiveness ratio;

[0034] Mutation operation: Introduce simulated annealing mechanism to prevent local optima.

[0035] Furthermore, the visualization interface includes:

[0036] 3D drill-down analysis: Supports data penetration from the hospital as a whole, departments, wards, and treatment groups.

[0037] Scenario Simulator: Predict the impact of budget adjustments on KPIs in real time by dragging and dropping parameters;

[0038] Collaborative annotation: Allows multiple departments to annotate abnormal data online and trigger a countersigning process.

[0039] An AI-based lean management approach for comprehensive hospital budgeting includes the following steps:

[0040] Step T1: Construct a multi-source heterogeneous data lake to integrate clinical, operational, and marketing data;

[0041] Step T2: Identify key influencing factors in budget preparation using the XGBoost algorithm;

[0042] Step T3: Establish a 12-month rolling budget forecasting model using an LSTM network;

[0043] Step T4: Implement closed-loop feedback control of budget-execution based on reinforcement learning;

[0044] Step T5: Generate an evaluation report that includes cost savings rate, ROI, and patient satisfaction indicators.

[0045] Furthermore, the method for extracting key influencing factors in step T2 includes:

[0046] The SHAP value was used to interpret the model output, and the variables with the highest feature importance were selected.

[0047] Eliminating multicollinearity using Pearson correlation coefficient;

[0048] Ultimately, the core factors of disease weighting, DRG payment standards, and equipment depreciation rate will be retained.

[0049] Furthermore, the closed-loop feedback control in step T4 specifically includes:

[0050] Establish the Budget Execution Index (BEI) as follows: BEI = Actual Expenditures / (Budget Baseline × Environmental Adjustment Factor);

[0051] The current policy is maintained when BEI ∈ [0.9, 1.1].

[0052] When BEI > 1.1, a cost reduction and efficiency improvement strategy is triggered;

[0053] The budget redistribution mechanism is activated when BEI < 0.9.

[0054] Furthermore, the evaluation report generation process includes:

[0055] The overall budget execution score for each department was calculated using the TOPSIS method.

[0056] Radar charts were used to compare the differences between planned and actual values ​​across six dimensions: scale, structure, schedule, quality, cost, and benefits.

[0057] We can discover hidden waste patterns and generate an improvement list based on association rules.

[0058] The beneficial effects of this invention are as follows:

[0059] This invention discloses an artificial intelligence-based lean management system and method for hospital comprehensive budgeting. Through advanced technology and scientific management methods, it comprehensively improves the scientific nature, accuracy, real-time performance, and effectiveness of hospital budget management. It helps hospitals optimize resource allocation, reduce operating costs, improve service quality, and enhance their core competitiveness, resulting in significant economic and social benefits. Attached Figure Description

[0060] Figure 1 This is a schematic diagram of the AI-based lean management system for comprehensive hospital budgeting according to the present invention.

[0061] Figure 2 This is a schematic diagram of the data acquisition module of the present invention;

[0062] Figure 3 This is a flowchart of the dynamic adjustment module of the present invention;

[0063] Figure 4 This is a diagram of the resource allocation algorithm architecture of the present invention;

[0064] Figure 5 This is a visual interface functional hierarchy diagram of the present invention;

[0065] Figure 6 This is a flowchart illustrating the entire process framework of the management method of the present invention;

[0066] Figure 7 This is a flowchart of the key factor extraction process of the present invention;

[0067] Figure 8 Generate an architecture diagram for the evaluation report of this invention. Detailed Implementation

[0068] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0069] like Figures 1 to 8 As shown, an AI-based lean budget management system for hospitals includes:

[0070] Data acquisition layer: Configured with a medical-grade edge computing gateway for real-time access to the HIS system, financial system and external data sources;

[0071] Algorithm layer: Deploy an NVIDIA DGXA100 server cluster to support distributed training of machine learning models;

[0072] Storage layer: Hadoop Distributed File System is used to store structured and unstructured data;

[0073] Application layer: A visual interface is developed based on the Spring Cloud framework, supporting access via Chrome / Firefox browsers.

[0074] Data acquisition module:

[0075] Structured data unit: Connects to the hospital ERP system via JDBC interface to synchronize financial items and operational indicators;

[0076] Unstructured data units: Deploy BERT-based NLP models to parse equipment maintenance logs and clinical records, and extract features such as equipment downtime and treatment cycles;

[0077] External data unit: It calls the National Healthcare Security Administration's API to obtain DRG payment standards and collects drug price index and epidemiological data through web crawling technology.

