Medical data matching system and method based on double indexes

Abstract

The invention provides a dual-index medical data matching system and method. In the system, medical equipment sends patient data and equipment data locally stored by the medical equipment to a cloud server; the cloud server performs feature extraction on the data, calculates the null value rate of the features, and determines candidate features according to the null value rate; calculating correction mutual information between the candidate features and the disease diagnosis result, and dynamically setting the weight of each candidate feature; determining key features, and respectively constructing a patient main index and an equipment main index according to the key features; the user terminal is used for sending a matching request to the cloud server; and the cloud server calculates a matching score between the to-be-matched patient number and each device feature in the device main index based on a cross-modal matching algorithm, takes the device feature of which the matching score is greater than a preset score threshold as a matching result corresponding to the to-be-matched patient number, and displays the matching result on the user terminal. The accuracy of medical data matching can be improved.

Description

Technical Field

[0001] The present invention belongs to the field of medical technology, and in particular relates to a medical data matching system and method based on dual indexes. Background Art

[0002] With the development of medical information technology, the Patient Master Index (PMI) system has become a key component of the Hospital Information System (HIS). The core function of the Patient Master Index is to uniquely confirm the patient's identity through a unique identifier (such as ID number, medical insurance number, patient ID, etc.), thereby integrating and managing the patient's medical records, examination results, treatment information, etc. in different medical institutions. The Patient Master Index system plays an important role in improving the efficiency of medical services, reducing medical errors and avoiding repeated examinations.

[0003] However, the existing patient master index technology has the following significant problems in application: (1) Incomplete information query: Currently, the query of patient information mainly relies on the patient master index, but this information is usually limited to the patient's basic identity information, such as name, gender, date of birth, etc. This means that in the actual query process, the patient information that can be obtained is still very limited, and it is impossible to fully present the patient's medical history, treatment records, examination results and other key medical data.

[0004] (2) Lack of integration of medical equipment usage information: In addition to the patient's basic information and medical records, the patient master index system is often unable to effectively integrate the medical equipment usage information (such as the use and results of CT, MRI, X-ray and other imaging examination equipment), resulting in the inability to query the medical equipment information used by the patient in different hospitals through the patient master index.

[0005] Due to the above factors, the patient information obtained through the existing patient master index query is usually incomplete, the query results are of low accuracy, and cannot provide accurate support for medical decision-making. This not only affects the efficiency of doctors' diagnosis and treatment, but also increases the risk of delayed diagnosis and treatment errors that patients may face. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide a medical data matching system and method based on dual indexes to improve the accuracy of medical data matching.

[0007] In a first aspect, the present invention provides a medical data matching system based on dual indexing, the system comprising a cloud server, a plurality of medical devices and a user terminal; Medical devices send their locally stored patient data and device data to the cloud server; the device data at least includes disease diagnosis results; The cloud server extracts features from the patient data and the device data to obtain multiple patient features and device features, respectively calculates the null value rate corresponding to each patient feature and device feature, and determines a candidate feature from the multiple patient features and device features according to the null value rate; the candidate feature is a patient feature or device feature whose null value rate is less than or equal to a preset threshold; The cloud server calculates the modified mutual information between the candidate features and the disease diagnosis results, and calculates the weight of each candidate feature based on the modified mutual information, conditional entropy constraints, and nonlinear mapping functions; the modified mutual information is used to quantify the synergistic effect between candidate features, and the conditional entropy constraints are used to avoid overfitting of candidate features with high mutual information but low discriminability; The cloud server determines the candidate features whose weights are greater than a preset threshold as key features, and constructs a patient primary index and a device primary index respectively according to the key features; wherein the key features include at least a patient number and a device number, and the patient primary index and the device primary index are associated with each other through the patient number; The user terminal is used to send a matching request to the cloud server; the matching request includes the patient number to be matched; The cloud server receives or responds to a matching request sent by a user terminal, obtains the patient data corresponding to the matching request from the patient master index, calculates the matching score between the patient number to be matched and each device feature in the device master index based on a cross-modal matching algorithm, takes the device feature whose matching score is greater than a preset score threshold as the matching result corresponding to the patient number to be matched, and displays the matching result on the user terminal.

