CN112258135A - Method and device for auditing prescription data and computer-readable storage medium - Google Patents
Method and device for auditing prescription data and computer-readable storage medium Download PDFInfo
- Publication number
- CN112258135A CN112258135A CN202010413774.2A CN202010413774A CN112258135A CN 112258135 A CN112258135 A CN 112258135A CN 202010413774 A CN202010413774 A CN 202010413774A CN 112258135 A CN112258135 A CN 112258135A
- Authority
- CN
- China
- Prior art keywords
- data
- prescription
- classification result
- model
- prescription data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
- G06Q10/103—Workflow collaboration or project management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Entrepreneurship & Innovation (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Databases & Information Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Economics (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Biomedical Technology (AREA)
- Epidemiology (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses a method and a device for auditing prescription data and a computer-readable storage medium. The auditing method of prescription data comprises the following steps: determining a first classification result for the prescription data based on structured data extracted from the prescription data, wherein the prescription data further comprises unstructured data; inputting unstructured data generated according to prescription data into one or more deep learning models to obtain a second classification result of the prescription data output by the deep learning models; and determining an auditing result of the prescription data according to the first classification result and the second classification result. The embodiment of the invention ensures that the auditing result is more accurate and improves the safety of prescription use.
Description
Technical Field
The present invention relates to the field of data processing, and in particular, to a method and an apparatus for auditing prescription data, and a computer-readable storage medium.
Background
The issuance of the medical institution prescription audit specification makes prescription audit rigid requirements. A good intelligent prescription-reviewing system should be able to solve the problems of scarce resources for pharmacists, a large number of medicines, complex contraindications for medicines, etc.
In the related art, the method of the intelligent auditor includes the following two.
The first is a method for performing an intelligent auditor based on a Hospital Information System (HIS). The HIS refers to an information system that uses the modernization means such as computer software and hardware technology and network communication technology to comprehensively manage the people flow, logistics and financial flow of hospitals and all the departments to which the hospitals belong, and collects, stores, processes, extracts, transmits and summarizes data generated at all the stages of medical activities to process the data to form various information, thereby providing comprehensive automatic management and various services for the overall operation of the hospitals. In the process of party checking, the HIS system firstly collects the electronic medical record of the patient, then inquires the medication guide and the medicine specification stored in the HIS system, and finally judges whether the medication is reasonable or not in a regular mode according to the retrieved information.
The second method is a method for carrying out party examination based on a Bayesian model. In the mode, a probability matrix of the model is obtained by learning historical prescription data and auditing data, and finally, an intelligent auditing model is established by the probability matrix of the medicines and the preliminary diagnosis. The content analyzed by the model mainly comprises the correlation analysis between the preliminary diagnosis and the medicine, the correlation between the user symptoms and the corresponding medicine components and the like.
Disclosure of Invention
After further analysis of the related technology, the inventor finds that the HIS system is mainly applied to large hospitals and is expensive. In addition, the HIS system only queries the drug data in an auditor link without any reasoning, so that the auditor result is not accurate. When the drug data in the HIS system is insufficient, its performance may be greatly affected. And the generalization capability of the Bayesian model-based model is limited, and the training of the Bayesian model-based model depends on structured data with limited information. Therefore, the result of the prescription is not accurate enough.
The embodiment of the invention aims to solve the technical problem that: how to improve the accuracy of intelligent auditing of prescriptions.
According to a first aspect of some embodiments of the present invention, there is provided a method for auditing prescription data, comprising: determining a first classification result for the prescription data based on structured data extracted from the prescription data, wherein the prescription data further comprises unstructured data; inputting unstructured data generated according to prescription data into one or more deep learning models to obtain a second classification result of the prescription data output by the deep learning models; and determining an auditing result of the prescription data according to the first classification result and the second classification result.
In some embodiments, the structured data includes at least one of numeric type data, enumerated type data, and prescription data.