[0078] Budgeting module:

[0079] Historical data analysis: The XGBoost algorithm was used to model budget execution data over the past three years. Input features included departmental revenue, cost ratio, and disease structure.

[0080] Dynamic budget generation: A 12-month rolling forecast model is built based on an LSTM network. Clinical data, operational data, and market data are input, and departmental budget allocation plans are output.

[0081] Execution monitoring module:

[0082] Deviation rate calculation: Real-time collection of expenditure data from the financial system to calculate the absolute deviation rate between actual expenditure and budget. Formula: |actual - budget| / budget × 100%.

[0083] Visual tracking: The 3D drill-down analysis interface supports drilling down from the hospital level to the treatment group, displaying budget execution progress, cost composition and abnormal fluctuations.

[0084] Dynamic adjustment module:

[0085] Resource demand forecasting: Based on the Prophet time series model, input historical resource usage data and external environmental parameters to forecast resource demand for the next 3 months;

[0086] Multi-objective optimization: The NSGA-II algorithm is used to generate the Pareto front solution set, and the objective functions include cost minimization, patient satisfaction maximization, and resource utilization equilibration.

[0087] Decision-maker preferences: Weights are determined using the Analytic Hierarchy Process (AHP), such as a cost weight of 0.4 and a satisfaction weight of 0.6, and the optimal solution is selected from the solution set.

[0088] Risk warning module:

[0089] Threshold setting: Set tolerance for differences by department / project, such as ±12% for surgical departments and ±8% for internal medicine departments;

[0090] Tiered early warning: When the deviation rate is ≥10%, it will be highlighted in yellow in the visualization interface; when the deviation rate is ≥15%, an orange warning will be triggered and pushed to the department head; when the deviation rate is ≥20%, a red warning will be activated and the budget will be frozen.

[0091] Response strategy: Match response measures from the response strategy library, such as budget freeze and resource allocation, and initiate multi-departmental review through collaborative annotation function.

[0092] Process optimization module:

[0093] Value stream mapping analysis: Identifying redundant steps in the budget process using ProcessMining technology and generating process optimization suggestions, such as merging approval steps;

[0094] Evaluation of optimization effect: Monte Carlo simulation was used to compare the budget execution efficiency before and after optimization.

[0095] Resource allocation module:

[0096] Chromosome encoding: Encode hospital beds, operating rooms, and medical staff into binary gene chains. For example, the length of the hospital bed gene chain is 100 bits, where 1 indicates occupied and 0 indicates idle.

[0097] Fitness functions: Overall bed turnover rate (target value ≥ 90%), patient waiting time (target value ≤ 2 hours), and cost-effectiveness ratio (target value ≥ 1.2) are used to construct an evaluation system.

[0098] Mutation operation: A simulated annealing mechanism is introduced, which accepts inferior solutions with probability P = exp(-ΔE / T) during the mutation process to prevent getting trapped in local optima.

[0099] Visual interface:

[0100] Scenario simulator: Supports drag-and-drop adjustment of budget allocation ratios, such as increasing the operating room budget by 10%, and predicts the impact on KPIs in real time;

[0101] Collaborative annotation: Allows multiple departments to annotate abnormal data online, such as when a department's costs exceed the budget, and triggers a joint approval process, such as joint approval by the finance department, medical department, and nursing department.

[0102] Implementation process of an AI-based lean budget management method for hospitals:

[0103] Data integration:

[0104] Construct a multi-source heterogeneous data lake and use the Flink real-time computing framework to integrate clinical data, such as electronic medical records, operational data, such as financial vouchers, and market data, such as medical insurance policies;

[0105] Data cleaning: Using the Python Pandas library to handle missing values, such as mean imputation; outliers, such as removal using the 3σ principle; and duplicate values, such as deduplication based on the primary key.

[0106] Key Influence Factor Extraction:

[0107] The SHAP value was used to interpret the XGBoost model output, and the top 20% of variables in terms of feature importance were selected, such as disease weight, DRG payment standard, and equipment depreciation rate.

[0108] Multicollinearity can be eliminated using Pearson correlation coefficient, such as retaining features with a correlation coefficient <0.7;

[0109] Ultimately, the core factors are retained, and rules are established to link these factors with budget preparation, such as increasing the budget by 0.8% for every 1% increase in the disease weight.