[0008] Optionally, patient characteristics include patient number, hospital information, identification information, and medical insurance information; Equipment characteristics include equipment number, hospital information, equipment name and equipment usage information.

[0009] Optionally, calculate the modified mutual information between the candidate features and the disease diagnosis results, including: By calculating the formula Get candidate features and disease diagnosis results Modified mutual information ;in, Represents candidate features Disease diagnosis results The first-order mutual information between them characterizes the candidate features Disease diagnosis results The correlation between represents the collaborative redundancy balance factor, , , represents the total number of candidate features, represents the entropy of the disease diagnosis result, Represents candidate features , candidate features Disease diagnosis results The third-order interaction information between Represents candidate features Disease diagnosis results The first-order mutual information between Represents candidate features , candidate features Synergistic effect and disease diagnosis results The mutual information between Represents candidate features Medium attribute value exist The frequency of occurrence in the diagnosis results of the disease class, Represents candidate features Medium attribute value The total frequency of occurrence, express The total frequency of the diagnosis results of the disease class, , Represents candidate features The total number of attribute values ​​in , Represents the total number of categories of disease diagnosis results.

[0010] Optionally, the weight of each candidate feature is calculated based on the modified mutual information, conditional entropy constraint, and nonlinear mapping function, including: By calculating the formula Get candidate features Weight ;in, Represents candidate features Corresponding disease diagnosis results The conditional entropy of , used to prevent division by zero errors, represents the penalty coefficient, Indicates disease diagnosis results Corresponding candidate features The conditional entropy of represents the conditional entropy penalty term, Used to control the nonlinear saturation intensity, .

[0011] Optionally, the expression for the patient primary index is: (Patient No. ) in, represents the patient primary index, Indicates MD5 encoding. Indicates other patient characteristics in the key features except the patient number. , Represents the total number of other patient characteristics.

[0012] Optionally, the device primary index is expressed as: (Device No. ) in, Indicates the device primary index, express coding, Indicates other equipment features in the key features except the equipment number. , Indicates the total number of other device characteristics.

[0013] Optionally, based on a cross-modal matching algorithm, a matching score between the patient number to be matched and each device feature in the device master index is calculated, including: By calculating the formula Get the patient number to be matched Device Characteristics The matching score between ; The joint embedding vector representing the patient is extracted from the patient data using the BERT neural network. The joint embedding vector representing the device is extracted from the device data using a ResNet neural network. represents the balance coefficient, Indicates the historical distribution of device usage in device data Distribution of patient medical records of Divergence.

[0014] In a second aspect, the present invention provides a medical data matching method based on dual indexes, which is applied to the above-mentioned medical data matching system based on dual indexes, and the method comprises: Access patient data and device data stored locally on medical devices; Extracting features from the patient data and the device data to obtain multiple patient features and device features, respectively calculating the null value rate corresponding to each patient feature and device feature, and determining a candidate feature from the multiple patient features and device features according to the null value rate; the candidate feature is a patient feature or device feature whose null value rate is less than or equal to a preset threshold; Calculate the modified mutual information between candidate features and disease diagnosis results, and calculate the weight of each candidate feature based on the modified mutual information, conditional entropy constraints and nonlinear mapping function; the modified mutual information is used to quantify the synergistic effect between candidate features, and the conditional entropy constraint is used to avoid overfitting of candidate features with high mutual information but low discriminability; Determine the candidate features whose weights are greater than a preset threshold as key features, and construct a patient primary index and a device primary index respectively according to the key features; wherein the key features include at least a patient number and a device number, and the patient primary index and the device primary index are associated with each other through the patient number; Based on the cross-modal matching algorithm, the matching score between the patient number to be matched and each device feature in the device primary index is calculated, and the device feature with a matching score greater than a preset score threshold is used as the matching result corresponding to the patient number to be matched.

