Patient primary index management method and system based on big data algorithm

Abstract

The invention provides a patient primary index management method and system based on a big data algorithm. The method comprises the following steps: acquiring original data of a plurality of historical patients; the method comprises the following steps: constructing a patient main index database containing multi-dimensional features through original data, and performing feature weight configuration on each feature of the multi-dimensional features based on a preset rule; the multi-dimensional features comprise social attribute features, biological attribute features and medical attribute features; multi-dimensional features of a patient to be recognized are obtained, and the comprehensive matching degree of the multi-dimensional features of the patient to be recognized and the existing EMPI in the patient main index database is calculated based on the feature weight, so that the operation of associating the multi-dimensional features to the existing EMPI or registering the multi-dimensional features as a new EMPI is executed; the repeated EMPIs are regularly combined, and the patient main index database is updated. The method solves the problems of feature variability and historical data merging, and has the advantages of high recognition precision and high matching strategy intelligence degree.

Description

A Patient Master Index Management Method and System Based on Big Data Algorithms Technical Field

[0001] This invention relates to the field of medical information technology, and more specifically, to a patient master index management method and system based on big data algorithms. Background Technology

[0002] With the deepening of medical informatization, patient medical data is scattered across multiple independent or heterogeneous systems, including outpatient, inpatient, physical examination, and laboratory testing systems. Due to differences in system construction standards, data collection processes (such as the error-prone nature of manual entry), and patient circumstances (such as seeking medical treatment without proper documentation or information changes), the same patient may have multiple unrelated identifiers and records within the hospital or even on regional medical information platforms. This "information silo" phenomenon severely hinders the construction of full-cycle patient health records, cross-departmental collaborative diagnosis and treatment, and the interconnection of regional medical information.

[0003] Existing patient master index solutions primarily rely on a single or a few key identifiers (such as ID card number or medical insurance card number) for precise matching. These solutions have significant limitations: incomplete coverage: they cannot effectively handle patients without identification documents, or whose documents contain incorrect or missing information; limited accuracy: they rely solely on social attributes and cannot comprehensively utilize high-value information such as biological attributes (e.g., blood type, height and weight trends) and medical attributes (e.g., specific disease history, surgical history) for cross-validation; rigidity: they use fixed matching thresholds and weights, failing to adapt to dynamic changes in data quality, patient group characteristics (e.g., children's growth and development), and medical scenarios (e.g., emergency and chronic disease management), leading to high rates of mismatches or missed matches.

[0004] Therefore, there is an urgent need for a patient master index management method that can integrate multi-dimensional features, dynamically assess feature credibility, and intelligently adjust matching strategies to achieve high-precision, robust, and flexibly configurable patient identification and data integration. Summary of the Invention

[0005] This invention aims to address the problems in existing technologies, such as low patient identification accuracy, difficulty in cross-system data association, and inability to adapt to complex real-world scenarios due to reliance on a single identifier, insufficient feature utilization, and rigid matching strategies. Specifically, it seeks to provide a method to address challenges such as identifying patients without identification documents, handling volatile features (e.g., children's height and weight), and merging historical data. To achieve these objectives, this invention proposes a patient master index management method and system based on big data algorithms.

[0006] Firstly, this application provides a patient master index management method based on big data algorithms, including: step S1, acquiring raw data of multiple historical patients; acquiring raw data of historical patients from various hospital business systems (such as hospital information systems and laboratory information systems), and performing data preprocessing on the raw data information. The preprocessing includes: marking missing feature fields in the raw data; intercepting or correcting obviously erroneous or abnormal feature values ​​in the raw data; and converting unstructured data into structured feature data.

