Intelligent quality evaluation and data completion method and system for medical bill data

Abstract

The present application relates to a method and system for intelligent quality evaluation and data completion of medical bill data, comprising the following steps: collecting medical bill data of medical billing units, constructing a medical bill data quality evaluation model, and determining evaluation indicators of the medical bill data according to the characteristics of the medical bill data; using the entropy weight method to calculate the weights of the evaluation indicators, and then calculating the final evaluation scores of the quality of the medical bill data of the medical billing units based on the weights; according to the medical bill data quality evaluation scores of the medical billing units, the medical billing units are divided into different levels, which are high quality, medium quality, and low quality respectively; using the constructed medical bill data gender content missing completion model and medical bill data department information completion model, the gender content missing, department information missing, and non-standard department information in the medical bill data of medical billing units with evaluation levels of medium quality and low quality are completed and corrected.

Description

Technical Field

[0001] The present application relates to the field of medical electronic receipts and artificial intelligence technology, and more specifically, to a method and system for intelligent quality evaluation and data completion of medical receipt data. Background Art

[0002] The data quality evaluation of medical bills is an important part of medical informationization. Traditional evaluation methods rely on rules or manual experience evaluation, but now intelligent evaluation models based on machine learning and statistics have become the mainstream. In practical applications, indicator selection and weight determination often rely on expert experience, which will produce subjective bias and affect the accuracy of the results.

[0003] In modern hospital management and medical services, there are problems such as missing department information, irregular department information, and errors in medical electronic receipts. Traditional department attribution methods often rely on manual rules or simple keyword matching, which are prone to errors when processing complex data and cannot efficiently adapt to differences between different hospitals or regions. Therefore, building an intelligent quality evaluation and data completion model for medical receipt data is an urgent problem to be solved.

[0004] Prior art, such as the Chinese patent with patent number "CN118069852A", discloses a data classification prediction method and system with multi-model fusion. It includes collecting and preprocessing labeled text data and classification lists. Use a large language model to enhance the classification description and save information through vectorization technology. Subsequently, the model is used to generate text summaries and perform data enhancement. For classifications with insufficient data, the model is used to synthesize data to meet the minimum data requirements. The fine-tuned BERT model and semantic similarity query are combined to verify each other to ensure correct classification of the text. The text data input by the user is cleaned and desensitized, and the text summary is extracted through the model. These summaries are semantically vectorized, and then the most matching classification is retrieved through the vector database. The invention combines the large language model technology with the traditional BERT text classification technology, utilizes the accuracy of BERT text classification for fixed standard classification, and improves the robustness of the system for unknown classifications through the large language model technology.

[0005] The problems with the above-mentioned prior art are that, although the large language model technology is combined with the traditional BERT text classification technology, the fusion mechanism of the two models may not be perfect, and problems such as unsmooth fusion and mutual interference may occur in actual operation; there is a lack of specific methods and standards for how to balance the accuracy of the BERT model and the robustness of the large language model, which may lead to unstable performance in different scenarios; retrieving the most matching classification through the vector database depends on the performance and accuracy of the vector database. If there is a problem with the database or it is inaccurate, it will directly affect the classification effect of the entire system. Summary of the invention

[0006] In order to solve the above technical problems, the present invention proposes a method and system for monitoring the operating efficiency of a water pump based on digital twin technology.

[0007] The technical solution of the present invention is as follows:

[0008] The present invention proposes a method for intelligent quality evaluation and data completion of medical bill data, comprising the following steps:

[0009] Step S1, collecting medical bill data from medical billing units, building a medical bill data quality evaluation model, and determining evaluation indicators for the medical bill data based on the characteristics of the medical bill data;

[0010] Step S2, using the entropy weight method to calculate the weights of the evaluation indicators, and then calculating the final evaluation score of the medical bill data quality of the medical billing unit based on the weights;

[0011] Step S3, according to the medical bill data quality evaluation scores of the medical billing units, the medical billing units are divided into different levels, and the different levels are high quality, medium quality, and low quality;

[0012] Step S4, constructing a model for completing missing gender content in medical bill data, and completing missing gender content in medical bill data of medical billing units with evaluation levels of medium quality and low quality through feature fusion and random forest model;

[0013] Step S5, construct a model for completing department information of medical bill data, predict corresponding department information through drug information, and complete and correct missing department information and non-standard department information in the medical bill data of medical billing units with medium quality and low quality evaluation levels.