In some embodiments, the structured data includes one or more fields and a value for each field; unstructured data generated from prescription data includes: the unstructured data corresponding to the structured data are obtained by filling a preset text corresponding to the value of a field in the structured data into a position to be filled in of a corresponding field in a preset text template; and, unstructured data in the prescription data.
In some embodiments, the review results of the prescription data include a category indicating that the prescription is correct and one or more categories indicating the type of error of the prescription.
In some embodiments, the prescription data includes patient data and medication data; in the first classification result, the error types of the prescription comprise a type that the medication data in the structured data is not matched with the patient data in the structured data and a type that the medication data in the structured data exceeds a limit amount; in the second classification result, the error type of the prescription includes a type that medication data in the unstructured data does not match patient data in the prescription data.
In some embodiments, determining the result of the audit on the prescription data based on the first classification result and the second classification result comprises: under the condition that the first classification result and the second classification result both indicate that the prescription is correct, the checking result of the prescription data is that the prescription is correct; in a case where the first classification result and the second classification result each include a category representing an error type of the prescription, the audit result of the prescription data is a set of categories representing the error type of the prescription in the first classification result and the second classification result.
In some embodiments, determining a first classification result for the prescription data based on the structured data extracted from the prescription data comprises: the method comprises the steps of inputting structured data into one or more machine learning-based integrated models, and obtaining a first classification result output by the integrated models, wherein the structured data comprises one or more fields and a value corresponding to each field, the integrated models are decision tree-based integrated models, each non-leaf node of the decision tree represents one field in the structured data and division values of the fields on the nodes, and each leaf node represents one classification result.
In some embodiments, the integrated model comprises one or more of a random forest model, or a gradient boosting iterative decision tree, GBDT, model.
In some embodiments, the deep learning model is a deep learning based text classification model; inputting unstructured data generated from prescription data into one or more deep learning models comprises: splicing unstructured data extracted from prescription data and unstructured data obtained by converting the structured data into texts; performing word segmentation processing on the text; converting each word in the text after word segmentation into a word vector to obtain a word vector sequence corresponding to the text; the word vector sequence is input into a deep learning model.
In some embodiments, the deep learning model includes one or more of a text classification model TextCNN using a convolutional neural network, a text classification model TextRNN using a recurrent neural network, a Transformer-based coding model transform.
In some embodiments, the auditing method further comprises: training one of the integrated model or the deep learning model for multiple times by using training data and sequentially using different alternative model parameters; testing the prediction effect of the trained model by using the test data; and determining the trained model corresponding to the candidate model parameter with the best prediction effect as the model for checking the prescription data.
According to a second aspect of some embodiments of the present invention, there is provided an auditing apparatus for prescription data, including: a structured data classification module configured to determine a first classification result for the prescription data based on structured data extracted from the prescription data, wherein the prescription data further includes unstructured data; the unstructured data classification module is configured to input unstructured data generated according to prescription data into one or more deep learning models to obtain a second classification result output by the deep learning models; and the auditing result determining module is configured to determine the auditing result of the prescription data according to the first classification result and the second classification result.
According to a third aspect of some embodiments of the present invention, there is provided an auditing apparatus for prescription data, including: a memory; and a processor coupled to the memory, the processor configured to perform an auditing method of any of the foregoing recipe data based on instructions stored in the memory.
According to a fourth aspect of some embodiments of the present invention, there is provided a computer readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements any one of the above-mentioned auditing methods for prescription data.
Some embodiments of the above invention have the following advantages or benefits: according to the embodiment of the invention, the characteristics of the structured data and the unstructured data in the prescription data can be considered simultaneously when the prescription data is audited, and the audit is carried out according to the classification result of the structured data in the prescription data and the classification result of the unstructured data corresponding to the prescription data as a whole, so that the audit result is more accurate, and the safety of prescription use is improved.
Other features of the present invention and advantages thereof will become apparent from the following detailed description of exemplary embodiments thereof, which proceeds with reference to the accompanying drawings.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 illustrates a flow diagram of a prescription data review method according to some embodiments of the invention.
