Dynamic DRG grouping system fusing ICD coding and multi-dimensional resource demand evaluation

Abstract

The invention provides a dynamic DRG grouping system integrating ICD coding and multi-dimensional resource demand evaluation. The system breaks through the limitation that traditional DRG grouping is single in basis and insufficient in dynamic adjustment, and more accurate case grouping is achieved by integrating multi-dimensional data such as patient conditions, treatment modes and resource consumption. A machine learning algorithm is introduced into the system, grouping standards are dynamically adjusted, and it is ensured that the system adapts to newest medical data and technical progress at any time. Meanwhile, a real-time feedback and self-learning mechanism is designed in the system, and a grouping algorithm is continuously optimized through feedback of medical staff. Through multi-level data integration and visual analysis, the system can generate a decision support report, and the hospital resource management and cost control efficiency is improved. According to the system, the accuracy of patient management and the rationality of resource utilization are remarkably improved, and the system has a wide application prospect.

Description

Technical Field

[0001] The present invention belongs to the field of medical artificial intelligence, and specifically relates to a dynamic DRG grouping system integrating ICD coding and multi-dimensional resource demand assessment. Background Art

[0002] The DRG (Diagnosis-Related Group) grouping system is a widely used patient classification method in hospital management, which aims to optimize resource allocation and control medical expenses by grouping patients according to their diagnosis and treatment methods. Although the existing DRG grouping system has achieved certain results in promoting the standardization of medical services and improving management efficiency, it still has several significant defects.

[0003] (1) Single grouping basis: The traditional DRG system mainly groups patients based on ICD codes, ignoring individual differences and treatment complexity. This single classification method may lead to significant differences in resource consumption among patients in the same DRG group, which in turn affects the hospital's resource allocation and cost control.

[0004] (2) Insufficient dynamic adjustment: Existing systems usually use static grouping standards, have long update cycles, and cannot respond to the development of medical technology and changes in treatment plans in a timely manner. This causes hospitals to face the problem of inaccurate grouping during use and cannot effectively reflect the actual medical needs of patients.

[0005] (3) Incomplete resource demand assessment: The resource demand assessment of the current DRG system is often incomplete, mainly relying on historical data, and unable to consider the diverse resource needs that patients may face during treatment. Such deficiencies may lead to waste and shortage of medical resources.

[0006] (4) Weak feedback mechanism: Most existing DRG systems lack an effective user feedback mechanism, and it is difficult for medical staff to provide real-time feedback during use. This situation limits the optimization and iteration of the system, resulting in the grouping algorithm being unable to improve with changes in actual usage. Summary of the invention

[0007] The purpose of the present invention is to provide a dynamic DRG grouping system integrating ICD coding and multi-dimensional resource demand assessment, comprising (1) a patient information input module, (2) a multi-dimensional resource demand assessment module, (3) a dynamic DRG grouping module, (4) a data analysis and feedback module, (5) a medical insurance connection and fee settlement module, and (6) a report generation module.

[0008] The patient information input module is used to record and store the patient's basic information and medical records, and can query and track the patient's medical history; the specific steps include: Medical staff enter the patient's basic information into the system, including name, age, gender, etc.; Enter the ICD codes for the primary and secondary diagnoses; At the same time, the input patient information, ICD codes and other relevant data are stored in the database for subsequent analysis.

[0009] The multi-dimensional resource demand assessment module is used to collect data, build a resource consumption model, calculate the resource consumption coefficient, and reflect the patient's medical resource consumption.

[0010] This system is not only based on ICD coding, but also integrates multi-dimensional resource demand assessment to more comprehensively reflect the actual needs of patients.

[0011] In the data input interface, the system obtains patient information from multiple data sources such as electronic medical records (EMR), drug management systems, and equipment management systems, including disease type, treatment plan, drug usage, and resource consumption records.

[0012] Construct a multi-dimensional resource demand assessment model, mainly considering the following factors: (1) Basic information of the patient, such as age, gender, etc.; (2) main diagnosis, main surgery, and disease severity; (3) treatment methods, including surgery, drug therapy, etc.; (4) expected length of hospital stay; Then use the following model to calculate: Among them, R is the total resource demand, C i is the unit cost of the i-th resource, U i is the estimated usage of the i-th resource, and n is the total number of resources.

[0013] At the same time, it is necessary to compare historical data with actual resource consumption data and dynamically adjust the resource consumption coefficient (RCC). The calculation formula is: RCC new =RCC old ×(1+Adjustment Factor) (2) The Adjustment Factor is an adjustment factor recalculated based on actual usage.

