Medical information card data identification system for hospital information department

Abstract

The invention relates to the technical field of data identification, in particular to a medical information card data identification system for a hospital information department. Data streams of a hospital information system are integrated, and the current medical path stage of a patient is tracked and identified in real time; loading a corresponding identification module for the identified path stage, and preferentially processing and identifying the identity and insurance information of the patient in the registration stage; the medical history, the diagnosis result and the medicine information are preferentially processed and recognized in the diagnosis stage, and each module is optimized according to the requirements of the stage; when the path of the patient is changed, for example, when the examination stage is switched to the treatment stage, the system adjusts the identification logic in real time, and automatically loads a data processing rule suitable for the current stage; meanwhile, linkage with an electronic medical record system of a hospital is achieved, the self-adaptive recognition method is used for accurately recognizing different path stages, and the accuracy and efficiency of data processing are improved; meanwhile, a dynamic adjustment mechanism enables the system to flexibly cope with changes of patient paths, and data redundancy and recognition errors are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of data identification, and in particular to a medical information card data identification system for a hospital information department. Background Art

[0002] In today's society, almost all hospitals are equipped with information systems such as HIS, LIS, PACS, CIS, etc.; these systems cover every aspect from patient registration, hospitalization to examination, medication, and even extend to the fields of health management and scientific research and teaching, realizing the initial informatization and digitization of various medical and management services in hospitals; therefore, all data generated by patients during the diagnosis and treatment process are recorded and circulated in these systems; these data not only reflect the patient's health status and disease manifestations, but also provide doctors with evidence-based basis and decision-making support; it can be said that data has become an indispensable core element in modern medical practice, and the hospital information department is responsible for the processing of such data.

[0003] When patients visit a hospital for treatment, they generally go through multiple processes, including registration, examination, diagnosis, treatment, hospitalization and settlement. During these stages in the hospital's information management process, the demand and processing requirements for medical information card data are different. For example, when registering, the demand point is the patient's identity and medical insurance information, while in the diagnosis and treatment stages, it is mainly the patient's detailed medical history, drug allergy history and treatment plan; at the same time, the data processed by the hospital information department comes from multiple different systems, including medical insurance systems, electronic medical record systems EMR, laboratory information management systems LIS, etc. There are differences in data formats, standards and contents of different systems, which leads to inconsistencies when identifying information card data.

[0004] The current medical information card data recognition method adopts a single recognition method, such as the OCR model based on template matching, the recognition system based on the rule engine, etc., and uses the same processing logic for all stages. This recognition method lacks specificity and cannot effectively distinguish the data requirements of different stages, resulting in insufficient accuracy and efficiency of information recognition. For example, in the diagnosis stage, the patient's medical history is irrelevant to the previous registration information. At this stage, only the patient's medical history needs to be processed, and the registration information does not need additional processing as long as the information is effectively associated. However, the existing unified processing method processes all information, and some information is repeated in the previous and subsequent stages and is irrelevant. It is impossible to effectively filter and prioritize important data, which will lead to data redundancy or interference; and for the processing process of different data in the hospital information department, there is a lack of a cross-system data consistency verification mechanism, and the data between the systems cannot be effectively integrated, resulting in inconsistent information. Specifically, if the patient's identity information in the medical insurance system does not completely match the information in the electronic medical record, it will lead to recognition errors or data loss, and the existing solution relies too much on manual data verification, which is inefficient and prone to errors. Summary of the invention

[0005] To this end, the present invention provides a medical information card data identification system for a hospital information department, which is used to overcome the problem in the prior art that the needs and processing requirements for medical information card data are different at different stages of the patient's medical treatment process, and a single identification method lacks specificity and cannot effectively distinguish the data needs at different stages; at the same time, the data processed by the hospital information department comes from multiple different systems, and there is a lack of a cross-system data consistency verification mechanism. The data between systems cannot be effectively integrated, which easily leads to information inconsistency.

[0006] To achieve the above object, the present invention provides a medical information card data recognition system for a hospital information department, comprising:

[0007] The dynamic path tracking module integrates the data stream of the hospital information system HIS in real time to dynamically track the patient's medical path; it identifies and outputs the patient's current medical stage information, including registration, diagnosis, examination, treatment, and settlement;

[0008] The phase identification module automatically loads the identification logic and algorithm corresponding to the phase based on the current phase information output by the dynamic path tracking module. In each medical phase, data closely related to the needs of the current phase are processed first. For example, in the registration phase, patient identity and medical insurance information are processed; in the diagnosis phase, medical history, diagnosis results and drug information are processed, and phase data identification results are output.