[0015] In a third aspect, the present invention provides a terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the above method is implemented when the processor executes the computer program.

[0016] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above method when executed by a processor.

[0017] The beneficial effects of the present invention are: The medical data matching system based on dual indexing provided by the present invention has a cloud server that can determine candidate features from multiple patient features and device features according to the null value rate, can effectively screen patient features, avoid interference of redundant features on matching, reduce noise, and is conducive to improving the accuracy of medical data matching; calculate the corrected mutual information between candidate features and disease diagnosis results, and calculate the weight of each candidate feature based on the corrected mutual information, conditional entropy constraints and nonlinear mapping functions, quantify the synergistic effect between features, and avoid overfitting of candidate features with high mutual information but low discriminability, thereby improving the accuracy of medical data matching; based on the cross-modal matching algorithm, calculate the matching score between the patient number to be matched and each device feature in the device main index, and use the device feature with a matching score greater than a preset score threshold as the matching result corresponding to the patient number to be matched, which solves the defects of independent construction of patient and device indexes and lack of dynamic association mechanism, enriches query information, is conducive to improving the accuracy of medical data matching, and has good application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a schematic diagram of the structure of a medical data matching system based on dual indexes in one embodiment of the present application; Figure 2 This is a flowchart of a medical data matching method based on dual indexes in one embodiment of the present application; Figure 3 This is a schematic diagram of the structure of a terminal device in one embodiment of the present application. DETAILED DESCRIPTION

[0019] In view of the technical problem of low cross-platform medical data matching accuracy in the prior art, the present invention provides a medical data matching system and method based on dual indexing, wherein the cloud server of the system can determine candidate features from multiple patient features and device features according to the null value rate, can effectively screen patient features, avoid interference of redundant features on matching, reduce noise, and help improve the accuracy of medical data matching; calculate the corrected mutual information between candidate features and disease diagnosis results, and calculate the weight of each candidate feature based on the corrected mutual information, conditional entropy constraints and nonlinear mapping functions, quantify the synergistic effect between features, and avoid overfitting of candidate features with high mutual information but low discriminability, thereby improving the accuracy of medical data matching; based on a cross-modal matching algorithm, calculate the matching score between the patient number to be matched and each device feature in the device main index, and use the device feature with a matching score greater than a preset score threshold as the matching result corresponding to the patient number to be matched, thereby solving the defects of independent construction of patient and device indexes and lack of a dynamic association mechanism, enriching query information, and helping to improve the accuracy of medical data matching, and having good application prospects.

[0020] The following is an explanation of the medical data matching system based on dual indexing provided by the present invention.

[0021] It should be noted that for the sake of ease of explanation, only two medical devices are used as an example for explanation. In other embodiments, the number of medical devices can be any number, and the hospitals or platforms to which the medical devices belong can be the same or different, and the present invention does not limit this.

[0022] like Figure 1 As shown, the medical data matching system based on dual indexing provided by the present invention includes a cloud server 101, a medical device 102, a medical device 103 and a user terminal 104; wherein the medical device 102 and the medical device 103 belong to different platforms, and illustratively, the medical device 102 belongs to Hospital A, and the medical device 103 belongs to Hospital B.

[0023] It should be noted that the cloud server, medical equipment, and user terminals are connected through network communication and data exchange and transmission are based on the HL7 protocol. The symmetric encryption algorithm AES is used in the data exchange and transmission process to utilize its high efficiency to quickly encrypt large amounts of data to ensure that the data is not stolen during network transmission.

[0024] The functions of each component in the medical data matching system are explained below.

[0025] Medical devices send their locally stored patient data and device data to the cloud server.