[0110] Rolling budget forecast:

[0111] A 12-month rolling budget forecasting model was established using an LSTM network. Input features included historical budget data, clinical data such as surgical volume, operational data such as bed utilization rate, and market data such as consumable prices.

[0112] Model training: Using the Adam optimizer, the learning rate was set to 0.001, the batch size to 32, and the number of training epochs to 100.

[0113] Model evaluation: The mean absolute percentage error (MASE) index is used, requiring a prediction error of ≤5%.

[0114] Closed-loop feedback control:

[0115] Establish the Budget Execution Index (BEI) as follows: BEI = Actual Expenditure / (Budget Baseline × Environmental Adjustment Factor), where the environmental adjustment factor is dynamically adjusted based on external factors.

[0116] When BEI∈[0.9,1.1], maintain the current strategy; when BEI>1.1, trigger cost reduction and efficiency improvement strategies, such as reducing unnecessary expenditures; when BEI<0.9, initiate the budget redistribution mechanism, such as increasing the budget for high-yield departments.

[0117] Assessment report generation:

[0118] The TOPSIS method was used to calculate the overall budget execution score for each department, with input indicators including cost savings rate, ROI, and patient satisfaction.

[0119] A radar chart was used to compare the differences between planned and actual values ​​across six dimensions: scale, structure, schedule, quality, cost, and benefits.

[0120] Based on the Apriori algorithm, hidden waste patterns are identified. For example, if the waste rate of consumables in a department is ≥15%, an improvement list is generated and pushed to the department head.

[0121] This invention discloses an AI-based lean management system and method for comprehensive hospital budgeting. The data acquisition module connects to the hospital's ERP system's financial and operational data through structured data units, ensuring the accuracy and standardization of basic data. The unstructured data unit utilizes NLP technology to analyze medical equipment maintenance logs and clinical records, uncovering key information that is difficult to obtain using traditional data collection methods. The external data unit acquires real-time medical insurance policies, drug price indices, and epidemiological data, broadening the data source channels. This multi-dimensional and comprehensive data acquisition approach enables hospitals to obtain more comprehensive and accurate data, providing a solid data foundation for all subsequent stages of budget management and effectively avoiding budget deviations caused by missing or inaccurate data.

[0122] The budget preparation module leverages machine learning algorithms to deeply analyze historical budget execution data, accurately capturing the patterns and trends behind the data to generate more scientific and rational dynamic budget plans. Compared to traditional experience-based budget preparation methods, this module fully considers the actual situation of hospital operations and historical data, making budget preparation more aligned with actual needs, reducing interference from human factors, and improving the accuracy and reliability of budget preparation. Furthermore, the dynamic budget plan can be adjusted promptly according to changes in the hospital's operational status, better adapting to the dynamic needs of hospital development.

[0123] The monitoring module tracks budget execution status in real time and accurately calculates the deviation rate between actual expenditures and the budget. Through real-time monitoring, hospital management can promptly grasp the budget execution situation and identify problems arising during the process. Accurate deviation rate calculation provides strong data support for subsequent decision-making, enabling the hospital to quickly take appropriate measures based on the actual situation, avoiding budget overruns or resource waste, and improving the efficiency and effectiveness of budget execution.

[0124] The dynamic adjustment module, based on a reinforcement learning model, intelligently and dynamically adjusts budget allocation by comprehensively considering bias rate and external environmental parameters. This module can automatically optimize budget allocation schemes based on real-time data and constantly changing environmental factors, making budget allocation more aligned with the hospital's actual needs and development strategy. Compared to traditional budget adjustment methods, the application of the reinforcement learning model makes budget adjustments more scientific and reasonable, better adaptable to complex and ever-changing market environments and hospital operational conditions, and improves the flexibility and adaptability of hospital budget management.

[0125] The risk warning module sets multi-level budget thresholds and can trigger different levels of warning signals based on the deviation rate. Setting tolerance for discrepancies by department / project allows for more precise targeting of warnings to different departments and projects, improving the effectiveness of the warnings. The tiered warning unit uses different colors to indicate the warning level based on the magnitude of the deviation rate, intuitively reflecting the degree of risk in budget execution. The response strategy library stores three types of response strategies: budget freeze, resource reallocation, and emergency approval. This provides hospital management with rapid and effective measures to address budget risks, helping the hospital to control risks and reduce losses in a timely manner.