[0007] Step S2: Construct a patient master index database containing multi-dimensional features using the raw data, and configure feature weights for each feature based on preset rules. The multi-dimensional features include social attribute features, biological attribute features, and medical attribute features. Social attribute features include: name, ID number, contact number, home address, and workplace. ID number is a highly stable and reliable feature, while contact number and workplace are variable features. Biological attribute features include: gender, date of birth, blood type, height, and weight. Blood type and gender are lifelong and reliable features, while height and weight (especially in children) are variable features that change over time. Medical attribute features include: disease diagnosis records, core medical history, medication history, and allergy history. Among them, core medical history and allergy history are reliable features; among them, ID number, blood type, gender and core medical history are stable features with high stability, and their initial weight is configured to be no less than 0.8; the variable features include at least one of height, weight, contact number and home address, and their initial weight is configured to be no more than 0.3.

[0008] An initial weight is preset based on the inherent stability of each feature, wherein the initial weight of stable features is greater than the initial weight of volatile features; the current effective weight is calculated based on the initial weight and a feature decay function, the expression of which is as follows: ;in, The effective weights of the features at time t; These are the initial weights; The time decay coefficient is the feature's time decay factor; t is the time since feature acquisition (in years); the time decay coefficient Based on feature stability settings: for stable features such as ID number, blood type, and gender, the time decay coefficient is set... =0; for medical history and medication history that fall between stable and variable characteristics, the time decay coefficient is 0 < λ ≤ 0.2; for variable characteristics such as height and weight, the time decay coefficient is λ > 0.2.

[0009] Step S3: When a patient to be identified seeks medical attention, extract their multi-dimensional features. Calculate the comprehensive matching degree between the multi-dimensional features and existing EMPIs in the patient master index based on feature weights. By comparing the comprehensive matching degree with a dynamic threshold, decide whether to associate the patient with an existing EMPI or register them as a new EMPI.

[0010] ;in, The dynamic threshold after the k-th successful match; : Minimum threshold, in some specific embodiments = 70%; Initial threshold =80%; k: number of consecutive successful matches, when k≥3, k=3; Threshold adjustment coefficient; Historical misjudgment rate (average misjudgment rate over the past 30 days); when When the threshold is ≥0.05 (false positive rate ≥5%), the threshold adjustment coefficient α is automatically halved to 0.0125 to control the risk of false positives by slowing down the threshold adjustment speed; when When the threshold is ≤0.01 (false positive rate ≤1%), the threshold adjustment coefficient α can be increased to 0.03, thereby improving matching efficiency by accelerating the threshold reduction speed.

[0011] Step S4: Regularly and automatically detect and merge duplicate EMPIs. During merging and daily updates, data conflicts are resolved according to conflict resolution rules, and feature weights are updated to achieve continuous self-optimization of the patient master index.

[0012] Step S5: For patients whose social attribute characteristics cannot be obtained, a temporary EMPI is generated, and the medical data of this visit is associated with this temporary EMPI. Specifically, this includes: generating a temporary EMPI based on the patient's biological attribute characteristics; after the temporary EMPI is supplemented with social attribute characteristics, an association operation is performed, which includes: performing precise matching using the social attribute characteristics as key identifiers; performing multi-dimensional feature cross-validation based on the biological attribute characteristics stored in the temporary EMPI and the characteristics in the patient's main index; according to the validation result, if the validation passes, all medical data under the temporary EMPI is migrated and associated with the corresponding EMPI; if the validation fails, the temporary EMPI is registered as a normal EMPI and entered into the patient's main index.

[0013] The beneficial effects of this invention are as follows: the recognition accuracy and coverage are greatly improved: by integrating social, biological and medical three-dimensional features and assigning dynamic weight values, high-precision identity recognition can still be achieved without relying on a single ID card, effectively covering complex scenarios such as no ID card or incorrect ID card, and it is expected to increase the cross-system patient data association rate by more than 40%.

[0014] It has strong intelligence and adaptability: the feature weight calculation and dynamic threshold judgment combine the time decay law of feature effectiveness, and adjust the strategy itself according to the actual operation effect (false judgment rate), which improves the long-term operation stability and reduces manual intervention.