[0014] The dynamic DRG grouping module comprehensively considers the severity of the patient's disease, treatment methods, and resource consumption, and simulates DRG grouping through machine learning. It also updates data based on monthly settlement conditions (DRG grouping conditions, medical insurance payment standards, etc.) to improve the accuracy of the model and extract effective information to help users develop response strategies.

[0015] Through machine learning algorithms, this module enables the system to analyze patient data in real time and dynamically adjust DRG groupings. Algorithm selection: Machine learning algorithms such as Random Forest or Gradient Boosting Trees are used to analyze the patient's ICD code and determine the DRG grouping model to which it belongs.

[0016] According to the mapping relationship between ICD codes and DRGs, the preliminary DRG group is determined, and the calculation formula is: DRG initial =f(ICD 1 ,ICD 2 ,…,ICD n ) (3) Here, f is a mapping function, which can determine the DRG group through a lookup table or a machine learning model.

[0017] Dynamic adjustment: The model is updated regularly (e.g. monthly) and the grouping criteria are adjusted based on real-time data feedback such as new cases. The feedback factor is: Among them, RCC i is the resource consumption coefficient of the i-th DRG group, and N is the number of DRG groups.

[0018] Data analysis and feedback module First, trend analysis is performed to analyze resource usage trends of different DRG groups and identify abnormal consumption; The calculation formula is: Among them, Total Cost DRG Total Cases is the total cost of a DRG group. DRG is the number of cases in this group; Secondly, feedback is collected, mainly collecting feedback from medical staff on DRG grouping and resource consumption assessment, and forming a feedback report.

[0019] The calculation formula for feedback satisfaction is as follows: Among them, Ratingk is the score of the kth feedback, and P is the total number of feedbacks.

[0020] By analyzing and classifying feedback texts, common problems and improvement directions are identified, and the model is regularly updated and optimized to form a closed-loop self-learning system.

[0021] Medical insurance connection and fee settlement module From the perspective of the medical insurance interface, when designing the interface with the medical insurance system, on the one hand, it is necessary to ensure the security and accuracy of data transmission through data encryption and security authentication, and on the other hand, it is necessary to ensure the accuracy of fee settlement.

[0022] The fees payable are: Payment DRG =Base Rate DRG ×RCC (7) Among them, Base Rate DRG To connect medical insurance to the benchmark payment standard for this DRG group, RCC is the resource consumption coefficient mentioned above.

[0023] Report generation module Design a standardized reporting module, mainly involving resource utilization and cost analysis reports. Detailed reports need to be generated, including the number of patients, total costs, average costs, RCC and other information corresponding to each DRG group, as well as resource consumption lists, cost trend lists and satisfaction scores for management decision-making.

[0024] At the same time, implement an automated report generation program to output reports regularly and send them to management.

[0025] Of course, it also includes user customization functions, allowing users to customize report content and format to meet different needs.

[0026] System maintenance and updates On the one hand, it is necessary to conduct regular data audits, establish a data audit mechanism and regularly update data, such as regularly checking and updating the ICD coding library and resource consumption database to ensure the timeliness of the system.

[0027] On the other hand, monitoring and evaluation are needed, and continuous optimization can be achieved through continuous monitoring of system performance, data analysis, and the creation of a user feedback platform to collect problems and suggestions during use.

[0028] Through the specific steps and calculation formulas of these modules, the dynamic DRG grouping system that integrates ICD codes and actual multi-dimensional resource needs can effectively improve the accuracy of medical management and resource utilization efficiency, and ultimately provide patients with better medical services.

[0029] The system of the present invention has been significantly improved in the following aspects: (1) Multi-dimensional resource assessment: This system is not only based on ICD codes, but also comprehensively considers the complexity of the patient's condition, treatment plan and its corresponding multi-type resource requirements, so as to achieve more accurate patient case grouping. This multi-dimensional assessment can effectively reduce the differences in resource consumption between different patients; (2) Dynamic grouping mechanism: By introducing advanced machine learning algorithms, the system can analyze and adjust DRG grouping in real time to ensure that the grouping criteria can quickly respond to new medical data and feedback. This can greatly improve the accuracy of grouping and ensure personalized medical services; (3) Real-time feedback and self-learning: A real-time feedback mechanism is designed to collect opinions and suggestions from medical staff and management. The system can continuously optimize the grouping algorithm and resource demand model based on this feedback. This self-learning ability makes the system more intelligent and adaptable; (4) Comprehensive data integration: This system can integrate data from different medical information systems and conduct multi-level data analysis, thereby providing more comprehensive patient management support and helping hospitals make scientific decisions.