[0014] As a preferred implementation, the evaluation index of the medical bill data is determined according to the characteristics of the medical bill data; the evaluation index content includes: completeness, accuracy, and effectiveness.

[0015] As a preferred implementation, the entropy weight method is used to calculate the weight of each evaluation index, and the specific process is as follows:

[0016] Data cleaning: pre-process the raw data, including deduplication, outlier processing, missing value processing, etc., to ensure the standardization of input data;

[0017] Standardization: In order to eliminate the dimensional differences between different indicators, each data quality indicator is standardized;

[0018] Weight calculation: By calculating the information entropy of the completeness, accuracy and effectiveness of the evaluation indicators, the difference coefficient of each indicator is obtained, thereby determining the weight of each indicator.

[0019] As a preferred implementation, the model for completing the missing gender content in medical bill data is constructed, and the missing gender content in the medical bill data of medical billing units with evaluation levels of medium quality and low quality is completed through feature fusion and random forest model, which includes the following steps:

[0020] Data collection and cleaning: Collect the original data set containing name and gender information from medical bill data to ensure the diversity and representativeness of data sources; process outliers in the data set, remove duplicate data, and standardize the name format;

[0021] Word2Vec word vector: Select the Word2Vec word vector model to vectorize the name. For each name, calculate the word vectors of all its characters and take the average or weighted average to convert the name into a fixed-length vector representation.

[0022] TF-IDF feature extraction: Perform TF-IDF feature extraction on name data; treat each name as a term, calculate the TF-IDF value of each term; generate a TF-IDF feature vector for each name;

[0023] Feature fusion: The vector generated by the Word2Vec word vector model is combined with the term frequency-inverse document frequency TF-IDF feature vector to obtain a comprehensive feature vector containing semantic and statistical information;

[0024] Dataset division: The dataset is divided into training set and test set according to 7:3;

[0025] Training random forest model: Use comprehensive feature vectors and gender labels to train the random forest model; the random forest model integrates multiple decision trees, each of which is trained based on different random samples and comprehensive feature vectors, and the final prediction result is obtained by voting or averaging;

[0026] Model verification and tuning: Use the test set to verify the performance of the model and calculate the accuracy and recall rate indicators; adjust the configuration of the random forest model according to the calculated accuracy and recall rate indicators;

[0027] Predict missing gender: Use the trained random forest model to predict the medical bill data with missing gender and fill in the missing gender labels.

[0028] As a preferred implementation, the constructed medical bill data department information completion model predicts the corresponding department information through drug information, and completes and corrects the missing department information and non-standard department information in the medical bill data of medical billing units with medium and low quality evaluation levels; specifically, it includes the following steps:

[0029] Data collection and cleaning: Extract key field data including drug list, drug item type and department information from the medical bill database; perform regularization cleaning on the data, remove all non-Chinese special characters and notes from the original data, process missing values, outliers, invalid values, and delete duplicate data;

[0030] Department information standardization: Use n-gram and Levenshtein distance to standardize department names and handle spelling errors and variants;

[0031] Drug and project type mapping dictionary construction: Build a mapping dictionary based on drug information and project type;

[0032] Medical bill data processing: secondary sorting based on the importance of item types, secondary sorting of the list content of each item type, and deduplication of text data;

[0033] Text data construction and classification: Construct a text data model to train the department information completion model for medical bill data, use FastText to classify departments, and obtain the trained department information completion model for medical bill data;

[0034] The trained medical bill data department information completion model is used to complete and correct missing department information and non-standard department information in the medical bill data.