FIG. 2 illustratively shows a data flow diagram during a prescription data review process.
FIG. 3 illustrates a block diagram of an audit device of prescription data according to some embodiments of the invention.
FIG. 4 is a block diagram of an audit device of prescription data according to further embodiments of the present invention.
FIG. 5 illustrates a schematic diagram of an audit device for processing data according to further embodiments of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The relative arrangement of the components and steps, the numerical expressions and numerical values set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.
Meanwhile, it should be understood that the sizes of the respective portions shown in the drawings are not drawn in an actual proportional relationship for the convenience of description.
Techniques, methods, and apparatus known to those of ordinary skill in the relevant art may not be discussed in detail but are intended to be part of the specification where appropriate.
In all examples shown and discussed herein, any particular value should be construed as merely illustrative, and not limiting. Thus, other examples of the exemplary embodiments may have different values.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, further discussion thereof is not required in subsequent figures.
After further analysis, the inventors have discovered that prescription data is particular in that it includes both structured data and unstructured data. Structured data includes, for example, numerical or enumerated data for patient gender, patient age, diagnosed illness, prescribed medication, whether pregnant, etc.; unstructured data includes, for example, patient notes, physician-described usage amounts, and the like. The invention provides a prescription auditing method combining structured data and unstructured data aiming at the characteristics of prescription data.
FIG. 1 illustrates a flow diagram of a prescription data review method according to some embodiments of the invention. As shown in fig. 1, the prescription data auditing method of this embodiment includes steps S102 to S106.
In step S102, a first classification result for the prescription data is determined based on the structured data extracted from the prescription data, wherein the prescription data further includes unstructured data.
The prescription data may be on-line data. For example, after a doctor or pharmacist submits electronic prescription data through a computer, the prescription data can be acquired and reviewed using the method of the embodiment of the present invention. If the audit result is that the prescription data is correct, continuing to send the data to the next subsystem or outputting the data; and if the prescription has errors as a result of the verification, feeding back error information to a computer which submits prescription data. Further, the prescription data may also be data stored in a database.
The data variable type of the structured data is a numerical variable or a categorical variable. Each category corresponds to a unique numerical value. In some embodiments, the structured data includes at least one of numeric type data, enumerated type data, and prescription data. That is, the structured data is of a numeric type, or of a type other than numeric, but whose contents are one of a plurality of options that can be expected. For example, the patient's age, the number of times of medical visits, the length of illness, etc. are numerical type data, and the patient's sex, the diagnosed disease, the prescribed medicine, whether pregnant woman, etc. are enumerated type data. The structured data may include a plurality of fields and a value for each field.
Unstructured data is data that is irregularly or incompletely structured and has no predefined data model. For example, the unstructured data may be text-type data without predefined content. In some embodiments, the unstructured data includes, for example, patient notes, physician notes, and the like. The content of these data is unpredictable and of variable length. For example, the patient notes "frequent sneezing during season change", the doctor notes "one time, three times a day, taken before meals, and deactivated if XX is allergic", and these items change with the habit of the patient or the doctor or the details, and thus cannot be preset.
In some embodiments, structured data is input into one or more machine learning-based integration models, obtaining a first classification result for an integration model output. The integration model includes, for example, one or more of a random forest model, or a gradient boosting iterative decision tree GBDT model.
The integrated model is a model formed by integrating a plurality of weak learners so as to obtain a better prediction effect.
In step S104, the unstructured data generated from the prescription data is input into one or more deep learning models, and a second classification result of the prescription data output by the deep learning models is obtained.
For example, the structured data in the prescription data can also be converted into unstructured data, and the unstructured data and the original structured data in the prescription data can be spliced together and input into the deep learning model. In some embodiments, the unstructured data is text data and the deep learning model is a deep learning based text classification model.
