Medical management method

Abstract

The invention discloses a medical management method, and relates to the technical field of medical management, and the method comprises the following steps: S1, employing a PACS system to obtain image data of a fracture region of a patient, employing an OCR technology to convert paper record data into electronic format data, synchronously obtaining pain numerical data, and uploading the image data, the electronic format data and the pain numerical data; the technical key points are as follows: potential laws in historical rehabilitation data can be mined by adopting a machine learning algorithm, an accurate prediction model is established, the accuracy of predicting the fracture rehabilitation progress of a patient is improved, and the prediction performance of the model is continuously optimized by comparing a prediction result with an actual result and adopting a model parameter fine tuning method. And meanwhile, specific arrangement is made for identification and adjustment of key features in the model, so that the prediction model is further optimized, and the accuracy of the model for the rehabilitation progress of the patient is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical management, and in particular to a medical management method. Background Art

[0002] Medical management refers to the organization, planning, coordination and control of the entire process of hospital medical system activities, aiming to keep the medical system in a state of regular operation and have strong adaptability to the changing objective environment, so as to achieve the best medical efficiency and medical effect. When managing medical data for patients in orthopedics, it is usually necessary to obtain or collect the basic information, condition and other relevant information of the corresponding patient as a medical data package to reflect the situation of the corresponding patient, so as to facilitate subsequent analysis or processing according to the corresponding medical data package.

[0003] In the past, the prediction of the rehabilitation progress of fracture patients mainly relied on the doctor's experience and intuition, lacking scientific basis. In the future, relevant algorithms were used to analyze historical rehabilitation data to establish a rehabilitation progress prediction model. The initially established model may only perform well for training data and have limited predictive ability for new patients. Routine adjustments were then made, which could not guarantee the accuracy and precision of the prediction results, and was not conducive to the management of medical data. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the shortcomings of the prior art, the present invention provides a medical management method, which adopts a machine learning algorithm to explore the potential patterns in historical rehabilitation data and establish an accurate prediction model. By comparing the predicted results with the actual results and fine-tuning the model parameters, the predictive performance of the model is continuously optimized. At the same time, specific arrangements are made in the model for the identification and adjustment of key features, which increases the accuracy of the model for the patient's rehabilitation progress and solves the problems raised in the background technology.

[0006] (II) Technical solution

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0008] A medical management method comprises the following steps:

[0009] S1. Use the PACS system to obtain image data of the patient's fracture area, use OCR technology to convert paper record data into electronic format data, and simultaneously obtain pain numerical data, and upload the image data, electronic format data and pain numerical data;

[0010] S2, receiving the image data, the electronic format data and the pain numerical data, integrating the image data, the electronic format data and the pain numerical data into a database, and unifying the format and standard of the data;

[0011] S3. Analyze the collected historical rehabilitation data using machine learning algorithms and establish a rehabilitation progress prediction model;

[0012] S4, inputting the corresponding patient data collected by real-time monitoring into the rehabilitation progress prediction model, and comparing the obtained prediction results with the actual results;

[0013] S5. Fine-tune the model parameters to optimize the model according to the comparison results until the comparison results meet expectations.

[0014] Furthermore, image data include but are not limited to X-rays and MRI scan results; paper records include surgical records and medication usage; pain numerical data are patients' self-ratings of the degree of pain in their fracture area, ranging from 1 to 10 points, with 1 point indicating the lowest pain value and 10 points indicating the highest pain value.

[0015] Furthermore, the image data, electronic format data and pain numerical data are integrated into the same database, and the data format and standard are unified by using FHIR to standardize the data.

[0016] Furthermore, historical rehabilitation data include, but are not limited to: basic information, fracture location and type, initial injury severity, and total length of time required for rehabilitation;

[0017] Among them, basic information includes the patient's age and gender; the initial injury severity indicates the hospital's preliminary diagnosis, including the degree of fracture displacement; the total time required for recovery indicates the time from injury to recovery.