[0026] Specifically, for each medical device, the medical device locally stores the patient data and device data of all patients of the platform / hospital to which it belongs. Patient data includes patient number, hospital information, identity information and medical insurance information, and device data includes device number, hospital information, device name and device usage information. The patient number is a unique identifier, and the patient numbers of different patients are different. The identity information includes identity card number, military officer certificate number, passport number and other information that can prove the identity. The medical insurance information includes name, gender, date of birth, blood type, medical data, etc. Medical data includes medical history, drug history, imaging results, test results and other medical instance data, and equipment usage information includes the department of use, usage status, etc. The equipment number is a unique identifier. When the medical device is produced, it is calibrated by the manufacturer, and the equipment numbers corresponding to different medical devices are different.

[0027] Patient data: PERSON (self-increment ID, patient ID, name, gender, date of birth, certificate number, ethnicity, telephone number, blood type, province, city, county, record status, update time, creation time), such as: Person (15, ZA2496497, Zhang San, male, XXXX-XX-XX, 6666666666666666666, Han nationality, XXXXXXXXXXX, A, XX province, XX city, XX district, 0, 2020-09-12 19:19:00, 2020-09-12 19:19:00).

[0028] Visit information data: VISIT_OCCURRENCE (self-increment ID, visit ID, patient ID, visit start time, visit end time, admission department, discharge department, admission diagnosis, admission condition, visit age, ward, bed, follow-up ID, payment type, record status, update time, creation time), such as: VISIT_OCCURRENCE(12,494,ZA2496497,2023-08-22 09:40:49,2023-08-23 16:00:00,Department of Oncology II,Department of Oncology II,'1. Lung malignancy (small cell lung cancer, T2N3M1a, extensive stage); 2. Idiopathic pulmonary fibrosis; 3. Type 2 diabetes. ','The patient underwent enhanced CT on 2023-08-22 due to "intermittent cough and sputum" and was considered to have "central lung cancer in the right upper lung and metastasis to the right lung and right pleura, right hilar and mediastinal lymph nodes: bilateral interstitial pneumonia". ',43,A-5 area,21A,No,Medical insurance payment,0,2023-08-22 09:40:49,2023-08-26 16:00:00).

[0029] Device data: DEVICE (self-increment ID, device number, device type, device name, using department, manufacturer, usage status, update time, creation time), such as: DEVICE (32, XD-0035, electrocardiograph, XD-120, cardiovascular department, Philips, in use, 2023-08-22 09:40:49, 2023-08-26 16:00:00).

[0030] Device medical data: DEVICE_WAVE (auto-increment ID, device number, data type, patient ID, visit ID, measurement value, measurement time, creation time), such as: DEVICE_WAVE (12, XD-0035, ECG, ZA2496497, 494, 132, 2023-08-22 09:40:49, 2023-08-22 09:40:50).

[0031] The cloud server extracts features from the patient data and device data to obtain multiple patient features and device features, calculates the null value rate corresponding to each patient feature and device feature, and determines candidate features from the multiple patient features and device features based on the null value rate.

[0032] In an embodiment of the present invention, the candidate feature is a patient feature or a device feature whose null value rate is less than or equal to a preset threshold.

[0033] It should be noted that the extracted patient features include patient number, hospital information, identity information, and medical insurance information; device features include device number, hospital information, device name, and device usage information. Since patient data and device data are usually structured data, feature extraction can be performed using common structured feature extraction methods, such as using the Scikit-learn method for feature extraction.

[0034] The calculation formula of void rate is:

[0035] Exemplarily, in one embodiment of the present invention, less than 5% of the patient features and device features are used as candidate features.

[0036] The cloud server calculates the modified mutual information between the candidate features and the disease diagnosis results, and calculates the weight of each candidate feature based on the modified mutual information, conditional entropy constraints and nonlinear mapping function.

[0037] In an embodiment of the present invention, the modified mutual information is used to quantify the synergistic effect between candidate features, and the conditional entropy constraint is used to avoid overfitting of candidate features with high mutual information but low discriminability.

[0038] The process of calculating the modified mutual information between the candidate features and the disease diagnosis results is described below.

