CN114420240A

CN114420240A - Prescription management system based on artificial intelligence, big data and block chain

Info

Publication number: CN114420240A
Application number: CN202210037685.1A
Authority: CN
Inventors: 袁方; 马晓飞; 沙卫国; 杨国熊; 杨冀宁
Original assignee: Individual
Current assignee: Individual
Priority date: 2022-01-13
Filing date: 2022-01-13
Publication date: 2022-04-29

Abstract

The present disclosure describes a prescription management system based on artificial intelligence, big data and blockchains, comprising a physician prescription end that utilizes a drug knowledge base to assist prescription and generate an electronic prescription, which is then stored to blockchains; the drug knowledge base is used for storing a plurality of drug information, wherein the instruction book information in the drug information is obtained by identifying interested fields in document contents in paper drug instruction books through artificial intelligence and natural language processing technologies and big data processing; automatically auditing the electronic prescription based on prescription audit rules to obtain first audit information and storing the first audit information to a block chain, wherein the prescription audit rules are established based on a drug knowledge base and are updated along with the drug knowledge base; the system comprises a positioning information acquisition terminal, a dispensing terminal and a user terminal, wherein the positioning information acquisition terminal is used for acquiring inventory information and price information of the dispensing terminal; and the dispensing terminal is used for dispensing corresponding medicines to the user according to the dispensing order and storing the dispensing information to the block chain.

Description

Prescription management system based on artificial intelligence, big data and block chain

Technical Field

The present disclosure relates generally to the field of electronic prescription circulation, and more particularly to a prescription management system based on artificial intelligence, big data, and blockchains.

Background

The rhythm of the national medical improvement policy is accelerated continuously, the hospital releases the out-hospital prescriptions gradually under the guidance of the 'zero difference rate' policy of the medicine, the foundation is laid for the separate landing of the medicine, the prescription out-flow policy is released continuously, a new mode that the patient independently purchases the medicine from the prescription to the medical institution or the retail pharmacy is advocated, and the medical institution is ensured to make a prescription and allocate the prescription according to the prescription management method, so that the prescription informed right and the medicine purchasing option right of the patient are guaranteed.

At present, a doctor generally makes a prescription first, and then a prescription checking pharmacist checks the prescription, and then a patient takes a medicine according to the prescription. However, the lack of big data supports that patient medication safety is not fully guaranteed. In addition, the number of prescription drug examiners is insufficient, the levels are uneven, and the prescription pre-examination cannot be done at all (namely, the patients take the medicines after examination), so that the patients are easy to see a doctor for the second time, and medical resources are wasted; and the medicine taking mode is complicated and inconvenient.

Disclosure of Invention

The present disclosure has been made in view of the above circumstances, and it is an object of the present disclosure to provide a prescription management system based on artificial intelligence, big data, and a blockchain, which can effectively perform automatic audit of electronic prescriptions to reduce labor costs and improve medication safety, convenience, and economy.