[0018] Furthermore, the process of establishing a rehabilitation progress prediction model is as follows: S301, cleaning historical rehabilitation data;

[0019] S302, selecting features related to fracture healing time from the cleaned historical rehabilitation data, including age, fracture type, and initial injury severity;

[0020] S303, selecting an algorithm suitable for time series analysis, including any one of ARIMA and LSTM;

[0021] S304, training the selected model using historical rehabilitation data, and adjusting model parameters to minimize prediction errors;

[0022] S305, testing the prediction performance of the model using an independent validation data set;

[0023] S306. Deploy the trained model to predict the fixed value range of fracture healing time of new patients in real time.

[0024] Furthermore, the prediction result indicates: the predicted fixed value range of the fracture healing time of the corresponding patient; the actual result indicates: the actual corresponding fracture healing time of the corresponding patient.

[0025] Furthermore, the comparison result represents the difference between the predicted result and the actual result. The model parameters are fine-tuned to optimize the model. The optimization process is as follows: identify key features, set the weights of key features in the model based on the comparison results, and limit the weights corresponding to each adjustment of key features in the following way:

[0026] Q = α*(|yc-jc|);

[0027] Where Q represents the added weight of each adjustment of the key features, α represents an adjustment coefficient, and 0<α<1, yc and jc represent the predicted value of the fracture healing time of the corresponding patient and the actual fracture healing time of the corresponding patient, respectively.

[0028] Furthermore, the process of identifying key features is: using any one of the analysis algorithm set to identify key features, and the analysis algorithm set includes but is not limited to feature importance evaluation algorithm, Pearson correlation coefficient, PDP and ICE graph and automatic feature selection algorithm.

[0029] Furthermore, the process of optimizing the model is:

[0030] According to the increased weight Q of the key features each time, the weight of the key features in the model is increased to obtain a new rehabilitation progress prediction model, and S4 is re-executed;

[0031] If the difference in the comparison result does not exceed the previous comparison result, the difference between the previous comparison result difference and the current comparison difference is calculated. When the difference is greater than the preset standard threshold, it does not meet expectations, and S4 is continued until the difference is less than or equal to the preset standard threshold, which meets expectations. The current rehabilitation progress prediction model is used to continue predicting and managing the next patient.

[0032] If the difference in the comparison result exceeds the previous comparison result, the algorithm replacement strategy is triggered to replace the analysis algorithm used to identify the key features until the difference in the comparison result does not exceed the previous comparison result.

[0033] (III) Beneficial effects

[0034] The present invention provides a medical management method, which has the following beneficial effects:

[0035] Digital, networked and information-based management methods allow doctors to obtain and analyze comprehensive patient data more quickly, improving diagnosis and treatment efficiency. Image data, electronic format data and pain numerical data are integrated into the same database, ensuring the relevance and consistency of the data, facilitating subsequent comprehensive analysis.

[0036] The use of machine learning algorithms can explore the potential patterns in historical rehabilitation data, establish accurate prediction models, and improve the accuracy of predicting patients' fracture rehabilitation progress. By comparing the predicted results with the actual results and fine-tuning the model parameters, the model's predictive performance is continuously optimized. At the same time, specific arrangements are made in the model for the identification and adjustment of key features, thereby further optimizing the prediction model and increasing the accuracy of the model for patients' rehabilitation progress. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 The figure is a flow chart of the overall steps of a medical management method in the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] See also Figure 1 This embodiment provides a medical management method, which includes the following steps:

[0040] S1. Use the PACS system to obtain image data of the patient's fracture area, use OCR technology to convert paper record data into electronic format data, and simultaneously obtain pain numerical data, and upload the image data, electronic format data and pain numerical data;