In some embodiments, a preset text corresponding to the value of the field in the structured data is filled in the position to be filled in of the corresponding field in a preset text template, so as to obtain the unstructured data generated after the structured data is converted. For example, the structured data included in the prescription data is { [ age:35], [ medicine _ id:123], [ medicine _ num:3], [ symptom:45] }, and the unstructured data is "the patient has a history of allergic rhinitis. The medicine is administered three times a day. It is administered twice per time after meal. ". The preset text template is, for example, "patient age < age >, symptom < symption >, and < medicine _ id > medicine < medicine _ num >, wherein" < > "represents the position to be filled in, and the content thereof represents the field corresponding to the position. The text converted from the above structured data is "patient age 35, symptom is cold, and XXX medication 3", where cold is text corresponding to 45 and XXX is text corresponding to 123. The unstructured data generated from the prescription data were "patient age 35 years, symptoms were cold, and 3 lots of XXX medication was prescribed. The patient had a history of allergic rhinitis. The medicine is administered three times a day. Two capsules at a time, taken after meals ".
Because of the unstructured data, some content needs to be determined depending on the information in the structured data. Therefore, by converting all the contents in the prescription data into unstructured data, a more accurate determination can be made as to whether there is an error in the part of the unstructured data in the prescription data.
In step S106, an audit result of the prescription data is determined according to the first classification result and the second classification result.
In some embodiments, the review results of the prescription data include a category indicating that the prescription is correct and one or more categories indicating the type of error of the prescription. The first and second classification results also include a category indicating that the prescription is correct and one or more categories indicating the type of error of the prescription. The types of errors include, for example, overdose, medication intolerance, presence of patient medication disabled, and the like. Therefore, the auditing result can not only give the result of whether the prescription is correct, but also give a specific error type, so that a doctor or a pharmacist can find problems in time, and the usability of prescription data processing is improved.
The structured data and the unstructured data have different characteristics and contain different information, so that the types of prescription errors which can be identified are different. In some embodiments, the prescription data includes patient data and medication data. In the first classification result, the error types of the prescription comprise a type that the medication data in the structured data is not matched with the patient data in the structured data and a type that the medication data in the structured data exceeds a limit amount; in the second classification result, the error type of the prescription includes a type that medication data in the unstructured data does not match patient data in the prescription data. For example, usage volumes are often present in physician notes and presented in text of uncertain content and length, which can only be determined by unstructured data. Therefore, the processing of the structured data and the processing of the unstructured data can respectively obtain corresponding classification results, so that the accuracy of prescription auditing is higher.
In some embodiments, in the case that both the first classification result and the second classification result indicate that the prescription is correct, the audit result of the prescription data is that the prescription is correct; in a case where the first classification result and the second classification result each include a category representing an error type of the prescription, the audit result of the prescription data is a set of categories representing the error type of the prescription in the first classification result and the second classification result. Since the prescription is related to the medication safety of the patient and is preferably carefully handled, the reliability of prescription verification can be further improved by feeding back all error types included in the first classification result and the second classification result, so that the medication safety of the patient is improved.
By the method, the characteristics of the structured data and the unstructured data in the prescription data can be considered when the prescription data is audited, and the audit is performed according to the classification result of the structured data in the prescription data and the classification result of the unstructured data corresponding to the prescription data as a whole, so that the audit result is more accurate, and the safety of prescription use is improved.
In classifying structured data, for example, a decision tree based integration model can be used. In some embodiments, the integrated model includes one or more of a random forest model, or a Gradient Boosting iterative Decision Tree (GBDT) model.
And randomly selecting samples and features for dividing the tree by the random forest model when the decision tree is constructed, and determining the classification result of the random forest model according to the simple voting result of the decision tree. The training speed and the prediction accuracy of the random forest model are high.
The GBDT can improve the performance of the model by reducing the deviation of the model according to the weight of the sample in the next iteration according to the predicted error rate in the iterative training process.
