Nuclear magnetic resonance appointment tardiness prediction method and device, and electronic equipment

Abstract

This invention relates to the field of artificial intelligence, specifically to a method, device, system, and storage medium for predicting late arrivals for MRI appointments. The method includes: acquiring MRI examination appointment data; classifying the appointment information according to different attribute characteristics; inputting the classified appointment information and labeled information into an initial late arrival prediction network; training the initial late arrival prediction network to obtain a target late arrival prediction model; after the target late arrival prediction model has been trained, evaluating and analyzing the model weights obtained through training for each type of appointment information to determine the importance score corresponding to the feature attributes included in each type of appointment information; and then using the target late arrival prediction model to predict whether a target patient will be late for their MRI appointment. This allows for the prediction of whether a patient will be late after booking a specific examination, thereby ensuring the normal operation of the medical imaging department and avoiding the waste of medical resources.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence, and specifically to a method, apparatus, and electronic device for predicting late arrivals for MRI appointments. Background Technology

[0002] Modern medical services often rely on appointment systems to ensure that every patient receives the fairest possible service, guarantee the quality of medical services, and improve the patient's medical experience.

[0003] Being late or missing an appointment for some MRI scans can delay patient treatment and disrupt the normal work order of the medical imaging department. Therefore, how to determine whether a patient scheduled for an MRI scan will be late has become an urgent problem to be solved in medical management. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method, device, and electronic device for predicting late arrivals for MRI appointments, aiming to solve the problem of how to determine whether a patient scheduled for an examination will be late.

[0005] According to a first aspect, embodiments of the present invention provide a method for predicting late arrivals for MRI appointments, comprising:

[0006] Obtain MRI examination appointment data, which includes appointment information for multiple historical patients and information indicating whether historical patients were late.

[0007] The appointment information in the MRI examination appointment data is classified according to different attribute characteristics;

[0008] The classified appointment information and labeling information are input into the initial late arrival prediction network, and the initial late arrival prediction network is trained to obtain the target late arrival prediction model.

[0009] After the target late arrival prediction model is trained, the model weights obtained through training are evaluated and analyzed for each type of appointment information to determine the importance score of the feature attributes included in each type of appointment information.

[0010] Obtain the target appointment information for the MRI scan corresponding to the target patient;

[0011] Classify target reservation information according to different attribute characteristics;

[0012] The categorized target appointment information is input into the target late arrival prediction model. The target late arrival prediction model outputs whether the target patient is late and the importance score of each target feature attribute included in the target appointment information.

[0013] Based on the importance scores of each target feature attribute, the lateness prediction model outputs the results of whether the target patient is late.

[0014] The MRI appointment lateness prediction method provided in this invention acquires MRI examination appointment data and classifies the appointment information in the MRI examination appointment data according to different attribute characteristics, ensuring the accuracy of classification. Then, the classified appointment information and annotation information are input into an initial lateness prediction network for training, obtaining a target lateness prediction model, ensuring the accuracy of the obtained target lateness prediction model. After the target lateness prediction model has been trained, the model weights obtained through training for each type of appointment information are evaluated and analyzed to determine the importance score corresponding to the feature attributes included in each type of appointment information, ensuring the accuracy of the determined importance score corresponding to the feature attributes included in each type of appointment information. Then, the target appointment information for the target patient's corresponding MRI is acquired; the target appointment information is classified according to different attribute characteristics, ensuring the accuracy of the classification of the target appointment information. The categorized target appointment information is input into a target lateness prediction model. This model outputs whether a target patient is late, along with importance scores for each target feature attribute included in the appointment information. This ensures the accuracy of both the output lateness prediction result and the importance scores of each target feature attribute. Then, based on the importance scores of each target feature attribute, the lateness prediction result output by the model is evaluated, ensuring the accuracy of this evaluation. Therefore, this method can predict whether a target patient will be late after booking an MRI examination, thus ensuring the normal operation of the medical imaging department and avoiding the waste of medical resources. Furthermore, the evaluation of the lateness prediction result further guarantees its accuracy.

[0015] In conjunction with the first aspect, in the first embodiment of the first aspect, the appointment information includes: the appointment creation time, the appointment time, and the registration time; obtaining MRI examination appointment data includes:

[0016] Obtain the original record data corresponding to each historical patient;

[0017] From the original record data, filter the creation time, appointment time, and registration time corresponding to each historical patient;

[0018] Subtract the corresponding appointment creation time from each appointment time to obtain the waiting time for each historical patient.

[0019] Based on the creation time, appointment time, registration time, waiting time, and whether each historical patient was late, MRI examination appointment data is generated.

[0020] The MRI appointment lateness prediction method provided in this invention includes appointment information such as appointment creation time, appointment time, and registration time. It obtains original record data for each historical patient and filters the appointment creation time, appointment time, and registration time for each historical patient from the original record data, ensuring the accuracy of the filtered appointment creation time, appointment time, and registration time. Then, it subtracts the corresponding appointment creation time from each appointment time to obtain the waiting time for each historical patient, ensuring the accuracy of the calculated waiting time. Based on the appointment creation time, appointment time, registration time, waiting time, and lateness information for each historical patient, MRI examination appointment data is generated, ensuring the accuracy of the generated MRI examination appointment data. This, in turn, ensures the accuracy of the target lateness prediction model trained using the MRI examination appointment data.

[0021] In conjunction with the first implementation method of the first aspect, in the second implementation method of the first aspect, MRI examination appointment data is generated based on the appointment creation time, appointment time, registration time, waiting time, and whether each historical patient was late, according to the corresponding information of each historical patient. This includes:

[0022] The appointment creation time is converted to a time value to determine the corresponding quarterly, weekly, and hourly values ​​for the appointment creation time.

[0023] Convert the appointment time into time values ​​to determine the corresponding quarterly, weekly, and hourly values.

[0024] The registration time is converted to a time value to determine the corresponding quarter value, week value, and pre-registration hour value.

[0025] Based on the quarterly, weekly, and hourly values ​​of appointments created for each historical patient, the MRI examination appointment data is generated.

[0026] The MRI appointment lateness prediction method provided in this invention performs time conversion on the appointment creation time to determine the corresponding quarterly, weekly, and hourly values ​​for the appointment, ensuring the accuracy of these values. Similarly, it performs time conversion on the appointment time to determine the corresponding quarterly, weekly, and hourly values ​​for registration, ensuring the accuracy of these values ​​as well. Then, based on the quarterly, weekly, and hourly values ​​of appointments created for each historical patient, as well as the registration quarterly, weekly, and hourly values, waiting times, and whether each historical patient was late, MRI examination appointment data is generated. This ensures the accuracy of the generated MRI examination appointment data, which in turn ensures the accuracy of the target lateness prediction model trained using the MRI examination appointment data.