[0041] Among them, the PACS system, which stands for Picture Archiving and Communication System, is an indispensable part of modern medical technology. The system is mainly used in hospital imaging departments, and is designed to store a large number of daily medical images (including images produced by MRI, CT, ultrasound, various X-ray machines, infrared devices, microscopes, etc.) in a digital manner through various interfaces (analog, DICOM, network, etc.); when needed, it can be quickly recalled for use under certain authorization, and some auxiliary diagnosis management functions can be added;

[0042] The following is a detailed analysis of the PACS system:

[0043] 1. System Overview

[0044] The PACS system is a professional medical information management system, which is mainly composed of image acquisition, transmission, storage, display and network. It is the foundation of digital imaging and an important component of the hospital information system (HIS). The system greatly improves the utilization rate of medical images and the efficiency of diagnosis and treatment by realizing the digitization, networking and information management of medical images.

[0045] 2. Main functions

[0046] Image storage and transmission: The PACS system can digitally store and transmit images produced by various medical imaging devices, reducing the demand for physical space and the maintenance costs of films and related equipment;

[0047] Auxiliary diagnosis management: The system supports a variety of image processing functions, such as magnification, rotation, contrast adjustment, etc., which help doctors make more accurate diagnoses; in addition, the PACS system also provides auxiliary diagnosis management functions, such as case analysis, report generation, etc.

[0048] Telemedicine: Supports remote access and consultation, allowing doctors to view patient imaging data in different locations, facilitating remote consultation and second opinions;

[0049] Data sharing and collaboration: It promotes data sharing and collaboration among different medical institutions and improves the continuity of medical services;

[0050] Patient services: Patients can more conveniently access their medical images and reports through the PACS system, which increases the transparency of medical services and patient participation;

[0051] 3. System composition

[0052] The PACS system mainly consists of the following parts:

[0053] Image acquisition equipment: including CT, MRI, X-ray machines, ultrasound and other medical imaging equipment, which convert medical images into digital signals through digital technology;

[0054] Image storage devices: used to store digital medical images for a long time. Common storage devices include disk arrays, CD libraries, and tape libraries.

[0055] Image transmission equipment: responsible for transmitting medical images from one device to another, or from one department to another. Common transmission equipment includes network switches, routers, fiber optic transmission equipment, etc.

[0056] Image processing equipment: used to perform various processing operations on digital medical images, such as image enhancement, image reconstruction, image analysis, etc.;

[0057] Image display device: used to display processed medical images to doctors or other medical staff. Common display devices include medical monitors, projectors, etc.

[0058] 4. Technical Features

[0059] Standardization: The PACS system complies with the DICOM (Digital Imaging and Communications in Medicine) standard, which is an international standard for medical imaging and image communication, ensuring compatibility and data interchangeability between different imaging devices;

[0060] Efficiency: Digital storage and transmission methods have greatly improved the utilization efficiency and diagnosis and treatment efficiency of medical images, and doctors can obtain and analyze image data more quickly;

[0061] Security: PACS systems typically include strict data security measures to ensure the privacy and security of patient data while meeting healthcare regulatory requirements;

[0062] In summary, the PACS system is an indispensable part of modern medical technology. It greatly improves the utilization rate and diagnosis and treatment efficiency of medical images by realizing the digitization, networking and information management of medical images, and provides doctors with a more convenient and efficient working method.

[0063] Image data includes but is not limited to X-rays and MRI scan results; Example: After a patient completes an MRI scan, the scan results are automatically uploaded to the PACS system and synchronized to the corresponding management system through the API interface;

[0064] Paper records, including surgical notes and medication usage, are converted to electronic format using OCR technology for subsequent entry into the system;

[0065] The pain numerical data are the patients' self-ratings of the pain level in their fracture area, ranging from 1 to 10 points, with 1 point indicating the lowest pain value and 10 points indicating the highest pain value;

[0066] S2, receiving the image data, the electronic format data and the pain numerical data, integrating the image data, the electronic format data and the pain numerical data into a database, and unifying the format and standard of the data;

[0067] The purpose of integrating image data, electronic format data, and pain numerical data into the same database is to: ensure the relevance and consistency of data; use unified data formats and standards, such as FHIR (Fast Healthcare Interoperability Resources), to facilitate data exchange between different systems;

[0068] Example: A patient’s X-rays, surgical notes, and pain scores are all stored in the same database and shared with other systems via the FHIR standard.