[0039] Specifically, by calculating the formula Get candidate features Disease diagnosis results Modified mutual information ;in, Represents candidate features Disease diagnosis results The first-order mutual information between them characterizes the candidate features Disease diagnosis results The correlation between represents the collaborative redundancy balance factor, , , represents the total number of candidate features, represents the entropy of the disease diagnosis result, Represents candidate features , candidate features Disease diagnosis results The third-order interaction information between Represents candidate features Disease diagnosis results The first-order mutual information between Represents candidate features , candidate features Synergistic effect and disease diagnosis results The mutual information between Represents candidate features Medium attribute value exist The frequency of occurrence in the diagnosis results of the disease class, Represents candidate features Medium attribute value The total frequency of occurrence, express The total frequency of the diagnosis results of the disease class, , Represents candidate features The total number of attribute values ​​in , Represents the total number of categories of disease diagnosis results.

[0040] It is worth mentioning that compared with the traditional method using first-order mutual information, the modified mutual information used in this invention also takes into account the synergy and redundancy between features, which can significantly improve the suppression rate of redundant features. By introducing a nonlinear compression function (Tanh) instead of Softmax, the weight discrimination is enhanced while maintaining interpretability.

[0041] The following is an explanation of the process of calculating the weight of each candidate feature based on the modified mutual information, conditional entropy constraint, and nonlinear mapping function.

[0042] Specifically, by calculating the formula Get candidate features Weight ;in, Represents candidate features Corresponding disease diagnosis results The conditional entropy of , used to prevent division by zero errors, represents the penalty coefficient, Indicates disease diagnosis results Corresponding candidate features The conditional entropy of represents the conditional entropy penalty term, Used to control the nonlinear saturation intensity, .

[0043] It is worth mentioning that the traditional technology uses fixed weights and cannot dynamically adjust feature weights according to data quality. When the equipment fails, it will lead to erroneous results and poor accuracy of medical data matching. However, this application calculates the weight of each candidate feature based on modified mutual information, conditional entropy constraints and nonlinear mapping functions, and can dynamically adjust feature weights, fundamentally solving the drawbacks of fixed weights and helping to improve the accuracy of medical data matching.

[0044] The cloud server determines the candidate features whose weights are greater than a preset threshold as key features, and constructs the patient primary index and the device primary index respectively based on the key features.

[0045] Among them, the key features include at least the patient number and the device number, and the patient primary index and the device primary index are associated through the patient number. In one embodiment of the present invention, the preset threshold is set to 0.6. This can further reduce the interference of redundant features on data matching and improve the accuracy of medical data matching.

[0046] Specifically, the expression of the patient's primary index is: (Patient No. ) in, represents the patient primary index, Indicates MD5 encoding. Indicates other patient characteristics in the key features except the patient number. , Represents the total number of other patient characteristics.

[0047] The expression for the device primary index is: (Device No. ) in, Indicates the device primary index, express coding, Indicates other equipment features in the key features except the equipment number. , Indicates the total number of other device characteristics.

[0048] The user terminal is used to send a matching request to the cloud server.

[0049] In an embodiment of the present invention, the matching request includes a patient number to be matched.

[0050] The cloud server receives or responds to a matching request sent by a user terminal, obtains the patient data corresponding to the matching request from the patient master index, calculates the matching score between the patient number to be matched and each device feature in the device master index based on a cross-modal matching algorithm, takes the device feature whose matching score is greater than a preset score threshold as the matching result corresponding to the patient number to be matched, and displays the matching result on the user terminal.

[0051] The following is an explanation of the process of calculating the matching score between the patient number to be matched and each device feature in the device master index based on the cross-modal matching algorithm.

[0052] Specifically, by calculating the formula Get the patient number to be matched Device Characteristics The matching score between ; The joint embedding vector representing the patient is extracted from the patient data using the BERT neural network. The joint embedding vector representing the device is extracted from the device data using a ResNet neural network. represents the balance coefficient, Indicates the historical distribution of device usage in device data Distribution of patient medical records The KL divergence of .