[0027] In conjunction with the second implementation method of the first aspect, in the third implementation method of the first aspect, the appointment information also includes patient type, which includes inpatients and outpatients. Based on the creation appointment quarter value, creation appointment week value, creation appointment hour value, appointment quarter value, appointment week value, appointment hour value, registration quarter value, registration week value, pre-registration hour value, waiting time, and whether each historical patient was late, MRI examination appointment data is generated, including:

[0028] Obtain the medical record number or patient card number corresponding to each historical patient;

[0029] Identify medical record numbers or patient card numbers to determine the patient type corresponding to each historical patient;

[0030] Based on the patient type corresponding to each historical patient, the following information is generated: appointment quarter value, appointment week value, appointment hour value, appointment quarter value, appointment week value, appointment hour value, registration quarter value, registration week value, pre-registration hour value, waiting time, and whether each historical patient was late.

[0031] The MRI appointment lateness prediction method provided in this invention obtains the medical record number or patient card number corresponding to each historical patient, identifies the medical record number or patient card number, and determines the patient type corresponding to each historical patient, ensuring the accuracy of the determined patient type. Then, based on the patient type corresponding to each historical patient, the created appointment quarter value, created appointment week value, created appointment hour value, the appointment quarter value, appointment week value, appointment hour value, registration quarter value, registration week value, pre-registration hour value, waiting time, and whether each historical patient was late, MRI examination appointment data is generated, ensuring the accuracy of the generated MRI examination appointment data, and thus ensuring the accuracy of the target lateness prediction model trained using the MRI examination appointment data.

[0032] In conjunction with the third implementation method of the first aspect, the fourth implementation method of the first aspect further includes the following appointment information: reimbursement type, patient gender, patient age, examination equipment, imaging sequence description, requesting department, examination fee, and registrar.

[0033] The MRI appointment lateness prediction method provided in this embodiment of the invention includes appointment information such as: reimbursement type, patient gender, patient age, examination equipment, imaging sequence description, requesting department, examination fee, and registrar. This ensures the accuracy of the generated MRI examination appointment data, and thus guarantees the accuracy of the target lateness prediction model trained using the MRI examination appointment data.

[0034] In conjunction with the first aspect, in the fifth embodiment of the first aspect, the evaluation of whether a target patient is late, as output by the target lateness prediction model, is based on the importance scores of each target feature attribute, including:

[0035] Obtain the preset importance level corresponding to each target feature attribute;

[0036] Compare the preset importance level and importance score corresponding to each target feature attribute;

[0037] When the preset importance level corresponding to each target feature attribute matches the importance score, the result of whether the target patient is late, as output by the target lateness prediction model, is determined to be an accurate result.

[0038] The MRI appointment lateness prediction method provided in this invention obtains a preset importance level corresponding to each target feature attribute, compares the preset importance level of each target feature attribute with its importance score, and ensures the accuracy of the comparison results. When the preset importance level of each target feature attribute matches the importance score, the lateness prediction model output is determined to be an accurate result for the target patient, thus ensuring the accuracy of the determined lateness prediction model output.

[0039] In conjunction with the fifth embodiment of the first aspect, in the sixth embodiment of the first aspect, the method further includes:

[0040] When the preset importance level and importance score corresponding to each target feature attribute do not match, the result of whether the target patient is late is determined to be an inaccurate result by the target lateness prediction model.

[0041] The target late arrival prediction model was trained again using MRI scan appointment data.

[0042] The MRI appointment lateness prediction method provided in this invention determines that the lateness prediction model output is inaccurate when the preset importance level and importance score of each target feature attribute do not match. This ensures the accuracy of the determination that the lateness prediction model output is inaccurate. Then, the lateness prediction model is retrained using MRI appointment data to ensure the accuracy of the retrained model, thereby ensuring the accuracy of predicting whether a target patient will be late after scheduling an MRI examination.

[0043] According to a second aspect, embodiments of the present invention also provide a magnetic resonance imaging (MRI) appointment lateness prediction device, comprising:

[0044] The first acquisition module is used to acquire MRI examination appointment data, which includes appointment information for multiple historical patients and information indicating whether the historical patients were late.

[0045] The first classification module is used to classify the appointment information in the MRI examination appointment data according to different attribute characteristics;

[0046] The training module is used to input the classified reservation information and annotation information into the initial late arrival prediction network, train the initial late arrival prediction network, and obtain the target late arrival prediction model.

[0047] The analysis module is used to evaluate and analyze the model weights obtained through training for each type of reservation information after the target late arrival prediction model has been trained, and to determine the importance score of the feature attributes included in each type of reservation information.

[0048] The second acquisition module is used to acquire the target appointment information for the scheduled MRI of the target patient.

[0049] The second classification module is used to classify target reservation information according to different attribute characteristics;

[0050] The first output module is used to input the classified target appointment information into the target late arrival prediction model. The target late arrival prediction model outputs whether the target patient is late and outputs the importance score of each target feature attribute included in the target appointment information.

[0051] The second output module is used to evaluate whether the target patient is late based on the importance scores of each target feature attribute.

[0052] The MRI appointment lateness prediction device provided in this invention acquires MRI examination appointment data and classifies the appointment information in the MRI examination appointment data according to different attribute characteristics, ensuring the accuracy of classification. Then, the classified appointment information and annotation information are input into an initial lateness prediction network for training, obtaining a target lateness prediction model, ensuring the accuracy of the obtained target lateness prediction model. After the target lateness prediction model completes training, the model weights obtained through training for each type of appointment information are evaluated and analyzed to determine the importance score corresponding to the feature attributes included in each type of appointment information, ensuring the accuracy of the determined importance score corresponding to the feature attributes included in each type of appointment information. Then, the target appointment information for the target patient's scheduled MRI is acquired; the target appointment information is classified according to different attribute characteristics, ensuring the accuracy of the classification of the target appointment information. The categorized target appointment information is input into a target lateness prediction model. This model outputs whether a target patient will be late, along with importance scores for each target feature attribute included in the appointment information. This ensures the accuracy of both the output lateness prediction result and the importance scores of each target feature attribute. Then, based on the importance scores of each target feature attribute, the lateness prediction result output by the model is evaluated, ensuring the accuracy of this evaluation. Therefore, this device can predict whether a target patient will be late after booking an MRI examination, thus ensuring the normal operation of the medical imaging department and avoiding waste of medical resources. Furthermore, the evaluation of the lateness prediction result further guarantees its accuracy.