[0009] Real-time adjustment module: When the dynamic path tracking module detects changes in the patient's medical stage, it adjusts the recognition logic in real time, uninstalls the recognition module of the completed stage, and loads the data processing rules corresponding to the new stage, so that the system always optimizes data processing for the current stage; the real-time adjustment module is linked with the electronic medical record system EMR to obtain real-time feedback and further optimize the accuracy of the recognition module;

[0010] The real-time adjustment module outputs the adjusted recognition logic and recognition results, so that the system can better recognize the staged medical process;

[0011] The multi-source data aggregation module aggregates patient-related data from different source systems, performs standardized processing and data cleaning on the aggregated data, and unifies the data format and coding standards;

[0012] The anomaly detection module performs real-time consistency verification based on the data provided by the multi-source data aggregation module.

[0013] Furthermore, in the dynamic path tracking module, the real-time tracking method is:

[0014] To model the data stream time series, the patient medical activity data obtained from the hospital information system is regarded as a time series S(t), where t is the timestamp, indicating the time point of the data stream, S(t) = {a1(t), a2(t), ..., a n (t)}, S(t) represents all medical activities at time t, a i (t) represents the state vector of the i-th medical activity, i refers to 1, 2, ..., n in the formula;

[0015] a i (t) is expressed as: a i (t) = {ID i ,T i ,P i ,C i ,L i}, where ID i represents the unique identifier of the patient, T i Indicates the type of medical activity (such as registration, diagnosis, examination, etc.), P i Indicates the relevant participants in medical activities (such as doctors and nurses), C i Indicates the specific content of medical activities (such as diagnosis results, treatment recommendations), L i Indicates the location where medical activities take place (such as department, ward);

[0016] In order to identify the medical stage of the patient, a dynamic state transition matrix M(t) is defined to describe the probability of the patient transitioning from one medical activity to another at different time points. Let the element m of the matrix M(t) be ij (t) represents the transition probability from medical activity i to medical activity j:

[0017] Where m ij (t) represents the probability that the patient transfers from activity i to activity j at time t, which is defined as: count(a i →a j ) represents the actual number of transfers from activity i to activity j, represents the total number of times the patient is transferred from activity i to all possible target activities k, where k represents all possible target activities and n represents the total number of medical activity types;

[0018] Through the state transition matrix M(t) and the current medical activity state S(t), the Markov chain is used to predict the next possible medical stage and identify the current stage C(t). For the current time point t, the patient's stage prediction is:

[0019] Where P(a i(t)) means that activity a is observed at time t i (t), C(t) represents the most likely medical stage predicted, corresponding to the stage the patient is currently in.

[0020] Furthermore, in the stage recognition module, the recognition logic matching the current stage is loaded according to the current medical stage information output by real-time tracking:

[0021] Define the adaptive loading function L(C t ), L(C t ) includes M1, M2, ..., M5 different medical stage identification modules, M1, M2, ..., M5 corresponding to the corresponding identification modules loaded by the currently identified medical stage C(t), including registration, diagnosis, examination, treatment and settlement;

[0022] Based on historical data feedback, a dynamic optimization function O(t) is defined. O(t) performs performance optimization every time a module is loaded:

[0023] Where W(t) is the time-weighted matrix, which gives higher weights to data at more recent time points. E(t,θ) is the error matrix, which reflects the deviation between the current module performance and actual requirements. The deviation is minimized by adjusting the parameter θ. t0 represents the time point when the module performance optimization starts.

[0024] Furthermore, the real-time adjustment module is linked with the electronic medical record system EMR and other related systems to obtain system feedback in real time and dynamically adjust the recognition logic:

[0025] The feedback receiving function is defined as F(t), which is used to receive feedback signals from the electronic medical record system EMR and other systems:

[0026] F(t)={f1(t),f2(t),...,f m (t)}, where f i (t) represents the importance of the feedback source (such as EMR) from the ith source, i refers to 1, 2, ... m in the formula,

[0027] m represents the number of feedback sources;

[0028] According to the feedback signal, a dynamic adjustment function D(t) is defined to update the recognition logic in real time:

[0029] where α i is the weight of the feedback signal, indicating the importance of different feedback sources, ▽ θ O(t) represents the gradient relative to the parameter θ, which guides how to adjust the recognition logic to best adapt to the current medical stage;

[0030] This effectively integrates HIS data streams, identifies and predicts patients’ medical stages in real time, and dynamically loads recognition logic that matches the current stage.