[0035] On the other hand, the present invention also provides a medical bill data intelligent quality evaluation and data completion system, including:

[0036] The evaluation model and evaluation index construction module collects medical bill data from medical billing units, builds a medical bill data quality evaluation model, and determines the evaluation index of medical bill data based on the characteristics of medical bill data;

[0037] The data quality score calculation module uses the entropy weight method to calculate the weights of the evaluation indicators, and then calculates the final evaluation score of the medical bill data quality of the medical billing unit based on the weights;

[0038] The medical billing unit classification module divides the medical billing units into different levels according to their medical bill data quality evaluation scores, namely high quality, medium quality and low quality;

[0039] The gender content completion module builds a model for completing the missing gender content in medical bill data. It uses feature fusion and random forest models to complete the missing gender content in the medical bill data of medical billing units with medium and low quality evaluation levels.

[0040] The department information completion and correction module builds a department information completion model for medical bill data, predicts the corresponding department information through drug information, and completes and corrects the missing and non-standard department information in the medical bill data of billing units with medium and low quality evaluation levels.

[0041] As a preferred implementation, the evaluation model and evaluation index building block determine the evaluation index of the medical bill data according to the characteristics of the medical bill data; the evaluation index content includes: completeness, accuracy, and effectiveness.

[0042] As a preferred implementation, the data quality score calculation module uses the entropy weight method to calculate the weight of each evaluation index. The specific process is as follows:

[0043] Data cleaning: pre-process the raw data, including deduplication, outlier processing, missing value processing, etc., to ensure the standardization of input data;

[0044] Standardization: In order to eliminate the dimensional differences between different indicators, each data quality indicator is standardized;

[0045] Weight calculation: By calculating the information entropy of the completeness, accuracy and effectiveness of the evaluation indicators, the difference coefficient of each indicator is obtained, thereby determining the weight of each indicator.

[0046] On the other hand, the present invention also provides an electronic device having a computer program stored thereon, which, when executed by a processor, implements the medical bill data intelligent quality evaluation and data completion method as described in any embodiment of the present invention.

[0047] On the other hand, the present invention also provides a computer-readable medium for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the medical bill data intelligent quality evaluation and data completion method as described in any embodiment of the present invention.

[0048] The present invention has the following beneficial effects:

[0049] 1. Construct a medical bill data quality evaluation model and determine the evaluation indicators based on the characteristics of the medical bill data, so that the evaluation is targeted and scientific and can accurately reflect the actual quality of the medical bill data.

[0050] 2. The entropy weight method is used to calculate the weight of each evaluation index, which avoids the interference of subjective factors and makes the weight determination more objective and reasonable, thereby ensuring the reliability of the final evaluation score.

[0051] 3. The billing units are divided into different levels to facilitate targeted management of billing units with different quality levels. High-quality units can be used as benchmarks, and medium-quality and low-quality units can clarify the direction of improvement and improve management efficiency.

[0052] 4. Build a medical bill data completion model to specifically complete and correct the data of medium-quality and low-quality billing units, thereby improving the accuracy of data completion and effectively improving the quality of medical bill data of these billing units.

[0053] 5. By completing and correcting low-quality data, it helps to improve the quality level of the entire medical bill data and provide a more accurate and reliable data foundation for bill data analysis and billing process optimization in the medical industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0055] Figure 2 The data quality score calculation flow chart of the present invention;

[0056] Figure 3 Flowchart of the model for completing missing gender content in medical bill data;

[0057] Figure 4 Flowchart of the model for completing department information for medical bill data. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0060] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0061] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0062] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0063] Embodiment 1:

[0064] In order to make the purpose, technical solutions and advantages of the present invention more clear, the following will be combined with the specific embodiments of the present application and refer to the attached Figure 1 , clearly and completely describe the technical solution of the present invention.