Therefore, the prescription management system based on artificial intelligence, big data and a block chain comprises the block chain, a doctor prescription end, a prescription auditing platform, a user end and a medicine taking end, wherein the prescription auditing platform comprises a medicine knowledge base and an auditing engine; the doctor prescription end utilizes the medicine knowledge base to carry out auxiliary prescription and generate an electronic prescription, and then the electronic prescription is stored in the block chain, wherein the electronic prescription has a unique prescription number; the medicine knowledge base is used for storing a plurality of kinds of medicine information, wherein the instruction information in the medicine information is obtained by extracting document contents in a paper medicine instruction by using an optical character recognition technology, recognizing unstructured data in the document contents by using artificial intelligence and a natural language processing technology, and performing big data processing on the unstructured data to obtain a structured field of interest; the auditing engine comprises a first auditing module, the first auditing module acquires the electronic prescription from the block chain according to the prescription number, automatically audits the electronic prescription based on prescription auditing rules to acquire first auditing information comprising a first auditing result and a first auditing proposal, stores the first auditing information into the block chain, desensitizes the electronic prescription and transfers the electronic prescription to the medicine taking end and the user end under the condition that the electronic prescription is qualified based on the first auditing result, wherein the prescription auditing rules are established on the basis of the medicine knowledge base and are updated along with the medicine knowledge base; the user side is used for acquiring the electronic prescription, selecting a medicine taking end according to self positioning information and inventory information and price information of the medicine taking end, creating a medicine dispensing order based on the electronic prescription and storing the medicine dispensing order to the block chain; and the medicine taking end is used for obtaining the electronic prescription and the dispensing order from the block chain according to the prescription number, and under the condition that the electronic prescription is verified to be effective, corresponding medicines are dispensed to a user according to the dispensing order and delivery information is stored in the block chain. Under the condition, the medicine knowledge base is updated intelligently so as to ensure the medicine use safety of the user through big data, the required quantity of a prescription checking pharmacist can be reduced under the condition of pre-checking of the prescription through automatic checking, the medicine taking fully considers the medicine use right and the selection right of the user, and the medicine taking is convenient and quick. Therefore, the electronic prescription can be automatically checked effectively, so that the labor cost is reduced, and the medication safety, convenience and economy are improved. In addition, the traceability of the electronic prescription circulation can be realized based on the block chain, and the safety of the electronic prescription circulation can be improved.

In addition, in the prescription management system of the present disclosure, optionally, the auditing engine further includes a second review module, the second review module is used for receiving the operation of a reviewer pharmacist to review the prescription to be reviewed including the first prescription to be reviewed and/or the updated rejected prescription based on the first review information so as to generate second review information including a second review result and a second review proposal and storing the second review information to the block chain, and under the condition that the prescription to be reviewed is qualified based on the second review result, after desensitization treatment is carried out on the prescription to be reviewed, the prescription to be reviewed is transferred to the medicine taking end and the user end, the first prescription to be reviewed is the electronic prescription whose first review result is intercepted, and the refund prescription is the electronic prescription whose first review result is non-transferable. In this case, the manual review of the screened electronic prescription can be simultaneously supported. This can improve the flexibility of electronic prescription verification.

In addition, in the prescription management system related to the present disclosure, optionally, the prescription management system further includes a supervision module, and the supervision module is configured to query the electronic prescription and the circulation information of the electronic prescription from the block chain for review by a supervision department. In this case, an effective management means can be provided for the monitoring department. Therefore, the safety of the medicine taking of the user can be further guaranteed.

In addition, in the prescription management system according to the present disclosure, optionally, the unstructured data in the document content is identified by artificial intelligence and natural language processing technology as follows: inputting target texts corresponding to the document contents into an artificial intelligence-based target model to identify unstructured data in the document contents, wherein the training process of the target model comprises the following steps: after filtering invalid characters in document content to be trained, segmenting a long text in the document content to be trained to obtain a target text corresponding to the document content to be trained, and labeling the target text to construct a domain prior dictionary as a training sample, wherein if the text length of the target text is smaller than a first preset text length, the target text is merged to enable the merged text length to be not smaller than the first preset text length and larger than a second preset text length; and training an artificial intelligence based model with the training samples to obtain the target model. As such, unstructured data can be identified based on artificial intelligence and natural language processing techniques.

In addition, in the prescription management system according to the present disclosure, optionally, when the artificial intelligence based model is trained, a basic model is trained by using the domain prior dictionary, pseudo-labeled data is obtained by predicting unlabeled data by using the basic model, and the pseudo-labeled data is added to the training sample to jointly train the artificial intelligence based model to obtain the target model. Therefore, the problem of scarce marking data can be relieved.