[0031] Furthermore, the source systems include the medical insurance system, the electronic medical record system EMR, and the laboratory information management system LIS;

[0032] The multi-source data aggregation module outputs standardized multi-source data.

[0033] Furthermore, in the multi-source data aggregation module, the standardized processing method is as follows:

[0034] For each source system S i , extract the time series data stream D i (t), where i represents the index of the system and is defined as:

[0035] D i (t) = {d i1 (t),d i2 (t),...,d im (t)}, where t is the timestamp, indicating the time point of data extraction, d ij (t) represents the time t from system S i The jth data record extracted from , j refers to 1, 2, ..., m in the formula;

[0036] Mapping data from different systems to a common data representation format M i (t), the mapping function is defined as φ i :

[0037] M i (t) = φ i (D i (t)),φ i Represents the mapping function from the source system data format to the common format;

[0038] Standardize the aggregated data and define the standardization function σ i Convert the data formats of each system into a unified format F(t):

[0039] F(t)=σ i (M i (t)), where σ i It represents the format standardization function, which is used to convert the data formats of different systems into a unified standard format F(t). F(t) represents the standardized data format, which is uniformly expressed as F(t) = {f1(t), f2(t), ..., f m (t)};

[0040] Use the encoding to normalize the matrix C iStandardize data encoding for each source system:

[0041] E i (t) = C i F(t), where C i represents the coding standardization matrix, which maps the coding of different systems to a unified coding standard. i (t) represents the standardized coded data, denoted as E i (t) = {e i1 (t),e i2 (t),...,e im (t)};

[0042] The semantic normalization function ψ is used to unify the semantics of the data:

[0043] S(t)=ψ(E i (t)), where ψ represents the semantic standardization function, which standardizes the semantics in different systems into a unified semantic representation, and S(t) represents the standardized semantic data, which can be expressed as S(t) = {s1(t), s2(t), ..., s m (t)}.

[0044] Furthermore, in the multi-source data aggregation module, the data cleaning method is as follows:

[0045] Use the data cleaning function to remove redundant and inconsistent data, and define S'(t) as the cleaned data set;

[0046] The check function is defined as the consistency check function χ, and χ is used to check the consistency of the data, expressed as χ(S'(t)). The output result returns a Boolean value to check whether the data is consistent. If not, the correction mechanism is automatically triggered or an alarm is sent to the administrator.

[0047] Furthermore, the consistency check method in the anomaly detection module is:

[0048] When identifying medical information card data, automatically compare it with data from different systems;

[0049] When inconsistencies are found in data across systems, anomaly detection is triggered, problematic data is flagged, and alerts are sent to information department staff;

[0050] The anomaly detection module automatically corrects errors based on historical data and rule base.

[0051] Furthermore, in the anomaly detection module, a correction mechanism is automatically triggered when data inconsistency is detected:

[0052] The anomaly detection module marks all detected inconsistent data and records the inconsistency type of each piece of data, including which data has mismatched encoding formats, which data has inconsistent semantics, and which data has incorrect formats;

[0053] Classify inconsistent data according to the tags and assign a specific inconsistency type to each piece of data;

[0054] Each inconsistency type corresponds to a different correction strategy. For coding inconsistencies, recoding or format conversion is performed; for semantic inconsistencies, reinterpretation or manual review is performed;

[0055] For each inconsistent data, a corresponding correction strategy is selected from the strategy library according to its type; by evaluating the effects of different strategies, a strategy to repair the data is automatically selected.

[0056] Furthermore, the data identification system workflow includes:

[0057] Step S1, start, the dynamic path tracking module identifies the patient's current medical stage and outputs stage information;

[0058] Step S2, identification, the phase identification module loads the corresponding identification logic and processes the phase data according to the output of the path tracking module;

[0059] Step S3, adjustment and optimization, when the patient's medical stage changes, the real-time adjustment module adjusts the recognition logic;

[0060] Step S4, data aggregation and standardization, the multi-source data aggregation module aggregates and standardizes data from different systems;

[0061] Step S5, consistency check, the anomaly detection module checks the aggregated data, ensures cross-system data consistency, and performs anomaly detection and correction.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] The present invention integrates the data flow of the hospital information system HIS, tracks and identifies the patient's current medical path stage (registration, diagnosis, settlement) in real time; loads the corresponding identification module for the identified path stage, prioritizes the processing and identification of the patient's identity and insurance information in the registration stage; prioritizes the identification of medical history, diagnosis results and drug information in the diagnosis stage, and each module is optimized according to the needs of the stage.