[0065] In order to solve the problems of the prior art, the present invention provides a method for intelligent quality evaluation and data completion of medical bill data, comprising the following steps:

[0066] Step S1, collecting medical bill data from medical billing units, building a medical bill data quality evaluation model, and determining the evaluation index of the medical bill data according to the characteristics of the medical bill data; specifically comprising the following steps:

[0067] Data preparation:

[0068] Data collection: Extract medical bill data of medical billing units from the bill database.

[0069] Data cleaning: Pre-process the medical bill data of medical billing units, including deduplication, outlier processing, missing value processing, etc., to ensure the standardization of input data.

[0070] Construction of data quality indicator system:

[0071] Determine the evaluation indicators: Based on the characteristics of medical bill data, define statistical indicators that can reflect the completeness, accuracy, and effectiveness of medical bill data. For example, in hospital bills, the distribution of indicators should be defined from the following attributes: billing unit name, payee name, payee code, payee gender, payee age, hospitalization time, and discharge time.

[0072] Completeness: the filling status of each field in the data set and whether there is any missing data.

[0073] Accuracy: whether the data content is consistent with the actual situation, such as whether the payer's gender is male or female.

[0074] Validity: The suitability and usability of the data.

[0075] Step S2, using the entropy weight method to calculate the weight of the evaluation index, and then calculating the final evaluation score of the medical bill data quality of the medical billing unit based on the weight.

[0076] like Figure 2 As shown in the figure, calculating the final evaluation score of the medical bill data quality of the medical billing unit includes the following steps:

[0077] Data cleaning: Preprocess the raw data, including deduplication, outlier processing, missing value processing, etc., to ensure the standardization of the input data.

[0078] Data standardization: In order to eliminate the dimensional differences between different indicators, each data quality indicator is standardized.

[0079] Weight calculation: By calculating the information entropy of each indicator, the difference coefficient of each indicator is obtained, thereby determining the weight of each indicator.

[0080] Step S3, according to the medical billing data quality evaluation scores of the medical billing units, the medical billing units are divided into different levels, and the different levels are high quality, medium quality and low quality.

[0081] By calculating the information entropy of each indicator, the difference coefficient of each indicator is obtained, and the weight of each indicator is determined. Then, the final evaluation score of the data quality of each invoicing unit is obtained based on the weight calculation. Each invoicing unit is divided into different levels, such as high quality, medium quality, and low quality, to facilitate subsequent classification processing and optimization.

[0082] Step S4, construct a model to complete the missing gender content in medical bill data, and use feature fusion and random forest model to complete the missing gender content in the medical bill data of medical billing units with medium and low quality evaluation levels.

[0083] like Figure 3 As shown in the figure, in order to solve the problem of missing payee gender information in medical electronic receipts, a gender attribute completion model based on word vector is proposed, which combines two text representation methods, TF-IDF and Word2Vec, and uses random forest classifier for classification. This method can not only effectively utilize the character-level information in the name, but also capture deeper semantic relationships, thereby improving the accuracy of prediction; it includes the following steps:

[0084] Data collection and cleaning: Collect the original data set containing name and gender information from the medical bill data to ensure the diversity and representativeness of the data source; process outliers in the data set, remove duplicate data, and standardize the name format.

[0085] Word2Vec word vector: Select the Word2Vec word vector model to vectorize the name. For each name, calculate the word vectors of all its characters and take the average or weighted average to convert the name into a vector representation of a fixed length.

[0086] TF-IDF feature extraction: Perform TF-IDF feature extraction on name data; treat each name as a term, calculate the TF-IDF value of each term; generate a TF-IDF feature vector for each name.

[0087] Feature fusion: The vector generated by the Word2Vec word vector model is concatenated with the term frequency-inverse document frequency TF-IDF feature vector to obtain a comprehensive feature vector containing semantic and statistical information.