In addition, in the prescription management system according to the present disclosure, optionally, the prescription management system further includes an inventory management module, and the inventory management module is configured to record warehousing information of the medicines when a new medicine is replenished, record ex-warehouse information of the medicines after the medicines are dispensed, and store the warehousing information and the ex-warehouse information in the block chain. In this case, medication safety of the user can be further ensured by stock management.

In addition, in the prescription management system according to the present disclosure, optionally, the medicine information further includes a medicine code, a medicine name, a medicine classification, a medicine toxicology classification and disease information, wherein the instruction information includes the document content and the field of interest, and the field of interest includes the medicine name, a medicine component, an indication content, a contraindication content, an adverse reaction and usage information; the auditing categories of the prescription auditing rules comprise indication medication, contraindication medication, administration, dosage, extreme amount, total amount, frequency, special population, pathophysiological state, gestational week, compatibility of medicines in the same group and incompatibility of medicines; the dispensing order comprises the prescription number, information of the dispensing end, a delivery mode, a payment mode and payment amount, and the dispensing end dispenses corresponding medicines to a user in the delivery mode.

In addition, in the prescription management system according to the present disclosure, the electronic prescription may further include user information, clinic information, department information, medical insurance information, diagnosis information, prescription time, expiration date, medication information, prescription information, trial information, dispensing information, and dispensing information. Therefore, the electronic prescription can be more comprehensive.

In addition, in the prescription management system according to the present disclosure, optionally, the medicine dispensing terminal checks validity of the electronic prescription based on the first or second result of the check and the self-information of the electronic prescription to determine whether the electronic prescription is valid, terminates dispensing of the medicine in response to invalidity of the electronic prescription, and dispenses the medicine in response to validity of the electronic prescription. Thus, the safety of medication can be improved.

In addition, in the prescription management system according to the present disclosure, optionally, the prescription auditing platform further includes a signature module and a complaint module; the signature module is used for signing the electronic prescription before the electronic prescription is transferred to the medicine taking end and the user end; and the complaint module is used for complaint when a doctor disagrees the examination result. This makes it possible to make the electronic prescription more reliable. In addition, the fluency of the electronic prescription flow can be improved.

According to the present disclosure, a prescription management system based on artificial intelligence, big data and a blockchain is provided, which can effectively perform automatic audit on electronic prescriptions to reduce labor cost and improve medication safety, convenience and economy.

Drawings

The disclosure will now be explained in further detail by way of example only with reference to the accompanying drawings, in which:

fig. 1 is a schematic diagram illustrating a block link point architecture according to an example of the present disclosure.

FIG. 2 is a schematic scenario illustrating a use prescription management system according to an example of the present disclosure.

FIG. 3 is a schematic block diagram illustrating a prescription management system according to an example of the present disclosure.

FIG. 4 is a schematic block diagram illustrating a prescription auditing platform according to examples of the present disclosure.

FIG. 5 is a flow chart illustrating a training process for a target model in accordance with an example of the present disclosure.

FIG. 6 is another schematic block diagram illustrating a prescription management system according to examples of the present disclosure.

Detailed Description

Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In the following description, the same components are denoted by the same reference numerals, and redundant description thereof is omitted. The drawings are schematic and the ratio of the dimensions of the components and the shapes of the components may be different from the actual ones.

It is noted that the terms "comprises," "comprising," and "having," and any variations thereof, in this disclosure, for example, a process, method, system, article, or apparatus that comprises or has a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include or have other steps or elements not expressly listed or inherent to such process, method, article, or apparatus. All methods described in this disclosure can be performed in any suitable order unless otherwise indicated herein or otherwise clearly contradicted by context.

The prescription management system based on artificial intelligence, big data and a block chain can effectively and automatically check electronic prescriptions so as to reduce labor cost and improve medication safety, convenience and economy. In addition, the prescription management system based on the block chain can verify the authenticity of the content in the block, and also has the property of being not tampered and traceability. The artificial intelligence, big data, and blockchain based prescription management systems to which the present disclosure relates may also sometimes be referred to as prescription management systems, prescription flow systems, or prescription systems, among others. Some concepts related to examples of the present disclosure are described below.