[0053] According to a third aspect, embodiments of the present invention provide an electronic device, including a memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to perform the MRI appointment lateness prediction method in the first aspect or any embodiment of the first aspect.

[0054] According to a fourth aspect, embodiments of the present invention provide a computer-readable storage medium storing computer instructions for causing a computer to perform the MRI appointment lateness prediction method in the first aspect or any embodiment of the first aspect. Attached Figure Description

[0055] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0056] Figure 1 This is a flowchart of the method for predicting late appointments for nuclear magnetic resonance imaging (NMR) provided in the embodiments of the present invention;

[0057] Figure 2 This is a schematic diagram of a confusion matrix provided by another embodiment of the present invention;

[0058] Figure 3 This is a schematic diagram of the process of training a target lateness prediction model using training data, provided by another embodiment of the present invention.

[0059] Figure 4 This is a schematic diagram illustrating the importance scores corresponding to the feature attributes provided in another embodiment of the present invention;

[0060] Figure 5 This is a schematic diagram illustrating the accuracy of the target lateness prediction model provided by another embodiment of the present invention;

[0061] Figure 6 This is a flowchart of a method for predicting late appointments for nuclear magnetic resonance imaging (NMR) provided by another embodiment of the present invention;

[0062] Figure 7 This is a flowchart of a method for predicting late appointments for nuclear magnetic resonance imaging (NMR) provided by another embodiment of the present invention;

[0063] Figure 8 This is a functional block diagram of the nuclear magnetic resonance imaging appointment lateness prediction device provided in the embodiments of the present invention;

[0064] Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of the present invention. Detailed Implementation

[0065] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0066] It should be noted that the method for predicting late MRI appointments provided in this application can be executed by a device for predicting late MRI appointments. This device can be implemented as part or all of a computer device through software, hardware, or a combination of both. The computer device can be a server or a terminal. In this application embodiment, the server can be a single server or a server cluster composed of multiple servers. The terminal in this application embodiment can be a smartphone, personal computer, tablet computer, wearable device, or other intelligent hardware device such as an intelligent robot. The following method embodiments will use an electronic device as an example for illustration.

[0067] In one embodiment of this application, such as Figure 1 As shown, a method for predicting late arrivals for MRI appointments is provided. Taking the application of this method to electronic devices as an example, the method includes the following steps:

[0068] S11. Obtain MRI examination appointment data.

[0069] The MRI appointment data includes appointment information for multiple historical patients and information indicating whether these patients were late.

[0070] Optionally, the electronic device can receive MRI examination appointment data input by the user; the electronic device can also receive MRI examination appointment data sent by other devices; the electronic device can also obtain MRI examination appointment data from an actual clinical RIS system.

[0071] This application does not specifically limit the method by which electronic devices acquire MRI examination appointment data.

[0072] This step will be explained in detail below.

[0073] S12. Classify the appointment information in the MRI examination appointment data according to different attribute characteristics.

[0074] Specifically, electronic devices can classify appointment information in MRI examination appointment data according to different attribute characteristics, which can be divided into category type features and continuous type features.

[0075] For example, Category (CAT) type features may include: reimbursement type, patient gender, MR equipment, imaging sequence description, requesting department, registrar, appointment date, appointment quarter, appointment hour, registration date, registration quarter, registration hour, appointment creation date, appointment creation quarter, and appointment creation hour; Continuous (CONT) type features may include: waiting time, patient age, and examination fee.

[0076] S13. Input the classified reservation information and labeling information into the initial late arrival prediction network, train the initial late arrival prediction network, and obtain the target late arrival prediction model.

[0077] Specifically, after classifying the appointment information in the MRI examination appointment data according to different attribute characteristics, the electronic device can input the classified appointment information and annotation information into the initial lateness prediction network. The network extracts features from the appointment information of multiple historical patients who have made MRI appointments and the annotation information on whether historical patients were late. The initial lateness prediction network is then trained based on the extracted features to obtain the target lateness prediction model.

[0078] The initial late arrival prediction network can be one of the following algorithms: XGBOOST algorithm, decision tree algorithm, SVM algorithm, ID3 algorithm, C4.5 algorithm, C5.0 algorithm, KNN algorithm, and ANN algorithm. This application does not impose specific limitations on the initial late arrival prediction network.

[0079] S14. After the target late arrival prediction model has been trained, evaluate and analyze the model weights obtained through training for each type of appointment information, and determine the importance score corresponding to the feature attributes included in each type of appointment information.

[0080] Specifically, after training the target late arrival prediction model, the electronic device can use the initial late arrival prediction network to evaluate and analyze the relationship between the appointment information and annotation information corresponding to each historical patient in the MRI examination appointment data, and to determine the importance score corresponding to the feature attributes included in each type of appointment information by using the model weights obtained through training.

[0081] During the training of the initial lateness prediction network using MRI scan appointment data, such as Figure 2 As shown, a confusion matrix can be generated, which is used to evaluate the accuracy of the target late arrival prediction model. The confusion matrix includes both the true and predicted values, and the accuracy of the target late arrival prediction model is then evaluated based on these values. The closer the predicted value is to the true value, the higher the accuracy of the target late arrival prediction model.

[0082] For example, such as Figure 3 As shown, the classified appointment information and annotation information can be input into the XGBOOST algorithm. During the training of the target late arrival prediction model, the model weights obtained through training for each type of appointment information are evaluated and analyzed for the appointment information and annotation information corresponding to each historical patient in the MRI examination appointment data. The importance score corresponding to the feature attributes included in each type of appointment information is determined, and then the late arrival result corresponding to each historical patient is output.

[0083] For example, such as Figure 4 The figure shown can be represented as the importance score of the feature attributes corresponding to a certain historical patient.

[0084] For example, such as Figure 5 The figure shows the accuracy of the target late arrival prediction model. The true positive rate represents the accuracy of the target late arrival prediction model, and the false positive rate represents the error rate of the target late arrival prediction model.

[0085] S15. Obtain the target appointment information for the MRI scan corresponding to the target patient.