[0088] Dataset division: The dataset is divided into training set and test set in a ratio of 7:3.

[0089] Training random forest model: Use comprehensive feature vectors and gender labels to train the random forest model; the random forest model integrates multiple decision trees, each of which is trained based on different random samples and comprehensive feature vectors, and the final prediction result is obtained by voting or averaging.

[0090] Model verification and tuning: Use the test set to verify the performance of the model and calculate the accuracy and recall rate indicators; adjust the configuration of the random forest model based on the calculated accuracy and recall rate indicators.

[0091] Predict missing gender: Use the trained random forest model to predict the medical bill data with missing gender and fill in the missing gender labels.

[0092] Step S5, construct a model for completing department information of medical bill data, predict corresponding department information through drug information, and complete and correct missing department information and non-standard department information in the medical bill data of medical billing units with medium quality and low quality evaluation levels.

[0093] like Figure 4 As shown in the figure, by introducing advanced natural language processing technology, for some data departments that are third-level departments, standard department similarity calculation based on n-gram and Levenshtein distance strings, and then using FastText to establish a department prediction model, department information can be effectively standardized and the accuracy and consistency of drug information and department affiliation can be improved. This not only helps to optimize the data management process within the hospital, but also improves the data quality of medical electronic receipts. Specifically, the following steps are included:

[0094] Data extraction: Extract valid data from the database, including drug lists, drug project types, and department information.

[0095] Data query: Write ClickHouse query statements to extract key field data from the database, including the drug list, the project type to which the drug belongs, and the department information.

[0096] Data cleaning: Regularize and clean the data, remove all non-Chinese special characters and comments from the original data, process missing values, outliers, invalid values, and delete duplicate data.

[0097] Department information standardization: Department names are standardized using n-gram and Levenshtein distance to handle spelling errors and variants. The standard names refer to the "Hospital Department Coding Standards Issued by the National Health Commission".

[0098] Department dictionary construction: According to the "Hospital Department Coding Standards Issued by the National Health Commission", a dictionary containing standardized secondary departments and standardized tertiary departments is established as a reference for mapping. As shown in Table 1:

[0099] Table 1 Mapping table of standardized secondary departments and standardized tertiary departments

[0100]

[0101] n-gram generation: Generate n-gram feature 2-gram for department names, which is used to calculate the string similarity between the department names in the data and the standard three-level departments, and use the ratio as the similarity score.

[0102] Distance similarity score calculation: Calculate the Levenshtein distance between the department name to be mapped and the standard department name, return the similarity ratio between the department name in the data and the standard three-level department, and thus match the most similar department name.

[0103] Mapping and replacement: The average similarity of n-gram and Levenshtein is used as the mapping basis. The closest candidate name is returned based on the average similarity of the department name in the original database, and it is mapped to the standard secondary department name to replace the original non-standard department name. As shown in Table 2:

[0104] Table 2 Standardized secondary department mapping replacement example

[0105]

[0106] Drug and project type mapping dictionary construction: Build a mapping dictionary based on drug information and project type to provide a reference for subsequent processing.

[0107] Data analysis: Analyze the relationship between project types and drug names, standardize the mapping of drugs outside the standard project types based on the existing bill content after removing non-Chinese content, and build a corresponding relationship between common project types and drugs.

[0108] Dictionary construction: Build a dictionary with the project type and the corresponding drug name to form a mapping. As shown in Table 3:

[0109] Table 3 Mapping table of project types and corresponding drug names

[0110]

[0111] Bill information data processing: Based on the secondary sorting of the importance of the item types, the list contents of each item type are sorted again to remove duplicate text data.

[0112] Item sorting: All medical electronic receipts without non-Chinese and remarks are sorted according to the importance of item types according to the list. For example, the item types in hospitalization receipts are sorted from most to least important as Western medicine fees, surgery fees, laboratory fees, examination fees, Chinese patent medicine fees, etc. At the same time, the items of each item type are sorted in the list as the final data order. The drug names, item types and processed item lists are integrated into a dictionary form, and then converted into text data. Secondary deduplication based on semantics is performed to remove data with word vector similarity greater than 90%.