[0015] High operational efficiency and reduced manual workload: The automated EMPI registration, matching, and merging process frees medical staff from tedious manual verification and data entry, potentially reducing related workload by over 60%. Intelligent merging based on graph clustering can efficiently handle complex duplicate record problems.

[0016] It offers excellent configurability and scalability: the feature system, initial weights, attenuation coefficients, dynamic threshold parameters, etc., can all be configured through the management platform, enabling this solution to flexibly adapt to the characteristics of patient groups and data governance levels in different hospitals and regions.

[0017] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 is a flowchart of the patient master index management method based on big data algorithm of the present invention. Detailed Implementation

[0020] 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, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0021] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0022] English Abbreviation Explanation Enterprise Master Patient Index (EMPI) Example 1: Referring to Figure 1, this example uses a tertiary-level Class A hospital as an example to provide a patient master index management method based on big data algorithms.

[0023] Step S1: Obtain raw data from multiple historical patients; extract historical patient data from the past 5 years from multiple existing systems such as HIS (Hospital Information System), LIS (Laboratory Information System), and PACS (Picture Archiving and Communication System). Use data cleaning tools to preprocess the raw data, such as standardizing ID number formats, converting disease diagnosis names to ICD-10 codes, and parsing unstructured medical record text to extract keywords.

[0024] Step S2: Construct a patient master index database containing multi-dimensional features using raw data, and configure feature weights for each feature based on preset rules. The multi-dimensional features include social attribute features, biological attribute features, and medical attribute features. Specifically, Step S2 includes: Step S201: Obtain multi-dimensional features including social attribute features, biological attribute features, and medical attribute features based on preprocessed raw data, and construct a patient master index database containing multi-dimensional features. The social attribute features include: name, ID number, contact number, home address, and workplace. ID number is a highly stable and reliable feature, while contact number and workplace are variable features. The biological attribute features include: gender, date of birth, blood type, height, and weight. Blood type and gender are lifelong and reliable features, while height and weight (especially in children) are variable features that change over time. The medical attribute features include: disease diagnosis records, core medical history, medication history, and allergy history. Core medical history and allergy history are reliable features.

[0025] Step S202: Calculate the feature weight for each feature of the multi-dimensional features based on preset rules; in some specific embodiments, an initial weight is preset based on the stability of each feature, wherein the initial weight of stable features is greater than the initial weight of volatile features; in this embodiment, the initial weight of the certificate number is... The initial weight of blood type is 0.95. The initial weight for gender is 0.90. The initial weight for core disease diagnosis is 0.85. The initial weight of the mobile phone number is 0.80. The initial weight of a child's height is 0.50. The value is 0.25; the current effective weight is calculated based on the initial weight and the feature decay function, the expression of which is as follows: ;in, The effective weights of the features at time t; These are the initial weights; t represents the time decay coefficient of the feature; t is the time since the feature was collected (in years); in this embodiment, the ID number, blood type, and gender are stable features that remain unchanged, and their time decay coefficients are... The time decay coefficient is 0, meaning no decay; for medical history and medication history, which fall between stable and volatile characteristics, the time decay coefficient is 0 < λ ≤ 0.2, where λ = 0.1 for disease diagnosis records; λ = 0.2 for mobile phone numbers; and λ = 0.5 for children's height.

[0026] In some specific embodiments, failure features are periodically detected. For each failure feature: a failure feature is identified and marked based on a preset failure rule, and the failure feature is reduced in weight or excluded in subsequent matching calculations; the failure rule includes time-based failure rules and / or statistical failure rules; the time-based failure rule is: when the current effective weight calculated by the feature decay function is lower than a preset first failure threshold, the feature is determined to be failed; that is, when the effective weight of the feature is lower than the current effective weight, the failure is determined to be lower than the current effective weight. If the value is ≤0.1, it is considered a failure feature and is automatically removed.