[0126] The process optimization module identifies redundant steps in the budgeting process through value stream mapping and generates targeted optimization suggestions. This module helps hospitals identify problems and bottlenecks in the budgeting process, reducing unnecessary steps and procedures and improving the efficiency of budget management. At the same time, the optimized process is simpler and more efficient, helping to improve collaboration between different departments within the hospital and promoting overall operational efficiency.

[0127] The resource allocation module combines queuing theory and game theory algorithms to optimize the allocation of medical resources. Chromosome encoding encodes resources such as hospital beds, operating rooms, and medical staff into binary gene chains. A fitness function integrates bed turnover rate, patient waiting time, and cost-effectiveness ratio to construct an evaluation system. The mutation operation introduces a simulated annealing mechanism to prevent local optima. This scientific resource allocation method fully considers the characteristics and interrelationships of various resources, achieving rational allocation and efficient utilization of resources, improving hospital operational efficiency and service quality, and providing patients with higher-quality medical services.

[0128] The visual interface provides multi-dimensional data dashboards and interactive decision support. 3D drill-down analysis supports data penetration across hospital-wide, departmental, ward, and treatment group levels, enabling hospital management to gain a deep understanding of budget execution at different levels. The scenario simulator predicts the impact of budget adjustments on KPIs in real time by dragging and dropping parameters, providing a scientific basis for decision-making. The collaborative annotation function allows multiple departments to annotate abnormal data online and trigger a countersigning process, strengthening inter-departmental communication and collaboration, and improving the scientific and democratic nature of decision-making.

[0129] The AI-based lean management approach for comprehensive hospital budgeting first constructs a multi-source heterogeneous data lake, integrating clinical, operational, and marketing data. This data integration method breaks down data silos, enabling centralized management and efficient utilization of data. By integrating data from different sources, hospitals can gain a more comprehensive understanding of their operational status and market environment, providing rich data resources for subsequent budget preparation and management, and helping to improve the accuracy and foresight of budget management.

[0130] The XGBoost algorithm was used to identify key influencing factors in budget preparation. The SHAP value was used to interpret the model output, and the top 20% of variables by feature importance were selected. Then, Pearson correlation coefficients were used to eliminate multicollinearity, ultimately retaining core factors such as disease weights, DRG payment standards, and equipment depreciation rates. This method accurately identifies the key factors affecting budget preparation, avoiding the consideration of too many irrelevant factors during the budget preparation process, improving the efficiency and accuracy of budget preparation, and making the budget plan more in line with the actual situation of the hospital.

[0131] A 12-month rolling budget forecasting model using LSTM networks can accurately predict the budget for the next 12 months based on historical data and current trends. Compared with traditional forecasting methods, LSTM networks have stronger nonlinear fitting capabilities and time-series data processing capabilities, enabling them to better capture complex patterns and trends in the data, improve the accuracy of budget forecasting, and provide a more reliable basis for hospital budget management.

[0132] This system utilizes reinforcement learning to implement closed-loop feedback control for budget execution. A Budget Execution Index (BEI) is established, and different strategies are adopted based on the BEI's value range. When BEI ∈ [0.9, 1.1], the current strategy is maintained; when BEI > 1.1, a cost-reduction and efficiency-enhancing strategy is triggered; and when BEI < 0.9, a budget reallocation mechanism is initiated. This closed-loop feedback control mechanism can promptly identify problems in the budget execution process and adjust accordingly, ensuring the effectiveness and rationality of budget execution and improving the precision of hospital budget management.

[0133] During the evaluation report generation process, the TOPSIS method was used to calculate the comprehensive budget execution score for each department. Radar charts were used to compare the planned and actual values ​​across six dimensions: scale, structure, schedule, quality, cost, and efficiency. Association rules were used to uncover hidden waste patterns and generate an improvement list. This comprehensive and in-depth evaluation method assesses budget execution from multiple perspectives, identifies potential problems and areas for improvement, and provides strong support for the hospital's continuous improvement.

[0134] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A lean management system for comprehensive budgeting in hospitals based on artificial intelligence, characterized in that, include: The data acquisition module is used to integrate data from the hospital's HIS system, financial system, and external market data. The budget preparation module uses machine learning algorithms to analyze historical budget execution data and generate dynamic budget plans. The execution monitoring module tracks the budget execution status in real time and calculates the deviation rate between actual expenditures and the budget. The dynamic adjustment module dynamically adjusts the budget allocation based on the bias rate and external environment parameters using a reinforcement learning model. The risk warning module allows setting multiple budget thresholds and triggering warning signals. The process optimization module identifies redundant steps in the budget process through value stream mapping and generates optimization suggestions. The resource allocation module combines queuing theory and game theory algorithms to optimize the allocation of medical resources. The visual interface provides multi-dimensional data dashboards and interactive decision support.