The prescription management system 10 (see fig. 3 for a specific structure) related to the present disclosure may have a blockchain 110, and each service using the prescription management system 10 may serve as a node of the blockchain 110. In this case, the nodes can communicate with each other to circulate the electronic prescription. In some examples, the business parties involved in the prescription management system 10 to which the present disclosure relates may include a medical facility C1, a prescription auditor C2, a regulatory facility C3, a user party C4, and a drug supply facility C5. As shown in fig. 1, in some examples, business parties such as medical facility C1, prescription auditor C2, regulatory facility C3, user party C4, and drug supply facility C5 may collectively establish a blockchain 110. In some examples, a node in blockchain 110 may be a server. Such as servers of various business parties. In some examples, the server may include, but is not limited to, a stand-alone physical server, a server cluster or distributed system of multiple physical servers, or a cloud server, among others. In some examples, the nodes in the blockchain 110 may also be terminal devices. For example, the terminal device of the user side C4.

In some examples, various service nodes may communicate with the nodes of blockchain 110 through terminal devices. In addition, the node of the blockchain 110 may perform a consensus process on each uplink request, and if the consensus passes, the uplink information in the uplink request may be stored separately (for example, the uplink information may be an electronic prescription, first audit information, second audit information, a dispensing order or delivery information, etc.). Therefore, the electronic prescription circulation tracing method can be beneficial to tracing the electronic prescription circulation in the follow-up process. In addition, uplink information may be processed and stored to the blockchain 110 via smart contracts. An intelligent contract may be a "computer trading agreement to execute contract terms". All users on blockchain 110, such as individual business parties, may see intelligent contracts based on blockchain 110.

The prescription management system 10 of the present disclosure also establishes a strict information security and institutional support system. Specifically, strict admission standards and management specifications are set, precautionary measures are set for drug stores, individuals and the like which do not operate in compliance and are processed in time, a perfect information security prevention and control system and a security emergency plan are also established, the security and confidentiality of the prescription management system 10 are improved, secret data is prevented from being decoded and accessed without authorization, and the system and the information are prevented from being leaked.

The present disclosure is described in detail below with reference to the attached drawings. In addition, the schematic scene of the present disclosure is for more clearly illustrating the technical solution of the present disclosure, and does not constitute a limitation on the technical solution provided by the present disclosure. FIG. 2 is a schematic scenario illustrating a use prescription management system 10 according to an example of the present disclosure.

In some examples, the prescription management system 10 to which the present disclosure relates may be used in a scenario as shown in fig. 2. In the scenario, medical facility C1, prescription review facility C2, regulatory facility C3 (e.g., health care, medical care, and food and drug administration), user party C4 (e.g., patient), and drug supply facility C5 (e.g., medical facility pharmacy, prescription sharing pharmacy, and self-service pharmacy) may manage the process of electronic prescription review, electronic prescription complaint, electronic prescription signature, drug withdrawal, dispensing, drug dispensing, or electronic prescription administration via prescription management system 10. Therefore, the electronic prescription can be automatically checked effectively, so that the labor cost is reduced, and the medication safety, convenience and economy are improved. In addition, medical facility C1 may include an online hospital C10 (e.g., an internet hospital) and an offline hospital C11 (e.g., secondary and tertiary medical facilities, primary health homes, and community service centers).

FIG. 3 is a schematic block diagram illustrating a prescription management system 10 according to an example of the present disclosure.