[0086] Specifically, the electronic device can receive user input of target appointment information for the MRI scan corresponding to the target patient; the electronic device can also receive target appointment information for the MRI scan corresponding to the target patient sent by other devices; the electronic device can also obtain target record data for the MRI scan corresponding to the target patient from the actual clinical RIS system, and then filter the target record data for the target patient based on reimbursement type, patient gender, patient age, examination equipment, imaging sequence description, requesting department, examination fee, registrar, appointment creation time, appointment time, and registration time. Then, by subtracting the appointment time from the appointment creation time, the waiting time for the target patient is obtained, and time conversion is performed on the appointment creation time, appointment time, and registration time to obtain the quarterly value, weekly value, hourly value, quarterly value, weekly value, hourly value, quarterly value, weekly value, and pre-registration hourly value for the target patient. Then, based on the reimbursement type, patient gender, patient age, examination equipment, imaging sequence description, requesting department, examination fee, registrar, waiting time, created appointment quarterly value, created appointment weekly value, created appointment hourly value, and appointment hourly value, registered quarterly value, registered weekly value, and pre-registered hourly value, the target appointment information for the target patient's MRI is generated.

[0087] S16. Classify the target reservation information according to different attribute characteristics.

[0088] Specifically, electronic devices can classify target reservation information according to different attribute characteristics, which can be divided into category type features and continuous type features.

[0089] For example, Category (CAT) type features may include: reimbursement type, patient gender, MR equipment, imaging sequence description, requesting department, registrar, appointment date, appointment quarter, appointment hour, registration date, registration quarter, registration hour, appointment creation date, appointment creation quarter, and appointment creation hour; Continuous (CONT) type features may include: waiting time, patient age, and examination fee.

[0090] S17. Input the classified target appointment information into the target late arrival prediction model. The target late arrival prediction model outputs whether the target patient is late and the importance score of each target feature attribute included in the target appointment information.

[0091] Specifically, after classifying the target appointment information according to different attribute characteristics, the electronic device can input the classified target appointment information into the target late arrival prediction model. The target late arrival prediction model extracts features from the target appointment information, outputs whether the target patient is late based on the extracted features, and outputs the importance score of each target feature attribute included in the target appointment information.

[0092] S18. Based on the importance scores of each target feature attribute, evaluate whether the target patient is late as output by the target lateness prediction model.

[0093] Specifically, the electronic device can acquire the preset importance levels of each target feature attribute corresponding to the target patient. Then, based on the relationship between the importance scores of each target feature attribute and the corresponding importance levels, it evaluates whether the target patient is late, as output by the target lateness prediction model.

[0094] This step will be explained in detail below.

[0095] The MRI appointment lateness prediction method provided in this invention acquires MRI examination appointment data and classifies the appointment information in the MRI examination appointment data according to different attribute characteristics, ensuring the accuracy of classification. Then, the classified appointment information and annotation information are input into an initial lateness prediction network for training, obtaining a target lateness prediction model, ensuring the accuracy of the obtained target lateness prediction model. After the target lateness prediction model has been trained, the model weights obtained through training for each type of appointment information are evaluated and analyzed to determine the importance score corresponding to the feature attributes included in each type of appointment information, ensuring the accuracy of the determined importance score corresponding to the feature attributes included in each type of appointment information. Then, the target appointment information for the target patient's corresponding MRI is acquired; the target appointment information is classified according to different attribute characteristics, ensuring the accuracy of the classification of the target appointment information. The categorized target appointment information is input into a target lateness prediction model. This model outputs whether a target patient is late, along with importance scores for each target feature attribute included in the appointment information. This ensures the accuracy of both the output lateness prediction result and the importance scores of each target feature attribute. Then, based on the importance scores of each target feature attribute, the lateness prediction result output by the model is evaluated, ensuring the accuracy of this evaluation. Therefore, this method can predict whether a target patient will be late after booking an MRI examination, thus ensuring the normal operation of the medical imaging department and avoiding the waste of medical resources. Furthermore, the evaluation of the lateness prediction result further guarantees its accuracy.

[0096] In one embodiment of this application, such as Figure 6 As shown, a method for predicting late arrivals for MRI appointments is provided. Taking the application of this method to electronic devices as an example, the method includes the following steps:

[0097] S21. Obtain MRI examination appointment data.

[0098] The MRI appointment data includes appointment information for multiple historical patients and information indicating whether these patients were late.

[0099] In an optional embodiment of this application, step S21, "acquiring MRI examination appointment data," may include the following steps:

[0100] S211. Obtain the original record data corresponding to each historical patient.

[0101] Optionally, the electronic device can receive the original record data corresponding to each historical patient input by the user; the electronic device can also receive the original record data corresponding to each historical patient sent by other devices; the electronic device can also obtain the original record data corresponding to each historical patient from the actual clinical RIS system.

[0102] The embodiments of the present application do not specifically limit the manner in which the electronic device obtains the original record data corresponding to each historical patient.

[0103] Exemplarily, the original data may include information such as the appointment time, creation appointment time, patient transfer registration time, patient ID, examination serial number, medical treatment card or medical insurance card number, patient gender, age, examination equipment model, examination item, application department, examination price, contact phone number, and registrar corresponding to each historical patient. The embodiments of the present application do not specifically limit the original data.

[0104] S212. Screen the creation appointment time, appointment time, and registration time corresponding to each historical patient from the original record data.

[0105] Specifically, the electronic device can identify the original record data, and then screen the creation appointment time, appointment time, and registration time corresponding to each historical patient from the various types of data included in the identified original record data.

[0106] Among them, the creation appointment time refers to the time when the historical patient first came to the registration room to create this examination, the appointment time refers to the date and time period of this examination appointment for the historical patient, and the registration time refers to the time when the historical patient actually arrived at the department and came to the registration room for registration.

[0107] Then, change all the Chinese content to English and remove all data units, such as "years old" in age and "year-month-day hour-minute" in time.

[0108] Since the examination items corresponding to different patients are different, there are a variety of examination item types. The examination items can be represented by the first English letter, and a simple inductive classification of the examination items is carried out according to medical background knowledge. Exemplarily, a simple inductive classification of each examination item can be carried out according to the detection site.

[0109] Then, make a dictionary for data conversion and mark the types of such data in the dictionary, mainly including category, integer, time, days, etc., to facilitate subsequent algorithm training.

[0110] S213. Subtract the corresponding appointment creation time from each appointment time to obtain the waiting time for each historical patient.

[0111] Specifically, electronic devices can subtract the corresponding appointment creation time from each appointment time to obtain the waiting time for each historical patient.

[0112] S214. Generate MRI examination appointment data based on the creation appointment time, appointment time, registration time, waiting time, and whether each historical patient was late.