[0064] In the present invention, when the patient's path changes, such as when the patient switches from the examination stage to the treatment stage, the system adjusts the recognition logic in real time and automatically loads the data processing rules suitable for the current stage; at the same time, it is linked with the hospital's electronic medical record system EMR, and the recognition model continuously obtains feedback from the electronic medical record system. The adaptive recognition method accurately recognizes different path stages, thereby improving the accuracy and efficiency of data processing; at the same time, the dynamic adjustment mechanism enables the system to flexibly respond to changes in the patient's path and reduce data redundancy and recognition errors.

[0065] The present invention adopts a cross-system consistency verification mechanism for multi-source data, aggregates relevant patient data from different source systems (medical insurance system, EMR, LIS), performs standardized processing and data cleaning during the data aggregation process, and unifies the data formats and coding standards of different systems.

[0066] The present invention, when identifying the medical information card data, automatically compares it with the aggregated data in real time to detect whether there is any inconsistency; when the patient's medical insurance information is identified, the system will immediately cross-verify with the relevant records in the medical insurance system and EMR to ensure the consistency and accuracy of the data. When the system finds inconsistencies in cross-system data, it will automatically trigger the anomaly detection module, mark the data entries that may have problems, and send an alarm to the information department staff.

[0067] The present invention integrates a verification mechanism for multi-source data, so that the hospital information department can significantly improve the accuracy of data identification, avoid the problem of data islands, reduce the workload and error rate of manual comparison, and ensure the consistency of cross-system data through real-time verification and intelligent correction mechanisms. This solves the problem in the prior art that the needs and processing requirements for medical information card data at different stages of the patient's medical treatment process are different, and a single identification method lacks specificity and cannot effectively distinguish the data needs at different stages; at the same time, the data processed by the hospital information department comes from multiple different systems, lacks a cross-system data consistency verification mechanism, and the data between systems cannot be effectively integrated, which easily leads to the problem of inconsistent information. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 This is a schematic diagram of the framework structure of the medical information card data identification system for the hospital information department of the present invention. DETAILED DESCRIPTION

[0069] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0070] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0071] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0072] Example 1

[0073] See also Figure 1 The present invention provides a medical information card data identification system for hospital information department, comprising:

[0074] The dynamic path tracking module integrates the data stream of the hospital information system HIS in real time to dynamically track the patient's medical path; it identifies and outputs the patient's current medical stage information, including registration, diagnosis, examination, treatment, and settlement;

[0075] In the dynamic path tracking module, the real-time tracking method is:

[0076] To model the data stream time series, the patient medical activity data obtained from the hospital information system is regarded as a time series S(t), where t is the timestamp, indicating the time point of the data stream, S(t) = {a1(t), a2(t), ..., a n (t)}, S(t) represents all medical activities at time t, a i (t) represents the state vector of the i-th medical activity, i refers to 1, 2, ..., n in the formula;

[0077] a i (t) is expressed as: a i (t) = {ID i ,T i ,P i ,C i ,L i}, where ID i represents the unique identifier of the patient, T i Indicates the type of medical activity (such as registration, diagnosis, examination, etc.), P i Indicates the relevant participants in medical activities (such as doctors and nurses), Ci Indicates the specific content of medical activities (such as diagnosis results, treatment recommendations), L i Indicates the location where medical activities take place (such as department, ward);

[0078] In order to identify the medical stage of the patient, a dynamic state transition matrix M(t) is defined to describe the probability of the patient transitioning from one medical activity to another at different time points. Let the element m of the matrix M(t) be ij (t) represents the transition probability from medical activity i to medical activity j:

[0079] Where m ij (t) represents the probability that the patient transfers from activity i to activity j at time t, which is defined as: count(a i →a j ) represents the actual number of transfers from activity i to activity j, represents the total number of times the patient is transferred from activity i to all possible target activities k, where k represents all possible target activities and n represents the total number of medical activity types;