[0069] Digital, networked and information-based management methods allow doctors to obtain and analyze comprehensive patient data more quickly, improve diagnosis and treatment efficiency, integrate image data, electronic format data and pain numerical data into the same database, ensure the relevance and consistency of the data, and facilitate subsequent comprehensive analysis.

[0070] S3. Analyze the collected historical rehabilitation data using machine learning algorithms and establish a rehabilitation progress prediction model;

[0071] Among them, historical rehabilitation data refers to various data records of orthopedic patients from injury to rehabilitation in the past period of time, and historical rehabilitation data includes but is not limited to:

[0072] Basic information, fracture location and type, initial injury severity, and total length of time required for recovery;

[0073] Specifically, basic information includes the patient's age and gender; the initial injury severity indicates the hospital's initial diagnosis, including the degree of fracture displacement, such as: moderate or moderate displacement; the total time required for rehabilitation indicates the time from injury to reaching a specific rehabilitation standard;

[0074] The process of establishing a rehabilitation progress prediction model is as follows:

[0075] S301, data preprocessing:

[0076] Clean historical rehabilitation data to remove incomplete or erroneous records;

[0077] Depending on actual needs, this may also include: converting data formats to ensure that all data are numerical for mathematical calculations; normalizing or standardizing data so that indicators of different magnitudes can be used in the same model;

[0078] S302, feature selection:

[0079] Select features associated with fracture healing time from preprocessed historical rehabilitation data, including age, fracture type, and initial injury severity; also include gender and total time required for rehabilitation;

[0080] S303, Model selection:

[0081] Select an algorithm suitable for time series analysis, including ARIMA (autoregressive integrated moving average model) and LSTM (long short-term memory network);

[0082] S304, model training:

[0083] The selected model is trained using historical rehabilitation data, and model parameters are adjusted to minimize prediction error;

[0084] S305, Model Verification:

[0085] Use an independent validation dataset to test the model’s predictive performance to ensure that the model has generalizability;

[0086] S306. Model deployment:

[0087] The trained model is deployed to predict the fracture healing time range of new patients in real time;

[0088] S4, inputting the corresponding patient data collected by real-time monitoring into the rehabilitation progress prediction model, and comparing the obtained prediction results with the actual results;

[0089] The prediction results are obtained by referring to the following content:

[0090] Prediction model: A mathematical model that can predict fracture healing time based on the patient's initial conditions and rehabilitation process data; Prediction results: For each new patient, the model will output a predicted fracture healing time constant range, for example: 135 days ± 5 days;

[0091] Example:

[0092] Suppose there is a historical rehabilitation data set containing the rehabilitation records of 1,000 fracture patients. We use this data to train an LSTM model to predict fracture healing time. After the model training is completed, we can input the data of a new patient, such as age 45, female, tibia fracture, and initial injury of moderate displacement. The model will output a predicted fracture healing time constant range, such as 135 days ± 5 days. In this way, doctors can develop more personalized treatment plans based on this prediction result and make timely adjustments during the rehabilitation process to improve the patient's rehabilitation effect and satisfaction.

[0093] The prediction results indicate: the predicted range of fracture healing time for the corresponding patient;

[0094] The actual result means: the actual corresponding fracture healing time of the patient.