[0113] In an optional embodiment of this application, the above-mentioned step S214, "generating MRI examination appointment data based on the appointment creation time, appointment time, registration time, waiting time, and whether each historical patient was late," may include the following steps:

[0114] (1) Convert the creation appointment time to a time value to determine the corresponding quarterly value, weekly value, and hourly value for the creation appointment time.

[0115] Specifically, electronic devices can perform time conversion on the appointment creation time to determine the corresponding quarterly value, weekly value, and hourly value for the appointment creation time.

[0116] The appointment time is divided into 5 time periods according to the 24-hour system (Time Period 1: 00:00-7:59, Time Period 2: 8:00-11:59, Time Period 3: 12:00-13:59, Time Period 4: 14:00-17:59, Time Period 5: 18:00-24:00).

[0117] For example, if the preset creation time is May 17, 2022, at 10:30, then the creation appointment time corresponds to the second quarter, the weekday of the creation appointment is 3, and the hour of the creation appointment is the second time period.

[0118] (2) Convert the reservation time to a time value to determine the corresponding quarterly value, weekday value, and hourly value.

[0119] Specifically, electronic devices can perform time conversion on the reservation time to determine the corresponding reservation quarter value, reservation week value, and reservation hour value.

[0120] (3) Convert the registration time to determine the corresponding quarter value, week value, and pre-registration hour value.

[0121] Specifically, electronic devices can perform time conversion on the registration time to determine the corresponding registration quarter value, registration week value, and pre-registration hour value.

[0122] (4) Generate MRI examination appointment data based on the quarterly value of appointment creation, weekly value of appointment creation, hourly value of appointment creation, quarterly value of appointment creation, weekly value of appointment creation, hourly value of appointment creation, quarterly value of registration, weekly value of registration, hourly value of pre-registration, waiting time, and whether each historical patient was late.

[0123] Specifically, electronic devices can generate MRI examination appointment data based on the quarterly, weekly, and hourly values ​​of appointments created for each historical patient, as well as the registration quarterly, weekly, and hourly values, pre-registration hourly values, waiting times, and whether each historical patient was late.

[0124] In one optional embodiment of this application, the appointment information also includes patient type, which includes inpatients and outpatients. The above step (4) "generating MRI examination appointment data based on the creation appointment quarter value, creation appointment week value, creation appointment hour value, appointment quarter value, appointment week value, appointment hour value, registration quarter value, registration week value, pre-registration hour value, waiting time, and whether each historical patient is late" may also include the following steps:

[0125] (41) Obtain the medical record number or patient card number corresponding to each historical patient.

[0126] Optionally, the electronic device can filter out the medical record number or patient card number corresponding to each historical patient from the original record data corresponding to each historical patient.

[0127] Optionally, the electronic device can also receive the medical record number or patient card number corresponding to each historical patient entered by the user, and the electronic device can also receive the medical record number or patient card number corresponding to each historical patient sent by other devices.

[0128] This application does not specifically limit the method by which electronic devices obtain the medical record number or patient card number corresponding to each historical patient.

[0129] (42) Identify the medical record number or patient card number to determine the patient type corresponding to each historical patient.

[0130] Specifically, electronic devices can use text recognition algorithms to identify medical record numbers or patient card numbers, and then determine the patient type corresponding to each historical patient based on the recognition results.

[0131] The text recognition algorithm can be based on handcrafted features, such as DPM (Deformable Parts Model), or it can be based on convolutional neural networks, such as the YOLO (You Only Look Once) detector, R-CNN (Region-based Convolutional Neural Networks) model, SSD (Single Shot MultiBox) detector, and Mask R-CNN (Mask Region-based Convolutional Neural Networks) model. This application does not specifically limit the text recognition algorithm used.

[0132] (43) Generate MRI examination appointment data based on the patient type corresponding to each historical patient, the creation of appointment quarterly value, appointment weekly value, appointment hourly value, appointment quarterly value, appointment weekly value, appointment hourly value, registration quarterly value, registration weekly value, pre-registration hourly value, waiting time, and whether each historical patient was late.

[0133] Specifically, electronic devices can generate MRI examination appointment data based on the patient type corresponding to each historical patient, the creation of appointment quarterly values, the creation of appointment weekly values, the creation of appointment hourly values, the appointment quarterly values, the appointment weekly values, the appointment hourly values, the registration quarterly values, the registration weekly values, the pre-registration hourly values, the waiting time, and the labeling information of whether each historical patient was late.

[0134] In one optional implementation of this application, the appointment information further includes: reimbursement type, patient gender, patient age, examination equipment, imaging sequence description, requesting department, examination fee, and registrar.

[0135] Specifically, electronic devices can identify the original record data and filter out the reimbursement type, patient gender, patient age, examination equipment, imaging sequence description, requesting department, examination fee, and registrar for each historical patient from the original record data.

[0136] Then, based on the reimbursement type, patient gender, patient age, examination equipment, imaging sequence description, requesting department, examination fee, registrar, patient type, creation of appointment quarterly value, creation of appointment weekly value, creation of appointment hourly value, appointment quarterly value, appointment weekly value, appointment hourly value, registration quarterly value, registration weekly value, pre-registration hourly value, waiting time, and whether each historical patient was late, MRI examination appointment data is generated.

[0137] Then, the electronic device categorizes the various feature attributes included in the appointment information corresponding to each historical patient. The categorical (CAT) features include: reimbursement type, patient age, patient gender, MR equipment, imaging sequence description, requesting department, registrar, appointment date, appointment quarter, appointment hour, registration date, registration quarter, registration hour, appointment creation date, appointment creation quarter, and appointment creation hour. The continuous (CONT) features include: waiting time, age, and examination fee. The electronic device inputs the categorical (CAT) and continuous (CONT) features into the initial late arrival prediction network. During the training of the target late arrival prediction model, the relationship between the appointment information and the labeled information corresponding to each historical patient in the MRI examination appointment data is analyzed to determine the importance score of the feature attributes included in each appointment information. Finally, the system outputs the late arrival result for each historical patient.