[0027] The statistical failure rule is as follows: when the value of the same feature of the same patient changes beyond the preset range of change obtained from historical data statistics in multiple consecutive collections, the feature is determined to be invalid or statistically abnormal, and the weight of the feature is automatically reduced.

[0028] A historical distribution model of features is constructed by grouping by age and gender to identify failures caused by abnormal changes in features. For each feature group, the mean μ, standard deviation σ, and interquartile range IQR = Q3 - Q1 (Q3 is the third quartile, and Q1 is the first quartile) are obtained. If the newly collected feature value of the group exceeds the interval [Q1 - 1.5×IQR, Q3 + 1.5×IQR] and the difference from the historical value exceeds 2σ, it is marked as "to be verified". If the difference exceeds 3σ for two consecutive collections, the feature is considered to be invalid. For example, if a child's height increases by 8cm in one year, which is far beyond 3σ = 4.5cm for the same age group, the weight of the feature is automatically reduced from 0.25 to 0.05.

[0029] The feature is included in the matching calculation only if it passes both the time decay rule (time decay is not invalid) and the statistical decay rule (statistical distribution is not abnormal); otherwise, the feature is marked as an invalid feature and is removed or its weight is reduced.

[0030] Step S203: Before calculating the comprehensive matching degree, the Softmax function is used to normalize the weights of all features participating in the current matching calculation. Let the initial weight set be... This yields the normalized initial weight set. This ensures that the sum of the initial weights of all features is 1. The expression for the weight normalization is as follows: .

[0031] Step S3: Obtain the multi-dimensional features of the patient to be identified, and calculate the comprehensive matching degree between the multi-dimensional features and the existing EMPIs in the patient master index based on the feature weights, so as to perform the operation of associating it with the existing EMPI or registering it as a new EMPI; Step S3 specifically includes: Step S301: Obtain feature matching values ​​based on the multi-dimensional features of the patient to be identified and the patient master index; In some specific embodiments, a patient named Zhang, who usually visits the clinic, comes to the outpatient clinic for treatment. The triage desk enters his name "Zhang Xiaoming", date of birth, mobile phone number, gender and blood type, etc.

[0032] We will first obtain its multi-dimensional features: name and mobile phone number (social attribute features), gender and date of birth (biological attributes), etc.

[0033] Step S302: Obtain the comprehensive matching degree based on the feature matching value and the effective weight corresponding to the matching feature. Search for similar EMPIs in the patient's main index database. A record from 3 years ago, EMPI-001, is found with the name "Zhang Xiaoming" (matching value 0.95) and blood type A (matching value 1.0), but the height at that time was 110cm (matching value 0.6). Based on the feature decay function, the effective weight of the "height" feature in EMPI-001 has decreased from 0.25 to a very low level due to 3 years of decay. The calculation expression based on the comprehensive matching degree is as follows: Comprehensive matching degree = Σ(feature matching value × feature weight) / Σ feature weight. Step S303: The comprehensive calculation shows that the matching degree between the current medical information and EMPI-001 is 82%, which is higher than the current dynamic threshold of 78%. Therefore, the automatic association is successful, and the current medical record is included in EMPI-001. The patient's height, weight, and other feature values ​​and weights are updated. If the comprehensive matching degree calculated this time is less than the dynamic threshold and less than the second threshold, it is registered as a new EMP.

[0034] The expression for the dynamic threshold is as follows: ;in, The dynamic threshold after the k-th successful match. Minimum threshold =70%, Initial threshold =80%, k: number of consecutive successful matches. When k≥3, it is calculated as 3 to prevent excessive downward adjustment; Threshold adjustment coefficient; Historical misjudgment rate; This represents the system's recent (e.g., within 30 days) historical misjudgment rate.