[0030] Through these improvements, this system can not only address the shortcomings of the existing DRG grouping system, but also provide hospitals with more efficient and accurate management tools, ultimately improving the quality of medical services and patient satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 : Flowchart of the dynamic DRG grouping module.

[0032] Figure 2 :Real-time feedback and self-learning process.

[0033] Figure 3 :Flowchart of dynamic DRG grouping module of NC1 group.

[0034] Figure 4 :DRG specific grouping real-time feedback and self-learning process.

[0035] Figure 5 :DRG improvement suggestions enhance system effectiveness diagram.

[0036] Figure 6 :Adjust treatment plans to reduce medical costs. DETAILED DESCRIPTION

[0037] The present invention is further described in conjunction with the accompanying drawings and embodiments, but the protection scope of the present invention is not limited thereto.

[0038] Example 1

[0039] In the data input interface, the system obtains patient information from multiple data sources such as electronic medical records (EMR), drug management systems, and equipment management systems, including disease type, treatment plan, drug usage, and resource consumption records.

[0040] To build a multi-dimensional resource demand model, we mainly consider the following factors: (1) Basic information of the patient, such as age, gender, etc.; (2) main diagnosis, main surgery, and disease severity; (3) treatment methods, including surgery, drug therapy, etc.; (4) Expected length of hospital stay.

[0041] Then use the following model to calculate: Among them, R is the total resource demand, C i is the unit cost of the i-th resource, U i is the estimated usage of the i-th resource.

[0042] At the same time, it is necessary to compare historical data with actual resource consumption data and dynamically adjust the resource consumption coefficient (RCC). The calculation formula is: RCC new =RCC old ×(1+Adjustment Factor) (2) The Adjustment Factor is an adjustment factor recalculated based on actual usage.

[0043] Through machine learning algorithms, this module enables the system to analyze patient data in real time and dynamically adjust DRG groupings.

[0044] Data processing flow Figure 1 shown.

[0045] Algorithm selection: Machine learning algorithms such as Random Forest or Gradient Boosting Trees are used to analyze the patient's ICD code and determine the DRG grouping model to which it belongs.

[0046] According to the mapping relationship between ICD codes and DRGs, the preliminary DRG group is determined, and the calculation formula is: DRG initial =f(ICD 1 ,ICD 2 ,…,ICD n ) (3) Here, f is a mapping function, which can determine the DRG group through a lookup table or a machine learning model.

[0047] Dynamic adjustment: The model is updated regularly (e.g. monthly) and the grouping criteria are adjusted based on real-time data feedback such as new cases.

[0048] The feedback factor is: Among them, RCC i is the resource consumption coefficient of the i-th DRG group, and N is the number of DRG groups.

[0049] Data analysis and feedback module First, trend analysis is performed to analyze resource usage trends of different DRG groups and identify abnormal consumption; The calculation formula is: Among them, Total Cost DRG Total Cases is the total cost of a DRG group. DRG is the number of cases in this group; Secondly, feedback is collected, mainly collecting feedback from medical staff on DRG grouping and resource consumption assessment, and forming a feedback report.

[0050] The calculation formula for feedback satisfaction is as follows: Among them, Rating k is the score of the kth feedback, and P is the total number of feedbacks.

[0051] By analyzing and classifying feedback texts, common problems and improvement directions are identified, and the model is regularly updated and optimized to form a closed-loop self-learning system.

[0052] The specific real-time feedback and self-learning process is as follows: Figure 2 shown.

[0053] Medical insurance connection and fee settlement module From the perspective of the medical insurance interface, when designing the interface with the medical insurance system, on the one hand, it is necessary to ensure the security and accuracy of data transmission through data encryption and security authentication, and on the other hand, it is necessary to ensure the accuracy of fee settlement.

[0054] The fees payable are: Payment DRG =Base Rate DRG ×RCC (7) Among them, Base Rate DRGTo connect medical insurance to the benchmark payment standard for this DRG group, RCC is the resource consumption coefficient mentioned above.

[0055] Report generation module Design a standardized reporting module, mainly involving resource utilization and cost analysis reports. Detailed reports need to be generated, including the number of patients, total costs, average costs, RCC and other information corresponding to each DRG group, as well as resource consumption lists, cost trend lists and satisfaction scores for management decision-making.