[0113] Text data construction and classification: Construct text data for model training and use FastText to classify departments. The specific steps are as follows:

[0114] Data division: The data is divided into training set and test set according to the ratio of 7:3.

[0115] Data labeling: Use the standardized department names as labels to form a training data set.

[0116] Model training: Use FastText to train the constructed text data to obtain a classification model.

[0117] Model verification: The test set was used to evaluate the performance of the model. After testing, the accuracy and recall of the department data completion model after model tuning in outpatient list notes reached 88.16% and 88.05% respectively, and in inpatient list notes, the performance was 95.07% and 95.02% respectively.

[0118] The trained medical bill data department information completion model is used to complete and correct missing department information and non-standard department information in the medical bill data.

[0119] The present invention uses a medical bill data completion model to complete and correct the missing gender information, department information, and non-standard department information in the medical bill data; improve the quality of medical bill data of the billing unit and provide a data basis for the subsequent billing process.

[0120] Embodiment 2:

[0121] This embodiment provides a medical bill data intelligent quality evaluation and data completion system, including:

[0122] The evaluation model and evaluation index construction module collects medical bill data from medical billing units, builds a medical bill data quality evaluation model, and determines the evaluation index of medical bill data based on the characteristics of medical bill data;

[0123] The data quality score calculation module uses the entropy weight method to calculate the weights of the evaluation indicators, and then calculates the final evaluation score of the medical bill data quality of the medical billing unit based on the weights;

[0124] The medical billing unit classification module divides the medical billing units into different levels according to their medical bill data quality evaluation scores, namely high quality, medium quality and low quality;

[0125] The gender content completion module builds a model for completing the missing gender content in medical bill data. It uses feature fusion and random forest models to complete the missing gender content in the medical bill data of medical billing units with medium and low quality evaluation levels.

[0126] The department information completion and correction module builds a department information completion model for medical bill data, predicts the corresponding department information through drug information, and completes and corrects the missing and non-standard department information in the medical bill data of billing units with medium and low quality evaluation levels.

[0127] Embodiment three:

[0128] This embodiment provides an electronic device having a computer program stored thereon, and when the computer program is executed by a processor, the method for intelligent quality evaluation and data completion of medical bill data as described in any embodiment of the present invention is implemented.

[0129] Embodiment 4:

[0130] This embodiment provides a computer-readable medium for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the medical bill data intelligent quality evaluation and data completion method as described in any embodiment of the present invention.

[0131] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0132] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0133] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0134] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.