[0035] A dual feedback loop is achieved through dynamic thresholds: when the number of successful matches increases (k increases), it indicates a greater certainty about the patient's identity, and the threshold can be appropriately relaxed to enhance the correlation ability; when the overall misjudgment rate increases (k increases), the threshold decreases. If the threshold is increased, the reduction will be automatically slowed down or even paused, prioritizing matching accuracy. This allows for an intelligent balance between precision and recall.

[0036] Step S4: Periodically merge duplicate EMPIs and update the patient master index database; Step S401: Periodically traverse the patient master index database, calculate the overall matching degree between different EMPIs, and screen out EMPI pairs with an overall matching degree not lower than the merging threshold as candidate merging pairs; In some specific embodiments, the patient master index database is traversed every morning to calculate the overall matching degree between different EMPIs. For example, it is found that the matching degree between EMPI-A and EMPI-B is 92%, the matching degree between EMPI-B and EMPI-C is 91%, but the direct matching degree between EMPI-A and EMPI-C is only 75%. Therefore, EMPI-A and EMPI-B are considered as one candidate merging pair, and EMPI-B and EMPI-C are considered as another candidate merging pair.

[0037] Step S402: Based on the candidate merging pairs, construct an association graph with EMPIs as nodes and matching relationships as edges, and identify the sets of mutually related EMPIs through a graph traversal algorithm to form association clusters to be processed; according to the graph traversal algorithm, identify EMPI-A, EMPI-B and EMPI-C as the same association cluster; Step S403: For each association cluster, calculate and compare the information integrity score of each EMPI in the cluster, and determine the EMPI with the highest integrity score as the core EMPI; after calculation, EMPI-A has the most complete information (with ID number, blood type and multiple medical histories), and is selected as the core EMPI.

[0038] Step S404: For each EMPI in the associated cluster other than the core EMPI, migrate all patient data associated with it to the core EMPI; if each EMPI in the associated cluster other than the core EMPI has a different value on the same feature as the core EMPI, determine the final value of the feature in the core EMPI according to the preset conflict arbitration rules. The conflict arbitration rules include at least one of the following priority orders: feature values ​​collected by automated equipment take precedence over feature values ​​entered manually; feature values ​​collected more recently take precedence over feature values ​​collected earlier; in the embodiment, it is found that the height of EMPI-B is different from that of EMPI-A during merging, but the height of EMPI-A is the most recently updated and obtained by the body weight collector. According to the conflict arbitration rules, the height data of EMPI-A is retained.

[0039] Step S405: Mark the EMPI status that has completed data migration as merged, and remove it from the effective index set of the patient master index library, while recording its merging relationship with the core EMPI; update the patient master index library; that is, migrate all medical records of EMPI-B and EMPI-C to EMPI-A, and mark the status of B and C as "merged".

[0040] In response to emergencies, such as when a patient arrives at a hospital unconscious and whose social attributes cannot be obtained, this method also includes generating a temporary EMPI for patients whose social attributes cannot be obtained, and associating the patient's medical data with this temporary EMPI. In some specific embodiments, at night, an unidentified patient is brought to the emergency room, and the following steps are performed: Step S501: Generate a temporary EMPI based on the unidentified patient's biometric characteristics; A temporary EMPI is generated using limited information provided by the ambulance personnel (approximate age, gender) and the results of the initial vital signs examination. All medical attributes such as resuscitation records and medication records are associated with this temporary EMPI; Step S502: After the temporary EMPI is supplemented with social attributes, an association operation is performed. The association operation includes: performing precise matching using the social attributes as a key identifier; Three days later, the unidentified patient's identity is confirmed, and the family provides an ID number. At this time, the ID number is used to find a formal EMPI of the patient from many years ago, thereby obtaining a precise match.

[0041] Step S503: In order to ensure a more accurate association, multi-dimensional feature cross-validation is performed based on the biological attribute features stored in the temporary EMPI and the features of the patient's main index. The "blood type O" and "rare drug allergy history" recorded in the temporary EMPI are completely consistent with the official EMPI history, with a matching degree of up to 96%.