[0080] Through the state transition matrix M(t) and the current medical activity state S(t), the Markov chain is used to predict the next possible medical stage and identify the current stage C(t). For the current time point t, the patient's stage prediction is:

[0081] Where P(a i (t)) means that activity a is observed at time t i (t), C(t) represents the most likely medical stage predicted, corresponding to the stage the patient is currently in;

[0082] Specifically, by real-time integration of hospital information system HIS data streams, accurately identifying and tracking patients' activities at different medical stages, and using time series modeling and dynamic state transition matrices, effectively predicting and identifying patients' current medical stages, accurate stage identification enables the system to dynamically load corresponding identification logic according to data requirements at different stages, thereby avoiding redundant data processing and information interference problems under traditional unified processing methods, making the data identification system more efficient in data processing and greatly improving identification accuracy, especially when the patient's path changes, the identification logic can be quickly adjusted to adapt to new stage requirements.

[0083] The phase identification module automatically loads the identification logic and algorithm corresponding to the phase based on the current phase information output by the dynamic path tracking module. In each medical phase, data closely related to the needs of the current phase are processed first. For example, in the registration phase, patient identity and medical insurance information are processed; in the diagnosis phase, medical history, diagnosis results and drug information are processed, and phase data identification results are output.

[0084] In the stage recognition module, the recognition logic matching the current stage is loaded according to the current medical stage information output by real-time tracking:

[0085] Define the adaptive loading function L(C t ), L(C t ) includes M1, M2, ..., M5 different medical stage identification modules, M1, M2, ..., M5 corresponding to the corresponding identification modules loaded by the currently identified medical stage C(t), including registration, diagnosis, examination, treatment and settlement;

[0086] Based on historical data feedback, a dynamic optimization function O(t) is defined. O(t) performs performance optimization every time a module is loaded:

[0087] Where W(t) is the time weighting matrix, which gives higher weight to data at more recent time points. E(t,θ) is the error matrix, which reflects the deviation between the current module performance and the actual demand. The deviation is minimized by adjusting the parameter θ. t0 represents the time point when the module performance optimization starts.

[0088] Specifically, the stage identification module loads specific identification logic based on the information of the current medical stage, and gives priority to processing data that is closely related to the current stage; for example, in the registration stage, the system centrally processes the patient's identity and medical insurance information, while in the diagnosis stage, it focuses on processing medical history and treatment-related information. Targeted processing significantly improves the efficiency of data identification, enables key data to be given priority in different processes, and reduces unnecessary data redundancy.

[0089] Real-time adjustment module: When the dynamic path tracking module detects changes in the patient's medical stage, it adjusts the recognition logic in real time, uninstalls the recognition module of the completed stage, and loads the data processing rules corresponding to the new stage, so that the system always optimizes data processing for the current stage; the real-time adjustment module is linked with the electronic medical record system EMR to obtain real-time feedback and further optimize the accuracy of the recognition module;

[0090] The real-time adjustment module outputs the adjusted recognition logic and recognition results, so that the system can better recognize the staged medical process;

[0091] The real-time adjustment module is linked with the electronic medical record system EMR and other related systems to obtain system feedback in real time and dynamically adjust the recognition logic:

[0092] The feedback receiving function is defined as F(t), which is used to receive feedback signals from the electronic medical record system EMR and other systems:

[0093] F(t)={f1(t),f2(t),...,f m (t)}, where f i (t) represents the importance of the feedback source (such as EMR) from the ith source, i refers to 1, 2, ... m in the formula,

[0094] m represents the number of feedback sources;

[0095] According to the feedback signal, a dynamic adjustment function D(t) is defined to update the recognition logic in real time:

[0096] where α i is the weight of the feedback signal, indicating the importance of different feedback sources, ▽ θ O(t) represents the gradient relative to the parameter θ, which guides how to adjust the recognition logic to best adapt to the current medical stage;

[0097] This effectively integrates HIS data streams, identifies and predicts the patient's medical stage in real time, and dynamically loads the recognition logic that matches the current stage;

[0098] Specifically, as the patient's stage in the medical pathway changes, the system adjusts the recognition logic in real time. When the patient enters the treatment stage from the examination stage, the system automatically loads the recognition rules applicable to the treatment stage to avoid processing data irrelevant to the current stage, thereby improving recognition efficiency and accuracy. The real-time adjustment module is linked to the electronic medical record system (EMR) to continuously optimize the recognition modules at each stage by continuously obtaining feedback, making them more suitable for the data processing needs of the current stage.