As shown in fig. 3, in some examples, prescription management system 10 may include a blockchain 110, a doctor prescription end 120, a prescription review platform 130, a user end 140, and a medication taking end 150. The doctor prescription end 120 may be used to assist in prescription and generate electronic prescriptions. The prescription audit platform 130 can automatically audit electronic prescriptions based on prescription audit rules created by an updatable drug knowledge base 131 (described later). The user terminal 140 may be used for the user to select the medication intake terminal 150 and create a medication order. The dispensing end 150 may be used to dispense medication to a user. In addition, the respective components in the prescription management system 10 can store and share the electronic prescription and the circulation information of the electronic prescription (i.e., information generated by the electronic prescription during the circulation process) based on the block chain 110. Under the condition, the medication safety of the user is guaranteed through big data, the demand of a prescription checking pharmacist can be reduced under the condition that the prescription is checked in front, the medication right and the selection right of the user are fully considered, and the medicine taking is convenient and fast. Therefore, the electronic prescription can be automatically checked effectively, so that the labor cost is reduced, and the medication safety, convenience and economy are improved. In addition, traceability of electronic prescription flow can be realized based on the block chain 110, and safety of electronic prescription flow can be improved.

As shown in fig. 3, in some examples, prescription management system 10 may include a doctor prescription end 120. In some examples, the doctor prescription end 120 may utilize the drug knowledge base 131 to assist in prescription. In this case, the prescription efficiency can be improved based on the drug knowledge base 131, and the number, variety, and quality of drugs can be ensured for rational administration. In some examples, the physician-prescription end 120 may automatically display the matched drug information (i.e., may prompt for an indication to take medication) for physician selection based on the diagnostic information via the drug knowledge base 131. In some examples, the doctor prescription end 120 may automatically display or generate usage amount information for the drug from the drug information and the user information through the drug knowledge base 131. For example, the doctor prescription end 120 may receive a doctor's operation of selecting a medicine and automatically prompt usage information based on the medicine information and user information (e.g., weight, age, physiological status) of the selected medicine. This can assist in the opening. In some examples, the doctor prescription end 120 may also be used to generate electronic prescriptions. In some examples, the physician prescription end may receive an operation of a physician to prescribe an electronic prescription to generate the electronic prescription.

In some examples, the physician-prescription end 120 may also provide medication associations based on diagnostic information, or assist in reminder functionality, at the time of physician prescription, through knowledge-based reasoning on a knowledge graph. In some examples, the knowledge-graph can be enriched with structured data and unstructured data. For example, a knowledge graph can be formed by attribute extraction techniques by parsing unstructured data (e.g., indications, contraindications, etc.) contained in the drug specification, or a knowledge graph of diagnosis and medication can be inferred in reverse from structured data such as prescription audit rules. Therefore, rich knowledge map assisted evolution can be obtained.

In some examples, the doctor prescription end 120 may store the electronic prescription to the blockchain 110. In some examples, doctor prescription end 120 may connect to blockchain 110 via medical facility C1 node and store the electronic prescription to blockchain 110. In some examples, blockchain 110 may store electronic recipes via encryption. In this case, after each component in the prescription management system 10 acquires the electronic prescription, the electronic prescription needs to be decrypted based on the agreed decryption method. Thus, the safety of electronic prescription circulation can be improved.

In some examples, the electronic prescription may have a unique prescription number. That is, the prescription number may uniquely identify an electronic prescription. In addition, the prescription number may be used to acquire the electronic prescription and the circulation information of the electronic prescription in the prescription management system 10.

In some examples, the electronic prescription may also include user information, clinic information, department information, medical insurance information, diagnostic information, time of prescription, expiration date, medication information, prescription information, trial information, dispensing information, and dispensing information. Therefore, the electronic prescription can be more comprehensive. In some examples, if the dispensing and the dispensing are the same party, the dispensing and dispensing information may be the same. In some examples, after the electronic prescription is circulated among the various components in the prescription management system 10, circulation information of the corresponding electronic prescription may also be generated. In some examples, the circulation information may include audit information (e.g., first audit information and/or second audit information), pharmacy orders, and delivery information.

FIG. 4 is a schematic block diagram illustrating a prescription auditing platform 130 according to examples of the present disclosure.