[0095] S5. Fine-tune the model parameters to optimize the model according to the comparison results until the comparison results meet expectations;

[0096] The comparison result is the difference between the predicted result and the actual result. The model is optimized by fine-tuning the model parameters. The optimization process is: identify the key features, set the weights of the key features in the model based on the comparison results, and limit the weights corresponding to each adjustment of the key features in the following way:

[0097] Q = α*(|yc-jc|);

[0098] In the formula, Q represents the added weight of each adjustment of the key features, α represents an adjustment coefficient, which is used to control the amplitude of the weight adjustment, and 0<α<1, yc and jc represent the predicted value of the fracture healing time of the corresponding patient and the actual fracture healing time of the corresponding patient, respectively. Combined with the above example, the predicted value of the fracture healing time of the corresponding patient is 135 days out of 135 days±5 days;

[0099] The process of identifying key features is as follows:

[0100] Using any one of the analysis algorithm set to identify key features; in this embodiment, an automated feature selection algorithm is used to identify key features; and the analysis algorithm set includes but is not limited to: feature importance assessment algorithm, Pearson correlation coefficient, PDP and ICE graph, and automated feature selection algorithm;

[0101] Among them, the feature importance evaluation algorithm:

[0102] Random Forest: The Random Forest algorithm can calculate the importance score of each feature, which is measured by observing the average depth of the feature in the tree or the average reduction in impurity; Gradient Boosting Tree (GBM): The GBM algorithm can also provide an assessment of feature importance, which is measured by observing the contribution of the feature in the boosting process;

[0103] Correlation analysis:

[0104] Pearson correlation coefficient: Calculate the Pearson correlation coefficient between each feature and the target variable to identify features that are strongly correlated with the target variable;

[0105] Partial Dependence Plot (PDP) and Individual Conditional Expectation (ICE):

[0106] Plot partial dependency plots for each feature to see how the feature affects the prediction. ICE plots can provide a more granular view of how individual data points are affected by feature values.

[0107] Automated feature selection algorithms:

[0108] Use automated feature selection algorithms, such as recursive feature elimination (RFE), model-based feature selection (such as linear models with L1 regularization), etc., to identify the most important features, i.e. key features;

[0109] In practical applications, we may combine multiple methods to comprehensively evaluate the importance of features;

[0110] For example: first use feature importance evaluation algorithm and correlation analysis to narrow the feature range, then use PDP and ICE diagrams to obtain how the corresponding features affect the prediction results, and finally use automated feature selection algorithm to identify key features;

[0111] The process of optimizing the model is:

[0112] According to the increased weight Q of the key features each time, the weight of the key features in the model is increased to obtain a new rehabilitation progress prediction model, and S4 is re-executed;

[0113] If the difference in the comparison result does not exceed the previous comparison result, the difference between the previous comparison result difference and the current comparison difference is calculated. When the difference is greater than the preset standard threshold, it does not meet expectations, and S4 is continued until the difference is less than or equal to the preset standard threshold, which meets expectations. The current rehabilitation progress prediction model is used to continue predicting and managing the next patient.

[0114] If the difference in the comparison result exceeds the previous comparison result, the algorithm replacement strategy is triggered to replace the analysis algorithm used to identify the key features until the difference in the comparison result does not exceed the previous comparison result.

[0115] The use of machine learning algorithms can explore the potential patterns in historical rehabilitation data, establish accurate prediction models, and improve the accuracy of predicting patients' fracture rehabilitation progress. By comparing the predicted results with the actual results and fine-tuning the model parameters, the model's predictive performance is continuously optimized. At the same time, specific arrangements are made in the model for the identification and adjustment of key features, thereby further optimizing the prediction model and increasing the accuracy of the model for patients' rehabilitation progress.

[0116] In the application, the several formulas involved are all calculated by taking their numerical values ​​after removing the dimensions, and the formulas are established by collecting a large amount of data and performing software simulation to obtain a formula for the most recent actual situation. Some coefficients or weights in the formulas are set by technical personnel in this field according to actual conditions, so they will not be elaborated here.