[0099] The multi-source data aggregation module aggregates patient-related data from different source systems, performs standardized processing and data cleaning on the aggregated data, and unifies the data format and coding standards;

[0100] Source systems include medical insurance system, electronic medical record system EMR, laboratory information management system LIS;

[0101] The multi-source data aggregation module outputs standardized multi-source data;

[0102] In the multi-source data aggregation module, the standardization processing method is as follows:

[0103] For each source system S i , extract the time series data stream D i (t), where i represents the index of the system and is defined as:

[0104] D i (t) = {d i1 (t),d i2 (t),...,d im (t)}, where t is the timestamp, indicating the time point of data extraction, d ij (t) represents the time t from system S i The jth data record extracted from , j refers to 1, 2, ..., m in the formula;

[0105] Mapping data from different systems to a common data representation format M i (t), the mapping function is defined as φ i :

[0106] M i (t) = φ i (D i (t)),φ i Represents the mapping function from the source system data format to the common format;

[0107] Standardize the aggregated data and define the standardization function σ i Convert the data formats of each system into a unified format F(t):

[0108] F(t)=σ i (M i (t)), where σ i It represents the format standardization function, which is used to convert the data formats of different systems into a unified standard format F(t). F(t) represents the standardized data format, which is uniformly expressed as F(t) = {f1(t), f2(t), ..., f m (t)};

[0109] Use the encoding to normalize the matrix C i Standardize data encoding for each source system:

[0110] E i (t) = C i F(t), where C i represents the coding standardization matrix, which maps the coding of different systems to a unified coding standard. i (t) represents the standardized coded data, denoted as E i (t) = {e i1 (t),e i2 (t),...,e im (t)};

[0111] The semantic normalization function ψ is used to unify the semantics of the data:

[0112] S(t)=ψ(E i (t)), where ψ represents the semantic standardization function, which standardizes the semantics in different systems into a unified semantic representation, and S(t) represents the standardized semantic data, which can be expressed as S(t) = {s1(t), s2(t), ..., s m (t)};

[0113] In the multi-source data aggregation module, the data cleaning method is as follows:

[0114] Use the data cleaning function to remove redundant and inconsistent data, and define S'(t) as the cleaned data set;

[0115] The check function is defined as consistency check function χ, and the consistency of data is checked using χ, which is expressed as χ(S'(t)). The output result returns a Boolean value to check whether the data is consistent. If not, the correction mechanism is automatically triggered or an alarm is sent to the administrator.

[0116] Specifically, the multi-source data aggregation module uniformly processes data from different source systems such as the medical insurance system, EMR, and laboratory information management system LIS to ensure the consistency of data format, coding standards, and semantics, so that the data recognition system can efficiently integrate multi-source data and reduce the problem of inconsistency in data across systems.

[0117] The anomaly detection module performs real-time consistency verification based on the data provided by the multi-source data aggregation module;

[0118] The consistency check method in the anomaly detection module is:

[0119] When identifying medical information card data, automatically compare it with data from different systems;

[0120] When inconsistencies are found in data across systems, anomaly detection is triggered, problematic data is flagged, and alerts are sent to information department staff;

[0121] The anomaly detection module automatically corrects errors based on historical data and rule base;

[0122] In the anomaly detection module, when data inconsistency is detected, the correction mechanism is automatically triggered:

[0123] The anomaly detection module marks all detected inconsistent data and records the inconsistency type of each piece of data, including which data has mismatched encoding formats, which data has inconsistent semantics, and which data has incorrect formats;

[0124] Classify inconsistent data according to the tags and assign a specific inconsistency type to each piece of data;

[0125] Each inconsistency type corresponds to a different correction strategy. For coding inconsistencies, recoding or format conversion is performed; for semantic inconsistencies, reinterpretation or manual review is performed;

[0126] For each inconsistent data, a corresponding correction strategy is selected from the strategy library according to its type; by evaluating the effects of different strategies, a strategy to repair the data is automatically selected;

[0127] Specifically, the anomaly detection module ensures the accuracy and consistency of data through real-time consistency verification and automatic correction mechanisms. When inconsistency in cross-system data is detected, the correction mechanism is automatically triggered to quickly repair inconsistent data through steps such as marking, classification, and selection of the best correction strategy, thereby reducing the need for manual intervention and improving data processing efficiency.