As shown in fig. 3, in some examples, the prescription management system 10 may include a prescription auditing platform 130. In some examples, as shown in FIG. 4, prescription review platform 130 may include a drug knowledge base 131 and a review engine 132. Drug knowledge base 131 may be used to store a variety of drug information. The review engine 132 may be used to perform at least one review of the electronic prescription to obtain a review result. In some examples, the review results may include first review results and/or second review results. In some examples, the prescription audit platform 130 may be connected to the blockchain 110 via a prescription audit C2 node.

In some examples, the drug knowledge base 131 may be updated on a regular or irregular basis to keep the drug knowledge base 131 active. Accordingly, effective and comprehensive prescription auditing rules can be subsequently established on the basis of the drug knowledge base 131, and development by doctors can be assisted. In some examples, the drug knowledge base 131 may be updated synchronously with a catalog of drugs that is periodically or aperiodically included from a third party platform (e.g., a food and drug administration). For example, the drug knowledge base 131 may cover at least 2 million and 6 million drug catalogs, and thereby establish over 170 million prescription audit rules.

In some examples, the drug information may include instructional information. In some examples, the drug information may also include drug codes, drug names, drug classifications (e.g., prescription and over-the-counter drugs), drug toxicology classifications, and disease information. Therefore, the medicine information can be more comprehensive. In some examples, the specification information may include document content in the drug specification and fields of interest in the drug specification. In some examples, the fields of interest may include drug name, drug composition, indication content, contraindication content, adverse reactions, and usage amount information.

In some examples, the drug information may be obtained based on artificial intelligence and Natural Language Processing (NLP) techniques to update the drug knowledge base 131. In some examples, the manual information in the drug knowledge base 131 may be obtained based on artificial intelligence and natural language processing techniques. Generally, the drug instruction book is an advantageous tool for describing the drug information more comprehensively and accurately, but the drug instruction book is often paper, the drug information in the drug instruction book cannot be obtained in batches, and the manual extraction of the drug information in the paper drug instruction book is time-consuming and is not beneficial to updating the drug knowledge base 131. In some examples, the document content in the paper drug specification may be extracted using Optical Character Recognition (OCR) and unstructured data in the document content may be identified by artificial intelligence and natural language processing techniques, and the unstructured data may be subjected to big data processing to obtain structured fields of interest. In this case, the field of interest in the paper drug specification can be quickly obtained, and the drug knowledge base 131 can be conveniently and quickly updated. This enables the drug knowledge base 131 to be effectively maintained.

In addition, unstructured data may be data that has an incomplete or irregular data structure, no predefined data model, and is inconvenient to represent with a database two-dimensional logical table. In addition, the structured data may be data logically expressed and implemented by a two-dimensional table structure, strictly following the data format and length specifications, and mainly stored and managed by a relational database.

In some examples, identifying unstructured data in the document content through artificial intelligence and natural language processing techniques may be for entering target text corresponding to the document content in the pharmaceutical specification into an artificial intelligence based target model to identify unstructured data in the document content. As such, unstructured data can be identified based on artificial intelligence and natural language processing techniques.

In some examples, big data processing on unstructured data may include data storage and data extraction. In some examples, in the data store, unstructured data may be stored into a data warehouse and processed for consistency, generality of the unstructured data. For example, in the consistency process, the data names may be kept consistent. For another example, in the general processing, the format of the attribute related to time in the unstructured data may be uniformly modified into a fixed arrangement format to specify the time attribute. In some examples, unstructured data may be stored to a data warehouse in a sliced store. For example, unstructured data may be stored in time by time. In some examples, unstructured data in the data warehouse may be modeled to determine the library tables of the drug specification itself and of the backend systems on which it depends, as well as the relationships between the various library tables. In some examples, in data extraction, multiple copies of data may be retrieved from the database tables of the data warehouse and combined in association with each other to generate a new database table, and then a query may be performed on the basis of the new database table or combined in association with other data. In this case, the manual information can be acquired by the big data, and the drug knowledge base can be built. Therefore, the medication safety of the user can be guaranteed through big data.