In some embodiments, structured data is input into one or more machine learning-based integration models to obtain a first classification result output by the integration model. The structured data comprises one or more fields and corresponding values of each field, the integrated model is a decision tree-based integrated model, each non-leaf node of the decision tree represents one field in the structured data and division values of the fields on the nodes, and each leaf node represents a classification result. For example, the structured data includes the age of the patient, the number of times of visit, the duration of illness, the number of medicines, and the name of the medicine, and a certain non-leaf node on a certain decision tree in the integrated model is used to determine whether the number of medicines is greater than 3, if so, the left sub-tree of the non-leaf node is entered, otherwise, the right sub-tree is entered.
In classifying unstructured data, the deep learning model may be one or more of TextCNN (text classification model using convolutional neural network), TextRNN (text classification model using recurrent neural network), Transformer (converter-based coding model).
The TextRNN starts from the first word in the text and calculates for each word in the text in turn, and each calculation contains the information from the current word to the first word.
TextCNN presents text in the form of a matrix, for example, each row of the matrix corresponds to one sentence in the text, so that key information in the text can be extracted in windows of different sizes.
The Transformer is an encoder in a model BERT based on an attention mechanism, and has the characteristics of high calculation speed and high prediction accuracy.
The unstructured data may be preprocessed to generate input data for the model. In some embodiments, inputting unstructured data generated from prescription data into one or more deep learning models comprises: splicing unstructured data extracted from prescription data and unstructured data obtained by converting the structured data into texts; performing word segmentation processing on the text; converting each Word in the text after Word segmentation into a Word vector to obtain a Word vector sequence corresponding to the text, for example, using a Word vector conversion tool Word2Vec and the like to obtain the Word vector; the word vector sequence is input into a deep learning model.
In some embodiments, the training data may be used to train each model separately.
In some embodiments, the training data may be utilized to train one of the integrated model or the deep learning model multiple times using different candidate model parameters in turn; testing the prediction effect of the trained model by using the test data; and determining the trained model corresponding to the candidate model parameter with the best prediction effect as the model for checking the prescription data. Therefore, the obtained model has better prediction effect.
FIG. 2 illustratively shows a data flow diagram during a prescription data review process. As shown in fig. 2, structured data and unstructured data are extracted from the square data on the line, respectively. Inputting the structured data into a GBDT model and a random forest model; the unstructured data extracted from the prescription data and the unstructured data generated by converting the structured data are spliced to obtain unstructured data, and the spliced unstructured data are input into TextRNN, TextCNN and transform. And finally, determining the auditing result of the prescription data according to the forecasting results of the five models.
FIG. 3 illustrates a block diagram of an audit device of prescription data according to some embodiments of the invention. As shown in fig. 3, the auditing apparatus 30 for prescription data of this embodiment includes: a structured data classification module 310 configured to determine a first classification result for the prescription data based on structured data extracted from the prescription data, wherein the prescription data further includes unstructured data; an unstructured data classification module 320 configured to input unstructured data generated from the prescription data into one or more deep learning models, and obtain a second classification result of the prescription data output by the deep learning models; an audit result determination module 330 configured to determine an audit result for the prescription data according to the first classification result and the second classification result.
In some embodiments, the structured data includes at least one of numeric type data, enumerated type data, and prescription data.
In some embodiments, the structured data includes one or more fields and a value for each field; unstructured data generated from prescription data includes: the unstructured data corresponding to the structured data are obtained by filling a preset text corresponding to the value of a field in the structured data into a position to be filled in of a corresponding field in a preset text template; and, unstructured data in the prescription data.
In some embodiments, the review results of the prescription data include a category indicating that the prescription is correct and one or more categories indicating the type of error of the prescription.
In some embodiments, the prescription data includes patient data and medication data; in the first classification result, the error types of the prescription comprise a type that the medication data in the structured data is not matched with the patient data in the structured data and a type that the medication data in the structured data exceeds a limit amount; in the second classification result, the error type of the prescription includes a type that medication data in the unstructured data does not match patient data in the prescription data.
In some embodiments, the audit result determination module 330 is further configured to determine that the prescription data is correct if the first classification result and the second classification result both indicate that the prescription is correct; in a case where the first classification result and the second classification result each include a category representing an error type of the prescription, the audit result of the prescription data is a set of categories representing the error type of the prescription in the first classification result and the second classification result.