[0138] The MRI appointment lateness prediction method provided in this invention obtains the original record data corresponding to each historical patient. From the original record data, it filters the appointment creation time, appointment time, and registration time corresponding to each historical patient, ensuring the accuracy of the filtered appointment creation time, appointment time, and registration time. Then, it subtracts the corresponding appointment creation time from each appointment time to obtain the waiting time corresponding to each historical patient, ensuring the accuracy of the calculated waiting time. Next, it performs time conversion on the appointment creation time to determine the corresponding appointment quarter value, appointment week value, and appointment hour value, ensuring the accuracy of the determined appointment quarter value, appointment week value, and appointment hour value. It also performs time conversion on the appointment time to determine the corresponding appointment quarter value, appointment week value, and appointment hour value, ensuring the accuracy of the determined appointment quarter value, appointment week value, and appointment hour value. Finally, it performs time conversion on the registration time to determine the corresponding registration quarter value, registration week value, and pre-registration hour value, ensuring the accuracy of the determined registration quarter value, registration week value, and pre-registration hour value. Then, based on the creation appointment quarter value, creation appointment week value, creation appointment hour value, appointment quarter value, appointment week value, appointment hour value, registration quarter value, registration week value, pre-registration hour value, waiting time, and whether each historical patient was late, MRI examination appointment data is generated. This ensures the accuracy of the generated MRI examination appointment data, thereby ensuring the accuracy of the target lateness prediction model trained using the MRI examination appointment data. Next, the medical record number or patient card number corresponding to each historical patient is obtained, and the patient type corresponding to each historical patient is identified, ensuring the accuracy of the determined patient type. Then, based on the patient type corresponding to each historical patient, the following information is generated: appointment quarter value, appointment week value, appointment hour value, appointment quarter value, appointment week value, appointment hour value, registration quarter value, registration week value, pre-registration hour value, waiting time, and whether each historical patient was late. This ensures the accuracy of the generated MRI examination appointment data, which in turn ensures the accuracy of the target lateness prediction model trained using the MRI examination appointment data.

[0139] In addition, the appointment information also includes: reimbursement type, patient gender, patient age, examination equipment, imaging sequence description, requesting department, examination fee, and registrar, ensuring the accuracy of the generated MRI examination appointment data, which in turn ensures the accuracy of the target lateness prediction model trained using the MRI examination appointment data.

[0140] In one embodiment of this application, such as Figure 7 As shown, a method for predicting late arrivals for MRI appointments is provided. Taking the application of this method to electronic devices as an example, the method includes the following steps:

[0141] S31. Obtain MRI examination appointment data.

[0142] The MRI appointment data includes appointment information for multiple historical patients and information indicating whether these patients were late.

[0143] For details on this step, please refer to [link / reference]. Figure 6 The details of S21 will not be elaborated here.

[0144] S32. Classify the appointment information in the MRI examination appointment data according to different attribute characteristics.

[0145] For details on this step, please refer to [link / reference]. Figure 6 The details of S22 will not be elaborated here.

[0146] S33. Input the classified reservation information and labeling information into the initial late arrival prediction network, train the initial late arrival prediction network, and obtain the target late arrival prediction model.

[0147] For details on this step, please refer to [link / reference]. Figure 6 The details of S23 will not be elaborated here.

[0148] S34. After the target late arrival prediction model has been trained, evaluate and analyze the model weights obtained through training for each type of reservation information, and determine the importance score corresponding to the feature attributes included in each type of reservation information.

[0149] For details on this step, please refer to [link / reference]. Figure 6 The details of S24 will not be elaborated here.

[0150] S35. Obtain the target appointment information for the MRI scan corresponding to the target patient.

[0151] For details on this step, please refer to [link / reference]. Figure 6 The details of the S25 will not be elaborated here.

[0152] S36. Classify the target reservation information according to different attribute characteristics.

[0153] For details on this step, please refer to [link / reference]. Figure 2 The details of S26 will not be elaborated here.

[0154] S37. Input the classified target appointment information into the target late arrival prediction model. The target late arrival prediction model outputs whether the target patient is late and outputs the importance score of each target feature attribute included in the target appointment information.

[0155] For details on this step, please refer to [link / reference]. Figure 6 The details of the S27 will not be elaborated here.

[0156] S38. Based on the importance scores of each target feature attribute, evaluate whether the target patient is late according to the output of the target lateness prediction model.

[0157] In an optional embodiment of this application, step S38, "evaluating whether the target patient is late based on the importance score of each target feature attribute," may include the following steps:

[0158] S381. Obtain the preset importance level corresponding to each target feature attribute.

[0159] Specifically, the electronic device can receive preset importance levels corresponding to each target feature attribute input by the user, and can also receive preset importance levels corresponding to each target feature attribute sent by other devices. The electronic device can also determine the preset importance level corresponding to the target feature attribute based on the importance score corresponding to the feature attribute included in each type of reservation information determined after the target late arrival prediction model has been trained.

[0160] This application does not specifically limit the method by which the electronic device obtains the preset importance level corresponding to the target feature attribute.

[0161] S382. Compare the preset importance level and importance score corresponding to each target feature attribute.

[0162] Specifically, after obtaining the preset importance level corresponding to the target feature attribute, the electronic device can compare the importance score of each target feature attribute output by the target lateness prediction model with the preset importance level corresponding to each target feature attribute.

[0163] S383. When the preset importance level corresponding to each target feature attribute matches the importance score, the result of whether the target patient is late, output by the target lateness prediction model, is determined to be the accurate result.

[0164] Specifically, when the preset importance level corresponding to each target feature attribute matches the importance score, the result of whether the target patient is late, as output by the target lateness prediction model, is determined to be an accurate result.

[0165] For example, when the importance score corresponding to one of the target feature attributes is 90 points and the preset importance level corresponding to the target feature attribute is level one, the electronic device determines that the preset importance level corresponding to each target feature attribute matches the importance score, and determines that the result of whether the target patient is late, output by the target lateness prediction model, is an accurate result.

[0166] S384. When the preset importance level and importance score corresponding to each target feature attribute do not match, the result of whether the target patient is late is determined to be an inaccurate result by the target lateness prediction model.

[0167] Specifically, when the preset importance level and importance score corresponding to each target feature attribute do not match, the result of whether the target patient is late is determined to be inaccurate by the target lateness prediction model.

[0168] For example, when the importance score corresponding to one of the target feature attributes is 58 points and the preset importance level corresponding to the target feature attribute is level one, the electronic device determines that the preset importance level and importance score corresponding to each target feature attribute do not match, and determines that the result of whether the target patient is late, output by the target lateness prediction model, is an inaccurate result.

[0169] S385. The target late arrival prediction model is trained again using MRI scan appointment data.

[0170] Specifically, when it is determined that the lateness prediction model outputs an inaccurate result regarding whether the target patient is late, the electronic device can retrain the lateness prediction model using MRI examination appointment data.