[0053] In order to verify the accuracy of the medical data matching system based on dual indexing provided by the present invention, in an embodiment where an emergency patient in a hospital needs to match an available monitor immediately, the patient ID and vital signs data (heart rate, blood pressure) are input to obtain the matching monitor number and historical calibration records. Compared with the traditional technology, the matching time of the present invention is reduced from 2 minutes to 8 seconds, the equipment utilization rate is increased by 40%, and the matching accuracy is increased by 35%.

[0054] In summary, the medical data matching system based on dual indexing provided by the present invention has a cloud server that can determine candidate features from multiple patient features and device features according to the null value rate, can effectively screen patient features, avoid the interference of redundant features on matching, reduce noise, and is conducive to improving the accuracy of medical data matching; calculate the corrected mutual information between candidate features and disease diagnosis results, and calculate the weight of each candidate feature based on the corrected mutual information, conditional entropy constraints and nonlinear mapping functions, quantify the synergistic effect between features, and can avoid overfitting of candidate features with high mutual information but low discriminability, thereby improving the accuracy of medical data matching; based on the cross-modal matching algorithm, calculate the matching score between the patient number to be matched and each device feature in the device main index, and use the device feature with a matching score greater than a preset score threshold as the matching result corresponding to the patient number to be matched, which solves the defects of independent construction of patient and device indexes and lack of dynamic association mechanism, enriches query information, is conducive to improving the accuracy of medical data matching, and has good application prospects.

[0055] The following is a description of the medical data matching method based on dual indexes provided by the present invention and applied to the medical data matching system based on dual indexes.

[0056] like Figure 2 As shown, the method comprises the following steps: Step 21, obtaining patient data and device data stored locally in the medical device.

[0057] Step 22, extract features from the patient data and device data to obtain multiple patient features and device features, calculate the null value rate corresponding to each patient feature and device feature respectively, and determine candidate features from the multiple patient features and device features based on the null value rate.

[0058] Candidate features are patient features or device features whose null value rate is less than or equal to a preset threshold.

[0059] Step 23, calculating the modified mutual information between the candidate features and the disease diagnosis results, and calculating the weight of each candidate feature based on the modified mutual information, conditional entropy constraints and nonlinear mapping function.

[0060] The modified mutual information is used to quantify the synergistic effect between candidate features, and the conditional entropy constraint is used to avoid overfitting of candidate features with high mutual information but low discriminability.

[0061] Step 24, determining the candidate features whose weights are greater than a preset threshold as key features, and constructing a patient primary index and a device primary index respectively according to the key features.

[0062] Among them, the key features include at least a patient number and a device number, and the patient primary index and the device primary index are associated with each other through the patient number.

[0063] Step 25, based on the cross-modal matching algorithm, calculate the matching score between the patient number to be matched and each device feature in the device main index, and take the device feature with a matching score greater than a preset score threshold as the matching result corresponding to the patient number to be matched.

[0064] The specific implementation process of each step has been described in detail in the system part, and will not be repeated here. The method disclosed in the present invention is applied to the above-mentioned medical data matching system based on dual indexes, and has the same technical effect as that of the above-mentioned medical data matching system.

[0065] like Figure 3 As shown, an embodiment of the present invention provides a terminal device, such as Figure 3 As shown, the terminal device D10 of this embodiment includes: at least one processor D100 ( Figure 3 Only one processor is shown in the figure), a memory D101, and a computer program D102 stored in the memory D101 and executable on the at least one processor D100, wherein the processor D100 implements the steps of any of the above-mentioned method embodiments when executing the computer program D102.