[0056] Example 2 Application example of the system of the present invention

[0057] 1. System deployment The system architecture includes: (1) Front-end user interface (UI): medical staff input patient information through the web application; (2) Back-end database: stores patient information, ICD codes, resource requirements, and grouping results; (3) Data processing module: executes machine learning algorithms and dynamic adjustment logic; (4) Visual analysis tools: used by management to view reports and analyze trends.

[0058] 2. Implementation steps

[0059] Step 1: Data input, including: (1) When a patient is admitted to the hospital, medical staff enter the patient's basic information, medical history, and ICD code into the system to form a data set; (2) The system verifies input information in real time to ensure the validity of the ICD code.

[0060] Step 2: Resource Needs Assessment The system calculates the estimated resource requirements based on the input patient information using formula (1) above.

[0061] For example, for a hospitalized patient, the system might assess resource requirements as follows: Drugs: C 1 = 10 yuan, estimated usage U 1 = 5 yuan, then the demand for medicine = 50 yuan Check: C 2 = 100 yuan, estimated usage U 2 = 2 yuan, then the inspection demand = 200 yuan Therefore, the total resource requirement is R=50+200=250 yuan.

[0062] Step 3: Dynamic DRG Grouping The system uses machine learning models to dynamically group DRGs based on patient information and resource demand data.

[0063] Import historical medical insurance DRG settlement data into the system, collect data such as patient age, diagnosis, surgery, medical expenses, and combine clinical experience to screen out grouping feature sets. Crop the collected data, clean up the data with incomplete information and wrong grouping according to the data volume and data characteristics, and perform noise reduction on the retained data. Identify the diagnosis code and surgical operation in the settlement list, determine which MDC main diagnosis table it belongs to according to the patient's main diagnosis, and determine whether it belongs to surgical operation ADRG, non-surgical operation ADRG or internal medicine diagnosis ADRG according to the patient's main surgical operation. Finally, combine the patient's age and complications / comorbidities to enter a specific DRG group. For example, the patient's main diagnosis is multiple uterine leiomyoma, which is judged to belong to MDCN (diseases and dysfunctions of the female reproductive system). The main surgery: laparoscopic abdominal hysterectomy belongs to NC1-uterus (except uterine cavity lesions) surgery group, and then combine the patient's age, complications / comorbidities, etc. to determine whether it belongs to group 11 (with severe complications and complications), group 13 (with general complications and complications) or group 15 (without complications and complications). Due to different resource consumption, the payment standards for the three DRG groups are 16,874.17, 14,861.67, and 11,450.92 yuan respectively.

[0064] Data processing flow Figure 3 shown.

[0065] Step 4: Real-time feedback and self-learning During the patient treatment process, medical staff can submit feedback through the system, such as "the grouping results are inaccurate" or "the resource demand assessment is low." In addition, the monthly grouping data and settlement status published by the medical insurance are regularly imported into the system. The system uses machine learning to determine abnormal data and resource consumption trends as well as changes in medical insurance payment standards. It continuously repeats the learning process and dynamically generates new models to improve the accuracy of grouping.

[0066] The system records and analyzes this feedback and regularly retrains the machine learning model to improve grouping accuracy.

[0067] The specific real-time feedback and self-learning process is as follows: Figure 4 shown.

[0068] Step 5: Data integration and analysis The system will integrate all relevant data of patients, including treatment results, resource usage, etc., and store them in the data warehouse.

[0069] Management uses Tableau to generate visual reports to analyze resource utilization, patient satisfaction, and grouping accuracy.

[0070] 3. Actual Results In the early stages of implementing the system, the hospital found that the accuracy of patients' DRG grouping increased by 20% and resource utilization efficiency increased by 15%.

[0071] The medical staff feedback system received a positive response, and the improvement suggestions were adopted in a timely manner, which enhanced the practicality of the system, such as Figure 5 shown.

[0072] Through data analysis, the management discovered some abnormal trends in resource consumption and promptly adjusted relevant treatment plans, further reducing medical costs, such as Figure 6 shown.

[0073] Through the above examples, the practical application of the dynamic DRG grouping system designed by the present invention in a hospital is demonstrated. The system successfully combines ICD coding and multi-dimensional resource demand assessment with a dynamic and real-time system, which not only improves the accuracy of grouping, but also enhances the resource management capabilities of the hospital and achieves better quality medical services.