[0171] The MRI appointment lateness prediction method provided in this invention obtains a preset importance level corresponding to each target feature attribute, and compares the preset importance level of each target feature attribute with the importance score to ensure the accuracy of the comparison results. When the preset importance level of each target feature attribute matches the importance score, the lateness prediction model output is determined to be an accurate result, ensuring the accuracy of the determined lateness prediction model output. When the preset importance level of each target feature attribute does not match the importance score, the lateness prediction model output is determined to be an inaccurate result, ensuring the accuracy of the determined lateness prediction model output. Then, the lateness prediction model is trained again using MRI examination appointment data, which ensures the accuracy of the retrained model and, consequently, the accuracy of predicting whether a target patient will have an MRI examination after the retraining.

[0172] It should be understood that, although Figure 1 as well as Figure 6-7 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 as well as Figure 6-7 At least some of the steps in the process may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but may be executed at different times. The execution order of these steps or stages is not necessarily sequential, but may be executed in turn or alternately with other steps or at least some of the steps or stages in other steps.

[0173] like Figure 8 As shown, this embodiment provides a device for predicting late arrivals for MRI appointments, including:

[0174] The first acquisition module 41 is used to acquire MRI examination appointment data, which includes appointment information for multiple historical patients and information indicating whether historical patients were late.

[0175] The first classification module 42 is used to classify the appointment information in the MRI examination appointment data according to different attribute characteristics;

[0176] Training module 43 is used to input the classified reservation information and annotation information into the initial late arrival prediction network, train the initial late arrival prediction network, and obtain the target late arrival prediction model.

[0177] Analysis module 44 is used to evaluate and analyze the model weights obtained through training for each type of reservation information after the target late arrival prediction model has been trained, and to determine the importance score corresponding to the feature attributes included in each type of reservation information.

[0178] The second acquisition module 45 is used to acquire the target appointment information of the target patient for the scheduled MRI.

[0179] The second classification module 46 is used to classify target reservation information according to different attribute characteristics;

[0180] The first output module 47 is used to input the classified target appointment information into the target late arrival prediction model. The target late arrival prediction model outputs whether the target patient is late and outputs the importance score of each target feature attribute included in the target appointment information.

[0181] The second output module 48 is used to evaluate whether the target patient is late based on the importance scores of each target feature attribute.

[0182] In one embodiment of this application, the appointment information includes: appointment creation time, appointment time, and registration time. The first acquisition module 41 is specifically used to acquire the original record data corresponding to each historical patient; filter the appointment creation time, appointment time, and registration time corresponding to each historical patient from the original record data; subtract the corresponding appointment creation time from each appointment time to obtain the waiting time corresponding to each historical patient; and generate MRI examination appointment data based on the appointment creation time, appointment time, registration time, waiting time, and whether each historical patient was late.

[0183] In one embodiment of this application, the first acquisition module 41 is specifically used to perform time conversion on the appointment creation time to determine the appointment creation quarter value, appointment creation week value, and appointment creation hour value corresponding to the appointment creation time; perform time conversion on the appointment time to determine the appointment quarter value, appointment week value, and appointment hour value corresponding to the appointment time; perform time conversion on the registration time to determine the registration quarter value, registration week value, and pre-registration hour value corresponding to the registration time; and generate MRI examination appointment data based on the appointment creation quarter value, appointment creation week value, appointment creation hour value, appointment quarter value, appointment week value, appointment hour value, registration quarter value, registration week value, pre-registration hour value, waiting time, and whether each historical patient was late.

[0184] In one embodiment of this application, the appointment information also includes patient types, including inpatients and outpatients. The first acquisition module 41 is specifically used to acquire the medical record number or medical card number corresponding to each historical patient; identify the medical record number or medical card number to determine the patient type corresponding to each historical patient; and generate MRI examination appointment data based on the patient type corresponding to each historical patient, the created appointment quarter value, created appointment week value, created appointment hour value, appointment quarter value, appointment week value, appointment hour value, registered quarter value, registered week value, pre-registered hour value, waiting time, and whether each historical patient was late.

[0185] In one embodiment of this application, the appointment information further includes: reimbursement type, patient gender, patient age, examination equipment, imaging sequence description, requesting department, examination fee, and registrar.

[0186] In one embodiment of this application, the second output module 48 is specifically used to obtain the preset importance level corresponding to each target feature attribute; compare the preset importance level corresponding to each target feature attribute with the importance score; and when the preset importance level corresponding to each target feature attribute matches the importance score, determine that the result of whether the target patient is late, output by the target lateness prediction model, is an accurate result.

[0187] In one embodiment of this application, the second output module 48 is specifically used to determine that the result of whether the target patient is late is an inaccurate result when the preset importance level and importance score corresponding to each target feature attribute do not match; and to retrain the target lateness prediction model using MRI examination appointment data.

[0188] For specific limitations and beneficial effects of the MRI appointment lateness prediction device, please refer to the limitations of the MRI appointment lateness prediction method above, which will not be repeated here. Each module in the aforementioned MRI appointment lateness prediction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the electronic device in hardware form or independently of the processor, or stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of each module.

[0189] This invention also provides an electronic device having the above-described features. Figure 8 The device shown is for predicting late appointments for MRI scans.

[0190] like Figure 9 As shown, Figure 9 This is a schematic diagram of the structure of an electronic device provided in an optional embodiment of the present invention, such as... Figure 9As shown, the electronic device may include: at least one processor 51, such as a CPU (Central Processing Unit), at least one communication interface 53, memory 54, and at least one communication bus 52. The communication bus 52 is used to enable communication between these components. The communication interface 53 may include a display screen or a keyboard; optionally, the communication interface 53 may also include a standard wired interface or a wireless interface. The memory 54 may be high-speed RAM (Random Access Memory) or non-volatile memory, such as at least one disk storage device. Optionally, the memory 54 may also be at least one storage device located remotely from the aforementioned processor 51. The processor 51 may be combined with... Figure 8 The described apparatus has an application program stored in memory 54, and the processor 51 calls the program code stored in memory 54 to perform any of the above method steps.

[0191] The communication bus 52 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The communication bus 52 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 9 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0192] The memory 54 may include volatile memory, such as random-access memory (RAM); the memory may also include non-volatile memory, such as flash memory, hard disk drive (HDD) or solid-state drive (SSD); the memory 54 may also include a combination of the above types of memory.

[0193] The processor 51 can be a central processing unit (CPU), a network processor (NP), or a combination of CPU and NP.

[0194] The processor 51 may further include a hardware chip. This hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0195] Optionally, memory 54 is also used to store program instructions. Processor 51 can invoke program instructions to implement the functions described in this application. Figure 1 as well as Figures 6-7 The method for predicting late appointments for MRI scans shown in the embodiments.