[0042] Step S504: Based on the verification results, the verification process involves migrating and associating all medical data under the temporary EMPI with the corresponding EMPI, automatically merging all emergency data under the temporary EMPI into the formal EMPI, and completing the historical archive. If the multi-dimensional feature cross-verification in step S503 fails and the matching degree is less than 70%, the temporary EMPI will be registered as a normal EMPI and entered into the patient master index database.

[0043] Example 2: This example provides a patient master index management system based on big data algorithms. The system includes a data acquisition module for acquiring raw data from multiple historical patients and multi-dimensional features of the patient to be identified; a master index library construction and maintenance module for constructing and maintaining a patient master index library containing multi-dimensional features based on the raw data of the historical patients; the multi-dimensional features include social attribute features, biological attribute features, and medical attribute features; a feature weight calculation module for calculating the feature weight of each feature in the multi-dimensional features based on preset rules; a patient matching and registration module for calculating the comprehensive matching degree between the multi-dimensional features of the patient to be identified and existing EMPIs in the patient master index library based on the feature weights, and associating the patient to be identified with a matched existing EMPI or registering it as a new EMPI according to the matching result; and an EMPI merging and updating module for updating the patient master index library by merging duplicate EMPIs in the patient master index library.

[0044] It should be noted that the specific manner in which each module performs its operation in the apparatus described in the above embodiments has been described in detail in the embodiments of the method, and will not be elaborated here.

[0045] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0046] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A patient master index management method based on big data algorithms, characterized in that, include: Obtain raw data from multiple historical patients; construct a patient master index library containing multi-dimensional features using the raw data, and configure feature weights for each feature based on preset rules; the multi-dimensional features include social attribute features, biological attribute features, and medical attribute features; The system acquires multi-dimensional features of the patients to be identified and calculates the overall matching degree between these features and existing EMPIs in the patient master index based on feature weights, in order to perform operations such as associating them with existing EMPIs or registering them as new EMPIs; the EMPIs represent enterprise-level patient master indexes; duplicate EMPIs are periodically merged and the patient master index is updated.

2. The patient master index management method based on big data algorithms according to claim 1, characterized in that, The step of calculating the feature weights for each feature of the multi-dimensional features based on preset rules includes: presetting an initial weight based on the stability of each feature, wherein the initial weight of stable features is greater than the initial weight of volatile features; and calculating the current effective weight based on the initial weights and a feature decay function, the expression of which is as follows: ;in, The effective weights of the features at time t; These are the initial weights; is the time decay coefficient of the feature; t is the time since the feature was acquired.

3. The patient master index management method based on big data algorithms according to claim 2, characterized in that, The preset rules also include: identifying and marking failure features based on preset failure rules, wherein the failure features are reduced in weight or excluded in subsequent matching calculations; the failure rules include time failure rules and / or statistical failure rules; the time failure rule is: when the current effective weight calculated by the feature decay function is lower than a preset first failure threshold, the feature is determined to be failure; the statistical failure rule is: when the value change of the same feature of the same patient exceeds a preset change range obtained from historical data statistics in multiple consecutive collections, the feature is determined to be failure or statistically abnormal, and the weight of the feature is automatically reduced.

4. The patient master index management method based on big data algorithms according to claim 2, characterized in that, The preset rule also includes: before calculating the comprehensive matching degree, using the Softmax function to normalize the weights of all features participating in the current matching calculation, so that the sum of the weights of all features is 1.

5. The patient master index management method based on big data algorithms according to claim 1, characterized in that, The calculation of the comprehensive matching degree between the multi-dimensional features of the patient and the existing EMPI in the patient master index database based on feature weights includes: obtaining feature matching values ​​based on the multi-dimensional features of the patient to be identified and the patient master index database; obtaining the comprehensive matching degree based on the feature matching values ​​and the effective weights corresponding to the matching features, the calculation expression of the comprehensive matching degree is as follows: comprehensive matching degree = Σ(feature matching value × feature weight) / Σ feature weight; comparing the comprehensive matching degree with a dynamic threshold, if the comprehensive matching degree is greater than or equal to the dynamic threshold, it is automatically associated with the existing EMPI, if the comprehensive matching degree is less than the dynamic threshold and less than the second threshold, it is registered as a new EMP.