In some embodiments, the structured data classification module 310 is further configured to input the structured data into one or more machine learning-based integration models, and obtain a first classification result output by the integration models, wherein the structured data includes one or more fields and a value corresponding to each field, the integration models are decision tree-based integration models, each non-leaf node of the decision tree represents one field in the structured data and a partition value of the field on the node, and each leaf node represents one classification result.
In some embodiments, the integrated model comprises one or more of a random forest model, or a gradient boosting iterative decision tree, GBDT, model.
In some embodiments, the deep learning model is a deep learning based text classification model; the unstructured-data classification module 320 is further configured to concatenate unstructured data extracted from the prescription data and unstructured data obtained by converting the structured data into text; performing word segmentation processing on the text; converting each word in the text after word segmentation into a word vector to obtain a word vector sequence corresponding to the text; the word vector sequence is input into a deep learning model.
In some embodiments, the deep learning model includes one or more of a text classification model TextCNN using a convolutional neural network, a text classification model TextRNN using a recurrent neural network, a Transformer-based coding model transform.
In some embodiments, the auditing apparatus 30 also includes: a training module 340 configured to train one of the integrated model or the deep learning model multiple times using different candidate model parameters in sequence using training data; testing the prediction effect of the trained model by using the test data; and determining the trained model corresponding to the candidate model parameter with the best prediction effect as the model for checking the prescription data.
FIG. 4 is a block diagram of an audit device of prescription data according to further embodiments of the present invention. As shown in fig. 4, the auditing apparatus 40 for prescription data of this embodiment includes: a memory 410 and a processor 420 coupled to the memory 410, the processor 420 being configured to perform an audit method of the prescription data in any of the foregoing embodiments based on instructions stored in the memory 410.
FIG. 5 illustrates a schematic diagram of an audit device for processing data according to further embodiments of the present invention. As shown in fig. 5, the auditing apparatus 50 for prescription data of this embodiment includes: the memory 510 and the processor 520 may further include an input/output interface 530, a network interface 540, a storage interface 550, and the like. These interfaces 530, 540, 550 and the connections between the memory 510 and the processor 520 may be, for example, via a bus 560. The input/output interface 530 provides a connection interface for input/output devices such as a display, a mouse, a keyboard, and a touch screen. The network interface 540 provides a connection interface for various networking devices. The storage interface 550 provides a connection interface for external storage devices such as an SD card and a usb disk.
An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to implement any one of the auditing methods for prescription data.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable non-transitory storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (14)
1. An auditing method for prescription data, comprising:
determining a first classification result for prescription data based on structured data extracted from the prescription data, wherein the prescription data further comprises unstructured data;
inputting unstructured data generated according to the prescription data into one or more deep learning models to obtain a second classification result of the prescription data output by the deep learning models;
and determining an auditing result of the prescription data according to the first classification result and the second classification result.
2. An auditing method according to claim 1 in which the structured data includes at least one of numeric type data, enumeration type data in the prescription data.
3. An auditing method according to claim 1 in which the structured data includes one or more fields and a value for each field;
the unstructured data generated from the prescription data comprises: the unstructured data corresponding to the structured data are obtained by filling a preset text corresponding to the value of a field in the structured data into a position to be filled in of a corresponding field in a preset text template; and unstructured data in the prescription data.
4. An auditing method according to claim 1 where the results of an audit of the prescription data includes a category indicating that the prescription is correct and one or more categories indicating the type of error of the prescription.
5. An auditing method according to claim 4 in which the prescription data includes patient data and medication data;
in the first classification result, the error types of the prescription comprise a type that medication data in the structured data does not match patient data in the structured data, and a type that the medication data in the structured data exceeds a limit amount;
in the second classification result, the error type of the prescription includes a type that medication data in unstructured data does not match patient data in the prescription data.