[0196] This invention also provides a non-transitory computer storage medium storing computer-executable instructions that can execute the MRI appointment delay prediction method and the delay prediction method in any of the above method embodiments. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc.; the storage medium may also include combinations of the above types of memory.

[0197] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for predicting late arrivals for nuclear magnetic resonance imaging (MRI) appointments, characterized in that, include: Obtain MRI examination appointment data, which includes appointment information for multiple historical patients and information indicating whether the historical patients were late; The appointment information in the MRI examination appointment data is classified according to different attribute characteristics; The categorized reservation information and the labeled information are input into the initial lateness prediction network, and the initial lateness prediction network is trained to obtain the target lateness prediction model. After the target late arrival prediction model is trained, the model weights obtained through training for each type of appointment information are evaluated and analyzed based on the relationship between the appointment information and the annotation information of each historical patient in the MRI examination appointment data, and the importance score of the feature attributes included in each type of appointment information is determined. Obtain the target appointment information for the MRI scan corresponding to the target patient; The target reservation information is classified according to different attribute characteristics; The categorized target appointment information is input into the target late arrival prediction model. The target late arrival prediction model outputs whether the target patient is late and outputs the importance score of each target feature attribute included in the target appointment information. The lateness prediction model outputs the lateness result of the target patient based on the importance score of each target feature attribute. The step of evaluating the lateness prediction model's output result for the target patient based on the importance scores of each target feature attribute includes: Receive the preset importance level corresponding to each target feature attribute input by the user; The preset importance level corresponding to each of the target feature attributes is compared with the importance score; When the preset importance level corresponding to each of the target feature attributes matches the importance score, the result of whether the target patient is late, output by the target lateness prediction model, is determined to be an accurate result.

2. The method according to claim 1, characterized in that, The appointment information includes: appointment creation time, appointment time, and registration time. Obtaining MRI examination appointment data includes: Obtain the original record data corresponding to each of the aforementioned historical patients; From the original record data, filter the creation appointment time, appointment time, and registration time corresponding to each of the historical patients; The waiting time for each historical patient is obtained by subtracting the corresponding appointment creation time from each appointment time. The MRI examination appointment data is generated based on the appointment creation time, appointment time, registration time, waiting time, and whether each historical patient was late.

3. The method according to claim 2, characterized in that, The process of generating MRI examination appointment data based on the appointment creation time, appointment time, registration time, waiting time, and whether each historical patient was late for each historical patient includes: The time for creating the appointment is converted to a time value to determine the corresponding quarterly value, weekly value, and hourly value for creating the appointment. The appointment time is converted to a time value to determine the corresponding quarterly value, weekday value, and hourly value. The registration time is converted to a time value to determine the corresponding registration quarter value, registration week value, and pre-registration hour value; The MRI examination appointment data is generated based on the quarterly value, weekly value, hourly value, quarterly value, weekly value, hourly value, quarterly value, weekly value, pre-registration hourly value, waiting time, and whether each historical patient was late.

4. The method according to claim 3, characterized in that, The appointment information also includes patient type, which includes inpatients and outpatients. The MRI examination appointment data is generated based on the following parameters for each historical patient: the quarterly value of appointment creation, the weekly value of appointment creation, the hourly value of appointment creation, the quarterly value of appointment creation, the weekly value of appointment creation, the hourly value of appointment creation, the quarterly value of appointment creation, the weekly value of appointment creation, the hourly value of appointment creation, the quarterly value of registration, the weekly value of registration, the hourly value of pre-registration, the waiting time, and whether each historical patient was late. Obtain the medical record number or patient card number corresponding to each of the aforementioned historical patients; Identify the medical record number or the patient card number to determine the patient type corresponding to each of the historical patients; The MRI examination appointment data is generated based on the patient type corresponding to each of the historical patients, the quarterly value of the appointment creation, the weekly value of the appointment creation, the hourly value of the appointment creation, the quarterly value of the appointment, the weekly value of the appointment, the hourly value of the appointment, the quarterly value of the registration, the weekly value of the registration, the hourly value of the pre-registration, the waiting time, and the labeling information of whether each of the historical patients was late.

5. The method according to claim 4, characterized in that, The appointment information also includes: reimbursement type, patient gender, patient age, examination equipment, imaging sequence description, requesting department, examination fee, and registrar.

6. The method according to claim 5, characterized in that, The method further includes: When the preset importance level corresponding to each of the target feature attributes does not match the importance score, the result of whether the target patient is late, output by the target lateness prediction model, is determined to be an inaccurate result. The target lateness prediction model was trained again using the MRI examination appointment data.

7. A device for predicting late arrivals for nuclear magnetic resonance imaging appointments, characterized in that, include: The first acquisition module is used to acquire MRI examination appointment data, which includes appointment information for multiple historical patients and information indicating whether the historical patients were late. The first classification module is used to classify the appointment information in the MRI examination appointment data according to different attribute characteristics; The training module is used to input the classified reservation information and the labeled information into the initial late arrival prediction network, train the initial late arrival prediction network, and obtain the target late arrival prediction model. The analysis module is used to evaluate and analyze the model weights obtained through training for each type of appointment information based on the relationship between the appointment information and the annotation information corresponding to each historical patient in the MRI examination appointment data after the target late arrival prediction model has been trained, and to determine the importance score corresponding to the feature attributes included in each type of appointment information. The second acquisition module is used to acquire the target appointment information for the scheduled MRI of the target patient. The second classification module is used to classify the target reservation information according to different attribute characteristics; The first output module is used to input the classified target appointment information into the target late arrival prediction model. The target late arrival prediction model outputs whether the target patient is late and outputs the importance score of each target feature attribute included in the target appointment information. The second output module is used to evaluate the lateness result of the target patient output by the target lateness prediction model based on the importance scores of each of the target feature attributes. The evaluation of the lateness result based on the importance scores of each of the target feature attributes includes: receiving preset importance levels corresponding to each target feature attribute input by the user; comparing the preset importance levels of each target feature attribute with the importance scores; and determining that the lateness result of the target patient output by the target lateness prediction model is an accurate result when the preset importance levels of each target feature attribute match the importance scores.

8. A late arrival prediction system for nuclear magnetic resonance imaging appointments, characterized in that, The method includes a memory and a processor, wherein the memory stores computer instructions, and the processor executes the computer instructions to perform the MRI appointment lateness prediction method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to execute the MRI appointment delay prediction method according to any one of claims 1-6.

Images