[0128] The overall workflow of the medical information card data identification system for hospital information departments includes:

[0129] Step S1, start, the dynamic path tracking module identifies the patient's current medical stage and outputs stage information;

[0130] Step S2, identification, the phase identification module loads the corresponding identification logic and processes the phase data according to the output of the path tracking module;

[0131] Step S3, adjustment and optimization, when the patient's medical stage changes, the real-time adjustment module adjusts the recognition logic;

[0132] Step S4, data aggregation and standardization, the multi-source data aggregation module aggregates and standardizes data from different systems;

[0133] Step S5, consistency check, the anomaly detection module checks the aggregated data, ensures cross-system data consistency, and performs anomaly detection and correction.

[0134] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A medical information card data recognition system for hospital information department, characterized in that: include: Dynamic path tracking module, real-time integration of hospital information system HIS data stream, dynamic tracking of patients' medical path; Identify and output the patient's current medical stage information, including registration, diagnosis, examination, treatment, and settlement; The stage recognition module automatically loads the recognition logic and algorithm corresponding to the stage based on the current stage information output by the dynamic path tracking module; in each medical stage, data closely related to the needs of the current stage are processed first; Real-time adjustment module: When the dynamic path tracking module detects changes in the patient's medical stage, it adjusts the recognition logic in real time, uninstalls the recognition module of the completed stage, and loads the data processing rules corresponding to the new stage; the real-time adjustment module is linked with the electronic medical record system EMR; The multi-source data aggregation module aggregates patient-related data from different source systems and performs standardization and data cleaning on the aggregated data; The anomaly detection module performs real-time consistency verification based on the data provided by the multi-source data aggregation module.

2. A medical information card data recognition system for hospital information department according to claim 1, characterized in that: In the dynamic path tracking module, the real-time tracking method is: To model the data stream time series, the patient medical activity data obtained from the hospital information system is regarded as a time series S(t), where t is the timestamp, indicating the time point of the data stream, S(t) = {a1(t), a2(t), ..., a n (t)}, S(t) represents all medical activities at time t, a i (t) represents the state vector of the i-th medical activity, i refers to 1, 2, ..., n in the formula; a i (t) is expressed as: a i (t) = {ID i ,T i ,P i ,C i ,L i }, where ID i represents the unique identifier of the patient, T i Indicates the type of medical activity, P i Indicates the relevant participants in medical activities, C i Indicates the specific content of medical activities, L i Indicates the location where the medical activity takes place; Define the dynamic state transition matrix M(t) to describe the probability of a patient transitioning from one medical activity to another at different time points. Let the element m of the matrix M(t) be ij (t) represents the transition probability from medical activity i to medical activity j: Where m ij (t) represents the probability that the patient transfers from activity i to activity j at time t, which is defined as: count(a i →a j ) represents the actual number of transfers from activity i to activity j, represents the total number of times the patient is transferred from activity i to all possible target activities k, where k represents all possible target activities and n represents the total number of medical activity types; Through the state transition matrix M(t) and the current medical activity state S(t), the Markov chain is used to predict the next possible medical stage and identify the current stage C(t). For the current time point t, the patient's stage prediction is: Where P(a i (t)) means that activity a is observed at time t i (t), C(t) represents the most likely medical stage predicted, corresponding to the stage the patient is currently in.

3. A medical information card data recognition system for hospital information department according to claim 2, characterized in that: In the stage recognition module, the recognition logic matching the current stage is loaded according to the current medical stage information output by real-time tracking: Define the adaptive loading function L(C t ), L(C t ) includes M1, M2, ..., M5 different medical stage identification modules, M1, M2, ..., M5 corresponding to the corresponding identification modules loaded by the currently identified medical stage C(t), including registration, diagnosis, examination, treatment and settlement; Based on historical data feedback, a dynamic optimization function O(t) is defined. O(t) performs performance optimization every time a module is loaded: Where W(t) is the time-weighted matrix, which gives higher weights to data at more recent time points. E(t,θ) is the error matrix, which reflects the deviation between the current module performance and actual requirements. The deviation is minimized by adjusting the parameter θ. t0 represents the time point when the module performance optimization starts.