[0117] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product. A person of ordinary skill in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented in electronic hardware, or in a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0118] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0119] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A medical management method, characterized in that: The steps include: S1. Use the PACS system to obtain image data of the patient's fracture area, use OCR technology to convert paper record data into electronic format data, and simultaneously obtain pain numerical data, and upload the image data, electronic format data and pain numerical data; S2, receiving the image data, the electronic format data and the pain numerical data, integrating the image data, the electronic format data and the pain numerical data into a database, and unifying the format and standard of the data; S3. Analyze the collected historical rehabilitation data using machine learning algorithms and establish a rehabilitation progress prediction model; S4, inputting the corresponding patient data collected by real-time monitoring into the rehabilitation progress prediction model, and comparing the obtained prediction results with the actual results; S5. Fine-tune the model parameters to optimize the model according to the comparison results until the comparison results meet expectations.

2. A medical management method according to claim 1, characterized in that: Image data include but are not limited to X-rays and MRI scan results; paper records include surgical records and medication usage; pain numerical data are patients' self-ratings of the pain level in their fracture area, ranging from 1 to 10 points, with 1 point indicating the lowest pain value and 10 points indicating the highest pain value.

3. A medical management method according to claim 2, characterized in that: The image data, electronic format data and pain numerical data are integrated into the same database, and the data format and standards are unified by standardizing the data using FHIR.

4. A medical management method according to claim 3, characterized in that: Historical rehabilitation data include, but are not limited to: basic information, fracture location and type, initial injury severity, and total length of time required for rehabilitation; Among them, basic information includes the patient's age and gender; the initial injury severity indicates the hospital's preliminary diagnosis, including the degree of fracture displacement; the total time required for recovery indicates the time from injury to recovery.

5. A medical management method according to claim 4, characterized in that: The process of establishing a rehabilitation progress prediction model is as follows: S301, cleaning historical rehabilitation data; S302, selecting features related to fracture healing time from the cleaned historical rehabilitation data, including age, fracture type, and initial injury severity; S303, selecting an algorithm suitable for time series analysis, including any one of ARIMA and LSTM; S304, training the selected model using historical rehabilitation data, and adjusting model parameters to minimize prediction errors; S305, testing the prediction performance of the model using an independent validation data set; S306. Deploy the trained model to predict the fixed value range of fracture healing time of new patients in real time.

6. A medical management method according to claim 5, characterized in that: The predicted result indicates: the predicted fixed value range of the patient's fracture healing time; the actual result indicates: the actual corresponding patient's fracture healing time.

7. A medical management method according to claim 6, characterized in that: The comparison result represents the difference between the predicted result and the actual result. The model is optimized by fine-tuning the model parameters. The optimization process is as follows: identify the key features, set the weights of the key features in the model based on the comparison results, and limit the weights corresponding to each adjustment of the key features in the following way: Q = α*(|yc-jc|); Where Q represents the added weight of each adjustment of the key features, α represents an adjustment coefficient, and 0<α<1, yc and jc represent the predicted value of the fracture healing time of the corresponding patient and the actual fracture healing time of the corresponding patient, respectively.

8. A medical management method according to claim 7, characterized in that: The process of identifying key features is: using any one of the analysis algorithm set to identify key features, and the analysis algorithm set includes but is not limited to feature importance assessment algorithm, Pearson correlation coefficient, PDP and ICE graph and automatic feature selection algorithm.

9. A medical management method according to claim 8, characterized in that: The process of optimizing the model is: According to the increased weight Q of the key feature each time the key feature is adjusted, the weight of the key feature in the model is increased, a new rehabilitation progress prediction model is obtained, and S4 is re-executed; If the difference in the comparison result does not exceed the previous comparison result, the difference between the previous comparison result difference and the current comparison difference is calculated. When the difference is greater than the preset standard threshold, it does not meet expectations, and S4 is continued until the difference is less than or equal to the preset standard threshold, which meets expectations. The current rehabilitation progress prediction model is used to continue predicting and managing the next patient. If the difference in the comparison result exceeds the previous comparison result, the algorithm replacement strategy is triggered to replace the analysis algorithm used to identify the key features until the difference in the comparison result does not exceed the previous comparison result.

Images