Claims

1. A dynamic DRG grouping system integrating ICD coding and multi-dimensional resource demand assessment, characterized in that: The system includes: (1) Patient information input module, which is used to record and store the patient's basic information and medical records, and can query and track the patient's medical history; (2) A multi-dimensional resource demand assessment module, which is used to collect data, build a resource consumption model, calculate the resource consumption coefficient, and reflect the patient's medical resource consumption; (3) Dynamic DRG grouping module, which uses machine learning algorithms to analyze patient data in real time and dynamically adjust DRG grouping; (4) Data analysis and feedback module, which analyzes and classifies feedback texts, identifies common problems and improvement directions, and regularly updates and optimizes the model to form a closed-loop self-learning system; (5) Medical insurance connection and fee settlement module: Design the interface with the medical insurance system and ensure the security and accuracy of data transmission and the accuracy of fee settlement through data encryption and security authentication; (6) Report generation module, which realizes the automated report generation process by generating detailed reports for each DRG group, outputs reports regularly and sends them to the management.

2. The dynamic DRG grouping system according to claim 1, characterized in that: The specific steps of the patient information input module include: (1) Medical staff enter the patient's basic information into the system, including name, age, and gender; (2) Enter the ICD codes for the primary and secondary diagnoses; (3) Store the input patient information, ICD codes, and other relevant data in the database for subsequent analysis.

3. The dynamic DRG grouping system according to claim 1, characterized in that: In the data input interface of the multi-dimensional resource demand assessment module, patient information is obtained from multiple data sources such as electronic medical records (EMR), drug management system, and equipment management system, including disease type, treatment plan, drug usage, and resource consumption records, and the following model is used for calculation: Among them, R is the total resource demand, C i is the unit cost of the i-th resource, U i is the estimated usage of the i-th resource, and n is the total number of resources; According to the comparison between historical data and actual resource consumption data, the resource consumption coefficient (RCC) is adjusted dynamically. The calculation formula is: RCC new =RCC old ×(1+Adjustment Factor) (2) The Adjustment Factor is an adjustment factor recalculated based on actual usage.

4. The dynamic DRG grouping system according to claim 1, characterized in that: The dynamic DRG grouping module comprehensively considers the severity of the patient's disease, treatment methods, and resource consumption, performs DRG grouping simulation through machine learning, and performs data update learning based on the monthly settlement situation; The algorithm: Use random forest or gradient boosting tree machine learning algorithm to analyze the patient's ICD code to determine the DRG grouping model to which it belongs. According to the mapping relationship between ICD code and DRG, determine the preliminary DRG group. The calculation formula is: DREG initial =f(ICD1,ICD2,…,ICD n ) (3) Wherein, f is a mapping function, which can determine the DRG group through a lookup table or a machine learning model; Dynamic adjustment: The model is updated regularly based on real-time data feedback from new cases, and the grouping criteria are adjusted. The feedback factor is: Among them, RCC i is the resource consumption coefficient of the i-th DRG group, and N is the number of DRG groups.

5. The dynamic DRG grouping system according to claim 1, characterized in that: The data analysis and feedback module (1) performs trend analysis to analyze resource usage trends of different DRG groups and identify abnormal consumption. The calculation formula is: Among them, Total Cost DRG Total Cases is the total cost of a DRG group. DRG is the number of cases in this group; (2) Collect feedback, mainly collecting feedback from medical staff on DRG grouping and resource consumption assessment, and forming a feedback report. The calculation formula for feedback satisfaction is as follows: Among them, Rating k is the score of the kth feedback, and P is the total number of feedbacks.

6. The dynamic DRG grouping system according to claim 1, characterized in that: The medical insurance connection and fee settlement module, in which the fees to be paid are: Payment DRG =Base Rate DRG ×RCC (7) Among them, Base Rate DRG To connect medical insurance to the benchmark payment standard for this DRG group, RCC is the resource consumption coefficient mentioned above.

7. The dynamic DRG grouping system according to claim 1, characterized in that: The report generation module mainly involves resource utilization and cost analysis reports, and needs to generate detailed reports, including the number of patients corresponding to each DRG group, total cost, average cost, RCC information and resource consumption list, cost trend list and satisfaction score.

8. The dynamic DRG grouping system according to claim 7, characterized in that: The report generation module includes a user customization function, allowing users to customize report content and format to meet different needs.

9. The dynamic DRG grouping system according to claim 1, characterized in that: The dynamic DRG grouping system also includes system maintenance and updates, regular data audits, establishment of a data audit mechanism and regular data updates, monitoring and evaluation, and continuous optimization through continuous monitoring of system performance, data analysis, and creation of a user feedback platform to collect problems and suggestions during use.

Images