4. A medical information card data recognition system for hospital information department according to claim 3, characterized in that: The real-time adjustment module is linked with the electronic medical record system EMR and other related systems to obtain system feedback in real time and dynamically adjust the recognition logic: The feedback receiving function is defined as F(t), which is used to receive feedback signals from the electronic medical record system EMR and other systems: F(t)={f1(t),f2(t),...,f m (t)}, where f i (t) represents the importance of the feedback source from the i-th source, i refers to 1, 2, ... m in the formula, m represents the number of feedback sources; According to the feedback signal, a dynamic adjustment function D(t) is defined to update the recognition logic in real time: where α i is the weight of the feedback signal, indicating the importance of different feedback sources, represents the gradient with respect to the parameter θ.

5. A medical information card data recognition system for hospital information department according to claim 4, characterized in that: Source systems include medical insurance system, electronic medical record system EMR, laboratory information management system LIS; The multi-source data aggregation module outputs standardized multi-source data.

6. A medical information card data recognition system for hospital information department according to claim 5, characterized in that: In the multi-source data aggregation module, the standardization processing method is as follows: For each source system S i , extract the time series data stream D i (t), where i represents the index of the system and is defined as: D i (t) = {d i1 (t),d i2 (t),...,d im (t)}, where t is the timestamp, indicating the time point of data extraction, d ij (t) represents the time t from system S i The jth data record extracted from , j refers to 1, 2, ..., m in the formula; Mapping data from different systems to a common data representation format M i (t), the mapping function is defined as φ i : M i (t) = φ i (D i (t)),φ i Represents the mapping function from the source system data format to the common format; Standardize the aggregated data and define the standardization function σ i Convert the data formats of each system into a unified format F(t): F(t)=σ i (M i (t)), where σ i It represents the format standardization function, which is used to convert the data formats of different systems into a unified standard format F(t). F(t) represents the standardized data format, which is uniformly expressed as F(t) = {f1(t), f2(t), ..., f m (t)}; Use the encoding to normalize the matrix C i Standardize data encoding for each source system: E i (t) = C i F(t), where C i represents the coding standardization matrix, which maps the coding of different systems to a unified coding standard. i (t) represents the standardized coded data, denoted as E i (t) = {e i1 (t),e i2 (t),...,e im (t)}; The semantic normalization function ψ is used to unify the semantics of the data: S(t)=ψ(E i (t)), where ψ represents the semantic standardization function, which standardizes the semantics in different systems into a unified semantic representation, and S(t) represents the standardized semantic data, which can be expressed as S(t) = {s1(t), s2(t), ..., s m (t)}.

7. A medical information card data recognition system for hospital information department according to claim 6, characterized in that: In the multi-source data aggregation module, the data cleaning method is as follows: Use the data cleaning function to remove redundant and inconsistent data, and define S'(t) as the cleaned data set; The check function is defined as the consistency check function χ, and χ is used to check the consistency of the data, expressed as χ(S'(t)). The output result returns a Boolean value to check whether the data is consistent. If not, the correction mechanism is automatically triggered or an alarm is sent to the administrator.

8. A medical information card data recognition system for hospital information department according to claim 7, characterized in that: The consistency check method in the anomaly detection module is: When identifying medical information card data, automatically compare it with data from different systems; When inconsistencies are found in data across systems, anomaly detection is triggered, problematic data is flagged, and alerts are sent to information department staff; The anomaly detection module automatically corrects errors based on historical data and rule base.

9. A medical information card data recognition system for hospital information department according to claim 8, characterized in that: In the anomaly detection module, when data inconsistency is detected, the correction mechanism is automatically triggered: The anomaly detection module marks all detected inconsistent data and records the inconsistency type of each piece of data, including which data has mismatched encoding formats, which data has inconsistent semantics, and which data has incorrect formats; Classify inconsistent data according to the tags and assign a specific inconsistency type to each piece of data; Each inconsistency type corresponds to a different correction strategy; For each inconsistent data, a corresponding correction strategy is selected from the strategy library according to its type.

10. A medical information card data recognition system for hospital information department according to claim 9, characterized in that: The data identification system workflow includes: Step S1, start, the dynamic path tracking module identifies the patient's current medical stage and outputs stage information; Step S2, identification, the phase identification module loads the corresponding identification logic according to the output of the path tracking module and processes the phase data; Step S3, adjustment and optimization, when the patient's medical stage changes, the real-time adjustment module adjusts the recognition logic; Step S4, data aggregation and standardization, the multi-source data aggregation module aggregates and standardizes data from different systems; Step S5, consistency check, the anomaly detection module checks the aggregated data and performs anomaly detection and correction.

Images