[0135] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. Intelligent quality evaluation and data completion method for medical bill data, characterized in that: The following steps are involved: Step S1, collecting medical bill data from medical billing units, building a medical bill data quality evaluation model, and determining evaluation indicators for the medical bill data based on the characteristics of the medical bill data; Step S2, using the entropy weight method to calculate the weights of the evaluation indicators, and then calculating the final evaluation score of the medical bill data quality of the medical billing unit based on the weights; Step S3, according to the medical bill data quality evaluation scores of the medical billing units, the medical billing units are divided into different levels, and the different levels are high quality, medium quality, and low quality; Step S4, constructing a model for completing missing gender content in medical bill data, and completing missing gender content in medical bill data of medical billing units with evaluation levels of medium quality and low quality through feature fusion and random forest model; Step S5, constructing a department information completion model for medical bill data, predicting corresponding department information through drug information, and completing and correcting missing department information and non-standard department information in the medical bill data of medical billing units with medium and low quality evaluation levels; Wherein: the construction of the model for completing the missing gender content in medical bill data, which completes the missing gender content in the medical bill data of medical billing units with evaluation levels of medium quality and low quality through feature fusion and random forest model, includes the following steps: Data collection and cleaning: Collect the original data set containing name and gender information from medical bill data to ensure the diversity and representativeness of data sources; process outliers in the data set, remove duplicate data, and standardize the name format; Word2Vec word vector: Select the Word2Vec word vector model to vectorize the name. For each name, calculate the word vectors of all its characters and take the average or weighted average to convert the name into a fixed-length vector representation. TF-IDF feature extraction: Perform TF-IDF feature extraction on name data; treat each name as a term, calculate the TF-IDF value of each term; generate a TF-IDF feature vector for each name; Feature fusion: The vector generated by the Word2Vec word vector model is combined with the term frequency-inverse document frequency TF-IDF feature vector to obtain a comprehensive feature vector containing semantic and statistical information; Dataset division: The dataset is divided into training set and test set according to 7:3; Training random forest model: Use comprehensive feature vectors and gender labels to train the random forest model; the random forest model integrates multiple decision trees, each of which is trained based on different random samples and comprehensive feature vectors, and the final prediction result is obtained by voting or averaging; Model verification and tuning: Use the test set to verify the performance of the model and calculate the accuracy and recall rate indicators; adjust the configuration of the random forest model according to the calculated accuracy and recall rate indicators; Predict missing gender: Use the trained random forest model to predict the medical bill data with missing gender and fill in the missing gender labels; Among them, the constructed medical bill data department information completion model predicts the corresponding department information through drug information, and completes and corrects the missing department information and non-standard department information in the medical bill data of medical billing units with medium and low quality evaluation levels; specifically includes the following steps: Data collection and cleaning: Extract key field data including drug list, drug item type and department information from the medical bill database; perform regularization cleaning on the data, remove all non-Chinese special characters and notes from the original data, process missing values, outliers, invalid values, and delete duplicate data; Department information standardization: Use n-gram and Levenshtein distance to standardize department names and handle spelling errors and variants; Drug and project type mapping dictionary construction: Build a mapping dictionary based on drug information and project type; Medical bill data processing: secondary sorting based on the importance of item types, secondary sorting of the list content of each item type, and deduplication of text data; Text data construction and classification: Construct a text data model to train the department information completion model for medical bill data, use FastText to classify departments, and obtain the trained department information completion model for medical bill data; The trained medical bill data department information completion model is used to complete and correct missing department information and non-standard department information in the medical bill data.

2. The medical bill data intelligent quality evaluation and data completion method according to claim 1 is characterized by: The evaluation index of the medical bill data is determined according to the characteristics of the medical bill data; the evaluation index content includes: completeness, accuracy, and effectiveness.

3. The medical bill data intelligent quality evaluation and data completion method according to claim 1 is characterized by: The entropy weight method is used to calculate the weight of each evaluation index. The specific process is as follows: Data cleaning: pre-process the raw data, including deduplication, outlier processing, missing value processing, etc., to ensure the standardization of input data; Standardization: In order to eliminate the dimensional differences between different indicators, each data quality indicator is standardized; Weight calculation: By calculating the information entropy of the completeness, accuracy and effectiveness of the evaluation indicators, the difference coefficient of each indicator is obtained, thereby determining the weight of each indicator.