In some examples, the training process of the target model includes constructing training samples (step S102) and training the artificial intelligence based model with the training samples to obtain the target model (step S104). Thus, the target model can be obtained through artificial intelligence and natural language processing technology.

In some examples, in step S102, the document content to be trained may be preprocessed to obtain a target text, and a domain prior dictionary may be constructed as a training sample based on the target text corresponding to the document content to be trained. In addition, the document content to be trained may be the document content in the drug manual for training.

In some examples, in the preprocessing, after filtering the invalid characters in the content of the document to be trained, the long text in the content of the document to be trained is segmented to obtain a target text corresponding to the content of the document to be trained, and the target text is labeled. In some examples, if the text length of the target text is less than a first preset text length (e.g., 128 bytes or 256 bytes), the target text may be merged such that the merged text length is not less than the first preset text length and is greater than a second preset text length (e.g., the maximum text length may be adjusted according to actual conditions). In some examples, the preprocessing may also include word segmentation, part-of-speech tagging, and stop-word. This can further improve the quality of the training sample.

In some examples, the target text may be converted into vectors to enable recognition by the artificial intelligence-based model (i.e., the training samples may be represented in vector form). For example, a Bag of words model (Bag of Word, BOW) may be used to convert the target text into a vector.

In some examples, in step S104, when training the artificial intelligence-based model, a basic model may be trained first by using a domain prior dictionary (i.e., a target text corresponding to a labeled document content to be trained), pseudo-labeled data may be obtained by predicting unlabeled data (i.e., a target text corresponding to an unlabeled document content to be trained) by using the basic model, and the pseudo-labeled data is added to a training sample to jointly train the artificial intelligence-based model to obtain the target model. Therefore, the problem of scarce marking data can be relieved. In some examples, a dynamic and learnable weight may be set for pseudo-annotation data when calculating the loss. This can reduce the noise of the pseudo label data. In some examples, the weight of the pseudo-annotation data may also be a fixed value, which may be obtained from multiple attempts (i.e., multiple training).

In some examples, the artificial intelligence based model may be pre-trained with a generic chinese corpus and then trained with training samples to obtain the target model. This can improve training efficiency.

In some examples, one training sample may be constructed for each field in the unstructured data. In this case, there are a large number of negative samples in training, a certain proportion (e.g. 30%) of the negative samples are randomly selected for training, and the rest are discarded. Therefore, training efficiency can be guaranteed.

As described above, in some examples, prescription review platform 130 may include an review engine 132 (see fig. 4). In some examples, the review engine 132 can include a first review module 1321. In some examples, the review engine 132 may also include a second review module 1322.

In some examples, the first review module 1321 can perform an automatic review of the electronic prescription. In some examples, the first review module 1321 can automatically review the electronic prescription based on the prescription review rules to obtain first review information. Therefore, the efficiency of checking the electronic prescription can be improved, and omission of manual checking can be avoided. In addition, the first review information may include the first review result and the first review suggestion. In some examples, the first review results may include flowable, non-flowable, and intercepted. In addition, the first review proposal can be a text description corresponding to the first review result. As an example, the first review proposal may be, for example, "reason for interception: caplet ophthalmic gel, with caution in children. Therefore, the explanation corresponding to the auditing result can be intuitively obtained.

In some examples, the first review module 1321 can obtain the electronic prescription from the blockchain 110 according to the prescription number, and automatically review the electronic prescription based on the prescription review rules to obtain the first review information. In some examples, the first review module 1321 can store the first review information to the blockchain 110. In this case, the efficiency of the electronic prescription audit can be improved, and the security of the electronic prescription flow can be improved based on the block chain 110.