[0066] Specifically, when the processor D100 executes the computer program D102, it obtains patient data and device data stored locally in the medical device; performs feature extraction on the patient data and device data to obtain multiple patient features and device features, calculates the null value rate corresponding to each patient feature and device feature respectively, and determines candidate features from the multiple patient features and device features based on the null value rate; calculates the modified mutual information between the candidate features and the disease diagnosis results, and calculates the weight of each candidate feature based on the modified mutual information, conditional entropy constraints, and nonlinear mapping functions; determines the candidate features whose weights are greater than a preset threshold as key features, and constructs a patient primary index and a device primary index respectively based on the key features; based on a cross-modal matching algorithm, calculates the matching score between the patient number to be matched and each device feature in the device primary index, and uses the device feature whose matching score is greater than the preset score threshold as the matching result corresponding to the patient number to be matched. Among them, candidate features are determined from multiple patient features and device features according to the null value rate, which can effectively screen patient features, avoid the interference of redundant features on matching, reduce noise, and help improve the accuracy of medical data matching; the modified mutual information between candidate features and disease diagnosis results is calculated, and the weight of each candidate feature is calculated based on the modified mutual information, conditional entropy constraints and nonlinear mapping functions, which quantifies the synergistic effect between features and can avoid overfitting of candidate features with high mutual information but low discriminability, thereby improving the accuracy of medical data matching; based on the cross-modal matching algorithm, the matching score between the patient number to be matched and the device features in the device main index is calculated, and the device features with a matching score greater than the preset score threshold are used as the matching results corresponding to the patient number to be matched, which solves the defects of independent construction of patient and device indexes and lack of dynamic association mechanism, enriches the query information, and is conducive to improving the accuracy of medical data matching, and has good application prospects.

[0067] The processor D100 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0068] In some embodiments, the memory D101 may be an internal storage unit of the terminal device D10, such as a hard disk or memory of the terminal device D10. In other embodiments, the memory D101 may also be an external storage device of the terminal device D10, such as a plug-in hard disk, a smart memory card (SMC, SmartMedia Card), a secure digital (SD, Secure Digital) card, a flash card (Flash Card), etc. equipped on the terminal device D10. Further, the memory D101 may also include both an internal storage unit of the terminal device D10 and an external storage device. The memory D101 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program, etc. The memory D101 may also be used to temporarily store data that has been output or is to be output.

[0069] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0070] An embodiment of the present application provides a computer program product. When the computer program product runs on a terminal device, the terminal device can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0071] A person skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of protection of the present application is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of different aspects of one or more embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.

[0072] One or more embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the present application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of the present application should be included in the protection scope of the present application.

Claims

1. A medical data matching system based on dual index, characterized in that: Includes cloud servers, multiple medical devices, and user terminals; The medical device sends the patient data and device data stored locally to the cloud server; the device data at least includes the disease diagnosis result; The cloud server extracts features from the patient data and the device data to obtain a plurality of patient features and device features, respectively calculates a null value rate corresponding to each patient feature and device feature, and determines a candidate feature from the plurality of patient features and device features according to the null value rate; the candidate feature is a patient feature or device feature whose null value rate is less than or equal to a preset threshold; The cloud server calculates the modified mutual information between the candidate features and the disease diagnosis results, and calculates the weight of each candidate feature based on the modified mutual information, conditional entropy constraints and nonlinear mapping function; The modified mutual information is used to quantify the synergistic effect between candidate features, and the conditional entropy constraint is used to avoid overfitting of candidate features with high mutual information but low discriminability; The cloud server determines the candidate features whose weights are greater than a preset threshold as key features, and constructs a patient primary index and a device primary index respectively according to the key features; wherein the key features include at least a patient number and a device number, and the patient primary index and the device primary index are associated with each other through the patient number; The user terminal is used to send a matching request to the cloud server; the matching request includes a patient number to be matched; The cloud server receives or responds to a matching request sent by the user terminal, obtains the patient data corresponding to the matching request from the patient master index, calculates the matching score between the patient number to be matched and each device feature in the device master index based on a cross-modal matching algorithm, takes the device feature with a matching score greater than a preset score threshold as the matching result corresponding to the patient number to be matched, and displays the matching result on the user terminal.

2. The medical data matching system according to claim 1, characterized in that: The patient characteristics include patient number, hospital information, identity information, and medical insurance information; The equipment characteristics include equipment number, hospital information, equipment name and equipment usage information.