4. Medical bill data intelligent quality evaluation and data completion system, characterized by: include: The evaluation model and evaluation index construction module collects medical bill data from medical billing units, builds a medical bill data quality evaluation model, and determines the evaluation index of medical bill data based on the characteristics of medical bill data; The data quality score calculation module uses the entropy weight method to calculate the weights of the evaluation indicators, and then calculates the final evaluation score of the medical bill data quality of the medical billing unit based on the weights; The medical billing unit classification module divides the medical billing units into different levels according to their medical bill data quality evaluation scores, namely high quality, medium quality and low quality; The gender content completion module builds a model for completing the missing gender content in medical bill data. It uses feature fusion and random forest models to complete the missing gender content in the medical bill data of medical billing units with medium and low quality evaluation levels. The department information completion and correction module builds a department information completion model for medical bill data, predicts the corresponding department information through drug information, and completes and corrects the missing and non-standard department information in the medical bill data of medical billing units with medium and low quality evaluation levels; Wherein: the construction of the model for completing the missing gender content in medical bill data, which completes the missing gender content in the medical bill data of medical billing units with evaluation levels of medium quality and low quality through feature fusion and random forest model, includes the following steps: Data collection and cleaning: Collect the original data set containing name and gender information from medical bill data to ensure the diversity and representativeness of data sources; process outliers in the data set, remove duplicate data, and standardize the name format; Word2Vec word vector: Select the Word2Vec word vector model to vectorize the name. For each name, calculate the word vectors of all its characters and take the average or weighted average to convert the name into a fixed-length vector representation. TF-IDF feature extraction: Perform TF-IDF feature extraction on name data; treat each name as a term, calculate the TF-IDF value of each term; generate a TF-IDF feature vector for each name; Feature fusion: The vector generated by the Word2Vec word vector model is combined with the term frequency-inverse document frequency TF-IDF feature vector to obtain a comprehensive feature vector containing semantic and statistical information; Dataset division: The dataset is divided into training set and test set according to 7:3; Training random forest model: Use comprehensive feature vectors and gender labels to train the random forest model; the random forest model integrates multiple decision trees, each of which is trained based on different random samples and comprehensive feature vectors, and the final prediction result is obtained by voting or averaging; Model verification and tuning: Use the test set to verify the performance of the model and calculate the accuracy and recall rate indicators; adjust the configuration of the random forest model according to the calculated accuracy and recall rate indicators; Predict missing gender: Use the trained random forest model to predict the medical bill data with missing gender and fill in the missing gender labels; Among them, the constructed medical bill data department information completion model predicts the corresponding department information through drug information, and completes and corrects the missing department information and non-standard department information in the medical bill data of medical billing units with medium and low quality evaluation levels; specifically includes the following steps: Data collection and cleaning: Extract key field data including drug list, drug item type and department information from the medical bill database; perform regularization cleaning on the data, remove all non-Chinese special characters and notes from the original data, process missing values, outliers, invalid values, and delete duplicate data; Department information standardization: Use n-gram and Levenshtein distance to standardize department names and handle spelling errors and variants; Drug and project type mapping dictionary construction: Build a mapping dictionary based on drug information and project type; Medical bill data processing: secondary sorting based on the importance of item types, secondary sorting of the list content of each item type, and deduplication of text data; Text data construction and classification: Construct a text data model to train the department information completion model for medical bill data, use FastText to classify departments, and obtain the trained department information completion model for medical bill data; The trained medical bill data department information completion model is used to complete and correct missing department information and non-standard department information in the medical bill data.

5. The medical bill data intelligent quality evaluation and data completion system according to claim 4 is characterized by: The evaluation model and evaluation index construction module determine the evaluation index of the medical bill data according to the characteristics of the medical bill data; the evaluation index content includes: completeness, accuracy, and effectiveness.

6. The medical bill data intelligent quality evaluation and data completion system according to claim 4 is characterized by: The data quality score calculation module uses the entropy weight method to calculate the weight of each evaluation index. The specific process is as follows: Data cleaning: pre-process the raw data, including deduplication, outlier processing, missing value processing, etc., to ensure the standardization of input data; Standardization: In order to eliminate the dimensional differences between different indicators, each data quality indicator is standardized; Weight calculation: By calculating the information entropy of the completeness, accuracy and effectiveness of the evaluation indicators, the difference coefficient of each indicator is obtained, thereby determining the weight of each indicator.

7. An electronic 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 program, the medical bill data intelligent quality evaluation and data completion method as described in any one of claims 1 to 3 is implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the medical bill data intelligent quality evaluation and data completion method as described in any one of claims 1 to 3 is implemented.

Images