6. An auditing method according to claim 4 where the determining an outcome of an audit of the prescription data from the first and second classification results comprises:
if the first classification result and the second classification result both indicate that the prescription is correct, the checking result of the prescription data is that the prescription is correct;
in a case where the first classification result and the second classification result both include a category representing an error type of the prescription, the review result of the prescription data is a set of categories representing the error type of the prescription in the first classification result and the second classification result.
7. An auditing method according to any one of claims 1 to 6 where said determining a first classification result for a recipe data based on structured data extracted from the recipe data comprises:
inputting the structured data into one or more machine learning-based integrated models, and obtaining a first classification result output by the integrated models, wherein the structured data comprises one or more fields and a value corresponding to each field, the integrated models are decision tree-based integrated models, each non-leaf node of the decision tree represents one field in the structured data and a division value of the field on the node, and each leaf node represents one classification result.
8. An auditing method according to claim 7 in which the ensemble model comprises one or more of a random forest model, or a gradient boosting iterative decision tree, GBDT, model.
9. An auditing method according to any one of claims 1 to 6 where the deep learning model is a deep learning based text classification model;
the inputting unstructured data generated from the prescription data into one or more deep learning models comprises:
splicing unstructured data extracted from the prescription data and unstructured data obtained by converting the structured data into text;
performing word segmentation processing on the text;
converting each word in the text after word segmentation into a word vector to obtain a word vector sequence corresponding to the text;
inputting the sequence of word vectors into a deep learning model.
10. An auditing method according to claim 9 in which the deep learning model includes one or more of a text classification model TextCNN using a convolutional neural network, a text classification model TextRNN using a recurrent neural network, a Transformer-based coding model Transformer.
11. An auditing method according to claim 1 further comprising:
training the integrated model or one of the deep learning models for multiple times by using training data and sequentially using different alternative model parameters;
testing the prediction effect of the trained model by using the test data;
and determining the trained model corresponding to the candidate model parameter with the best prediction effect as the model for checking the prescription data.
12. An audit device of prescription data, comprising:
a structured data classification module configured to determine a first classification result for prescription data based on structured data extracted from the prescription data, wherein the prescription data further comprises unstructured data;
the unstructured data classification module is configured to input unstructured data generated according to the prescription data into one or more deep learning models to obtain a second classification result output by the deep learning models;
an audit result determination module configured to determine an audit result for the prescription data according to the first classification result and the second classification result.
13. An audit device of prescription data, comprising:
a memory; and
a processor coupled to the memory, the processor configured to perform an auditing method of the prescription data of any of claims 1-11 based on instructions stored in the memory.
14. A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, implements an auditing method for prescription data according to any of claims 1 to 11.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010413774.2A CN112258135A (en) | 2020-05-15 | 2020-05-15 | Method and device for auditing prescription data and computer-readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010413774.2A CN112258135A (en) | 2020-05-15 | 2020-05-15 | Method and device for auditing prescription data and computer-readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112258135A true CN112258135A (en) | 2021-01-22 |
Family
ID=74224274
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010413774.2A Pending CN112258135A (en) | 2020-05-15 | 2020-05-15 | Method and device for auditing prescription data and computer-readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112258135A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113539414A (en) * | 2021-07-30 | 2021-10-22 | 中电药明数据科技(成都)有限公司 | Method and system for predicting rationality of antibiotic medication |