3. The medical data matching system according to claim 2, characterized in that: The calculating of the modified mutual information between the candidate feature and the disease diagnosis result comprises: By calculating the formula Get candidate features and disease diagnosis results Modified mutual information ;in, Represents candidate features and disease diagnosis results The first-order mutual information between them characterizes the candidate features and disease diagnosis results The correlation between represents the collaborative redundancy balance factor, , , represents the total number of candidate features, represents the entropy of disease diagnosis results, Represents candidate features , candidate features and disease diagnosis results The third-order interaction information between Represents candidate features and disease diagnosis results The first-order mutual information between Represents candidate features , candidate features Synergistic effect and disease diagnosis results The mutual information between Represents candidate features Medium attribute value exist The frequency of occurrence in the diagnosis results of the disease class, Represents candidate features Medium attribute value The total frequency of occurrence, express The total frequency of the diagnosis results of the disease class, , Represents candidate features The total number of attribute values ​​in , Represents the total number of categories of disease diagnosis results.

4. The medical data matching system according to claim 3, characterized in that: The calculating the weight of each candidate feature based on the modified mutual information, the conditional entropy constraint and the nonlinear mapping function includes: By calculating the formula Get candidate features Weight ;in, Represents candidate features Corresponding disease diagnosis results The conditional entropy of , used to prevent division by zero errors, represents the penalty coefficient, Indicates disease diagnosis results Corresponding candidate features The conditional entropy of represents the conditional entropy penalty term, Used to control the nonlinear saturation intensity, .

5. The medical data matching system according to claim 4, characterized in that: The expression of the patient primary index is: (Patient No. ) in, represents the patient primary index, Indicates MD5 encoding. Indicates other patient characteristics in the key features except the patient number. , represents the total number of other patient characteristics described.

6. The medical data matching system according to claim 5, characterized in that: The expression of the device primary index is: (Device No. ) in, represents the primary index of the device, express coding, Indicates other equipment features in the key features except the equipment number. , Indicates the total number of other device features described.

7. The medical data matching system according to claim 6, characterized in that: The step of calculating the matching score between the patient number to be matched and each device feature in the device primary index based on the cross-modal matching algorithm includes: By calculating the formula Get the patient number to be matched Device Characteristics The matching score between ; The joint embedding vector representing the patient is extracted from the patient data using the BERT neural network. The joint embedding vector representing the device is extracted from the device data using a ResNet neural network. represents the balance coefficient, Indicates the historical distribution of device usage in device data Distribution of patient medical records The KL divergence of .

8. A medical data matching method based on dual indexing, characterized in that: The method is applied to the medical data matching system based on dual index as described in any one of claims 1 to 7, and the method comprises: Access patient data and device data stored locally on medical devices; Performing feature extraction on the patient data and the device data to obtain a plurality of patient features and device features, respectively calculating the null value rate corresponding to each patient feature and device feature, and determining a candidate feature from the plurality of patient features and device features according to the null value rate; the candidate feature is a patient feature or device feature whose null value rate is less than or equal to a preset threshold; Calculating the modified mutual information between the candidate features and the disease diagnosis results, and calculating the weight of each candidate feature based on the modified mutual information, conditional entropy constraints and nonlinear mapping functions; the modified mutual information is used to quantify the synergistic effect between the candidate features, and the conditional entropy constraints are used to avoid overfitting of candidate features with high mutual information but low discriminability; Determine the candidate features whose weights are greater than a preset threshold as key features, and construct a patient primary index and a device primary index respectively according to the key features; wherein the key features include at least a patient number and a device number, and the patient primary index and the device primary index are associated with each other through the patient number; Based on the cross-modal matching algorithm, the matching score between the patient number to be matched and each device feature in the device main index is calculated, and the device feature with a matching score greater than a preset score threshold is used as the matching result corresponding to the patient number to be matched.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to claim 8 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to claim 8 is implemented.

Images