2. The hospital comprehensive budget lean management system based on artificial intelligence according to claim 1, characterized in that: The data acquisition module includes: Structured data units that interface with the hospital's ERP system for financial and operational data; Unstructured data units are used to parse medical equipment maintenance logs and clinical records using NLP technology; External data units acquire medical insurance policies, drug price indices, and epidemiological data in real time.

3. The hospital comprehensive budget lean management system based on artificial intelligence according to claim 1, characterized in that: The dynamic adjustment module implements budget reallocation using the following steps: Step S1: Calculate the fluctuation of resource demand in the next 3 months based on the time series forecasting model; Step S2: Evaluate the risk values ​​of different allocation strategies through Monte Carlo simulation; Step S3: Generate the Pareto optimal solution set using a multi-objective optimization algorithm; Step S4: Select the final allocation scheme based on the decision-maker's preference weight.

4. The hospital comprehensive budget lean management system based on artificial intelligence according to claim 1, characterized in that: The risk warning module includes: Threshold setting unit, set the tolerance for differences according to department / project; The tiered early warning unit triggers a yellow alert if the deviation rate is ≥10%, an orange alert if the deviation rate is ≥15%, and a red alert if the deviation rate is ≥20%. The response strategy library stores three types of response strategies: budget freeze, resource allocation, and emergency approval.

5. The hospital comprehensive budget lean management system based on artificial intelligence according to claim 1, characterized in that: The resource allocation module employs an improved genetic algorithm, including: Chromosome coding: Encoding hospital beds, operating rooms, and medical staff into binary gene chains; Fitness function: An evaluation system is constructed by combining bed turnover rate, patient waiting time, and cost-effectiveness ratio; Mutation operation: Introduce simulated annealing mechanism to prevent local optima.

6. The hospital comprehensive budget lean management system based on artificial intelligence according to claim 1, characterized in that: The visualization interface includes: 3D drill-down analysis: Supports data penetration from the hospital as a whole, departments, wards, and treatment groups. Scenario Simulator: Predict the impact of budget adjustments on KPIs in real time by dragging and dropping parameters; Collaborative annotation: Allows multiple departments to annotate abnormal data online and trigger a countersigning process.

7. A lean management method for comprehensive budgeting in hospitals based on artificial intelligence, characterized in that: Includes the following steps: Step T1: Construct a multi-source heterogeneous data lake to integrate clinical, operational, and marketing data; Step T2: Identify key influencing factors in budget preparation using the XGBoost algorithm; Step T3: Establish a 12-month rolling budget forecasting model using an LSTM network; Step T4: Implement closed-loop feedback control of budget-execution based on reinforcement learning; Step T5: Generate an evaluation report that includes cost savings rate, ROI, and patient satisfaction indicators.

8. The hospital comprehensive budget lean management method based on artificial intelligence according to claim 7, characterized in that: The method for extracting key influencing factors in step T2 includes: The SHAP value was used to interpret the model output, and the variables with the highest feature importance were selected. Eliminating multicollinearity using Pearson correlation coefficient; Ultimately, the core factors of disease weighting, DRG payment standards, and equipment depreciation rate will be retained.

9. The hospital comprehensive budget lean management method based on artificial intelligence according to claim 7, characterized in that: The closed-loop feedback control in step T4 is specifically as follows: Establish the Budget Execution Index (BEI) as follows: BEI = Actual Expenditures / (Budget Baseline × Environmental Adjustment Factor); The current policy is maintained when BEI ∈ [0.9, 1.1]. When BEI > 1.1, a cost reduction and efficiency improvement strategy is triggered; The budget redistribution mechanism is activated when BEI < 0.

9.

10. A hospital comprehensive budget lean management method based on artificial intelligence according to claim 7, characterized in that: The process of generating the assessment report includes: The overall budget execution score for each department was calculated using the TOPSIS method. Radar charts were used to compare the differences between planned and actual values ​​across six dimensions: scale, structure, schedule, quality, cost, and benefits. We can discover hidden waste patterns and generate an improvement list based on association rules.

Images