| CN113836128A (en) * | 2021-09-24 | 2021-12-24 | 北京拾味岛信息科技有限公司 | Abnormal data identification method, system, equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106599933A (en) * | 2016-12-26 | 2017-04-26 | 哈尔滨工业大学 | Text emotion classification method based on the joint deep learning model |
| CN109493977A (en) * | 2018-11-09 | 2019-03-19 | 天津新开心生活科技有限公司 | Text data processing method, device, electronic equipment and computer-readable medium |
| CN110264342A (en) * | 2019-06-19 | 2019-09-20 | 深圳前海微众银行股份有限公司 | A kind of business audit method and device based on machine learning |
| CN111091350A (en) * | 2019-12-12 | 2020-05-01 | 中国银行股份有限公司 | Method, device and equipment for auditing and processing service data and storage medium |
-
2020
- 2020-05-15 CN CN202010413774.2A patent/CN112258135A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106599933A (en) * | 2016-12-26 | 2017-04-26 | 哈尔滨工业大学 | Text emotion classification method based on the joint deep learning model |
| CN109493977A (en) * | 2018-11-09 | 2019-03-19 | 天津新开心生活科技有限公司 | Text data processing method, device, electronic equipment and computer-readable medium |
| CN110264342A (en) * | 2019-06-19 | 2019-09-20 | 深圳前海微众银行股份有限公司 | A kind of business audit method and device based on machine learning |
| CN111091350A (en) * | 2019-12-12 | 2020-05-01 | 中国银行股份有限公司 | Method, device and equipment for auditing and processing service data and storage medium |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113539414A (en) * | 2021-07-30 | 2021-10-22 | 中电药明数据科技(成都)有限公司 | Method and system for predicting rationality of antibiotic medication |
| CN113836128A (en) * | 2021-09-24 | 2021-12-24 | 北京拾味岛信息科技有限公司 | Abnormal data identification method, system, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111986770B (en) | Prescription medication auditing method, device, equipment and storage medium | |
| US10818397B2 (en) | Clinical content analytics engine | |
| AU2011247830B2 (en) | Method and system for generating text | |
| CN114026651A (en) | Automatic generation of structured patient data records | |
| US8600772B2 (en) | Systems and methods for interfacing with healthcare organization coding system | |
| US20210157979A1 (en) | Systems and Methods for Extracting Form Information Using Enhanced Natural Language Processing | |
| Hsu et al. | Automated extraction of reported statistical analyses: towards a logical representation of clinical trial literature | |
| CN111066033A (en) | Machine learning method for generating labels of fuzzy results | |
| US20230178199A1 (en) | Method and system of using hierarchical vectorisation for representation of healthcare data | |
| US8676800B2 (en) | Method and system for generating text | |
| EP3329403A1 (en) | Reliability measurement in data analysis of altered data sets | |
| WO2020180679A1 (en) | Interactive treatment pathway interface for guiding diagnosis or treatment of a medical condition | |
| CN113674858A (en) | Online medical prescription medication intelligent checking method, device, equipment and storage medium | |
| CN112258135A (en) | Method and device for auditing prescription data and computer-readable storage medium | |
| US20140019090A1 (en) | Systems, methods, and logic for generating statistical research information | |
| CN113724830B (en) | Medication risk detection method based on artificial intelligence and related equipment | |
| US11901052B2 (en) | System and method for handling exceptions during healthcare record processing | |
| Liang et al. | Knowledge representation in patient safety reporting: an ontological approach | |
| CN115938608A (en) | Clinical decision early warning method and system based on prompt learning model | |
| CN109522331A (en) | Compartmentalization various dimensions health data processing method and medium centered on individual | |
| CN111986815B (en) | Project combination mining method based on co-occurrence relation and related equipment | |
| Wunnava et al. | One Size Does Not Fit All: An Ensemble Approach Towards Information Extraction from Adverse Drug Event Narratives. | |
| CN113688854A (en) | Data processing method and device and computing equipment | |
| US20140278519A1 (en) | Expression of Clinical Logic with Positive and Negative Explainability | |
| JP5764914B2 (en) | Program, analysis apparatus and method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20210308 Address after: 100176 room 701, 7 / F, building 1, yard 18, Kechuang 11th Street, Beijing Economic and Technological Development Zone, Daxing District, Beijing Applicant after: Beijing Jingdong tuoxian Technology Co.,Ltd. Address before: Room A402, 4th floor, building 2, No.18, Kechuang 11th Street, Daxing District, Beijing, 100176 Applicant before: BEIJING WODONG TIANJUN INFORMATION TECHNOLOGY Co.,Ltd. Applicant before: BEIJING JINGDONG CENTURY TRADING Co.,Ltd. |