6. The patient master index management method based on big data algorithms according to claim 5, characterized in that, The expression for the dynamic threshold is as follows: ;in, The dynamic threshold after the k-th successful match. Minimum threshold : Initial threshold, k: Number of consecutive successful matches; Threshold adjustment coefficient; Historical misjudgment rate.

7. The patient master index management method based on big data algorithms according to claim 1, characterized in that, The social attributes include name, ID number, contact number, and address; the biological attributes include gender, date of birth, blood type, height, and weight; and the medical attributes include health card number, medical history keywords, surgical history, and diagnostic records.

8. The patient master index management method based on big data algorithms according to claim 4, characterized in that, The periodic merging of duplicate EMPIs and updating the patient master index database includes: periodically traversing the patient master index database, calculating the overall matching degree between different EMPIs, and selecting EMPI pairs with an overall matching degree not lower than the merging threshold as candidate merging pairs; based on the candidate merging pairs, constructing an association graph with EMPIs as nodes and matching relationships as edges, and identifying mutually related EMPI sets through a graph traversal algorithm to form association clusters to be processed; for each association cluster, calculating and comparing the information integrity score based on each EMPI within the cluster, and determining the EMPI with the highest integrity score as the core EMPI; for each other EMPI within the association cluster except the core EMPI, its information integrity score is determined as the core EMPI. All associated patient data are migrated to the core EMPI; if each EMPI other than the core EMPI in the associated cluster has a different value for the same feature as the core EMPI, the final value of that feature in the core EMPI is determined according to a preset conflict arbitration rule, which includes at least one of the following priority orders: feature values ​​collected by automated equipment take precedence over manually entered feature values; feature values ​​collected more recently take precedence over feature values ​​collected earlier; the EMPI status of the data migration completed is marked as merged, and it is removed from the valid index set of the patient master index database, while recording its merging relationship with the core EMPI; the patient master index database is updated.

9. The patient master index management method based on big data algorithms according to claim 1, characterized in that, It also includes generating a temporary EMPI for patients whose social attribute characteristics cannot be obtained. Specifically, this includes: generating a temporary EMPI based on the patient's biological attribute characteristics; after the temporary EMPI is supplemented with social attribute characteristics, performing an association operation, which includes: performing precise matching using the social attribute characteristics as key identifiers; performing multi-dimensional feature cross-validation based on the biological attribute characteristics stored in the temporary EMPI and the characteristics in the patient's main index; and according to the validation results, if the validation passes, migrating and associating all medical data under the temporary EMPI with the corresponding EMPI; if the validation fails, registering the temporary EMPI as a normal EMPI and entering it into the patient's main index.

10. A patient master index management system based on big data algorithms, employing the method described in any one of claims 1 to 9, characterized in that, include: The data acquisition module is used to acquire raw data from multiple historical patients and multi-dimensional characteristics of the patient to be identified. The main index library construction and maintenance module is used to construct and maintain a patient main index library containing multi-dimensional features based on the original data of the historical patients; the multi-dimensional features include social attribute features, biological attribute features, and medical attribute features; feature The weight calculation module is used to calculate the feature weight of each feature in the multi-dimensional features based on preset rules; the patient matching and registration module is used to calculate the comprehensive matching degree between the multi-dimensional features of the patient to be identified and the existing EMPIs in the patient master index based on the feature weights, and to associate the patient to be identified with the existing matching EMPIs or register the patient as a new EMPI according to the matching results. The EMPI merge update module is used to update the patient master index by merging duplicate EMPIs in the patient master index.

Images