In some examples, after storing the first review information to the blockchain 110, the electronic prescription stream can be transferred to the medication intake 150 and the user terminal 140. Specifically, in the case where the electronic prescription is confirmed to be qualified based on the first review result, the electronic prescription is desensitized and then transferred to the medicine dispensing terminal 150 and the user terminal 140. Thus, the privacy of the user can be protected. In some examples, if the first review result of the electronic prescription is transferable, the electronic prescription may be qualified. In some examples, the electronic prescription may be rejected if the first review result of the electronic prescription is non-transferable. That is, the rejected prescription may refer to an electronic prescription whose first review result is non-transferable. In some examples, if the first review result of the electronic prescription is an interception, the electronic prescription may be reviewed by the second review module 1322 as the first review prescription. That is, the first pending prescription may refer to an electronic prescription whose first result of review is an interception.

As described above, the first review module 1321 may perform automatic review on the electronic prescription based on the prescription review rule to acquire first review information. In some examples, prescription audit rules may be created based on the drug knowledge base 131 and updated with the drug knowledge base 131. In this case, the prescription audit rules can be updated in time. This can maintain the validity of the prescription audit rule. In some examples, the drug information in the drug knowledge base 131 may be classified to obtain audit categories, a conversion rule for converting the drug information corresponding to each audit category into a prescription audit rule is defined for each audit category, and then the prescription audit rule is automatically updated based on the conversion rule after the drug knowledge base 131 is identified to be updated. Therefore, prescription audit rules can be updated conveniently based on the drug knowledge base 131, and omission of manual analysis of audit rules can be avoided. Examples of the disclosure are not limited thereto and in other examples, review rules for prescriptions that are not related to drug knowledge base 131 may also be manually set. For example, the validity rules of the electronic prescription may be set manually.

In some examples, audit categories for prescription audit rules may include, but are not limited to, indication medications, contraindication medications, usage (e.g., intravenous drip or oral), amount, maximum, total, frequency, specific population, pathophysiological state, week of pregnancy, co-drug compatibility and drug incompatibility, and the like. In this case, prescription audit rules can be set based on a plurality of audit categories. This enables prescription audit rules to be managed easily.

As described above, in some examples, the review engine 132 may also include a second review module 1322 (see fig. 4). In some examples, the second review module 1322 may be configured to receive an operation of a reviewer pharmacist reviewing a prescription to be reviewed based on the first review information to generate second review information. In this case, the manual review of the screened electronic prescription can be simultaneously supported. This can improve the flexibility of electronic prescription verification. In addition, the second review information may include second review results and second review suggestions. In some examples, the second audit results may include both flowable (i.e., pass) and non-flowable (i.e., not pass). In addition, the second review proposal may be a text description corresponding to the second review result.

In some examples, the pending prescription may include a first pending prescription and/or an updated refund prescription. That is, the first pending prescription and/or the rejected prescription may be reviewed by the second review module 1322 after being updated. In some examples, the pending review prescription may also include a pre-set proportion of eligible electronic prescriptions. In this case, the qualified electronic prescription that is automatically checked can be subjected to quality control. Therefore, the quality of automatic audit can be improved.

In some examples, the second review module 1322 may store the second review information to the blockchain 110. In this case, the security of the electronic prescription flow can be improved based on the block chain 110. In some examples, the second review module 1322 may pass the flow of prescriptions to be reviewed to the medication intake 150 and the user end 140 after storing the second review information in the blockchain 110. Specifically, in the case that the prescription to be reviewed is confirmed to be qualified based on the second review result, the desensitization processing is performed on the prescription to be reviewed, and then the desensitization processing is transferred to the medicine dispensing terminal 150 and the user terminal 140. Thus, the privacy of the user can be protected. In some examples, if the second review result of the recipe to be reviewed is flowable, then the recipe to be reviewed may be qualified. In some examples, if the second review result for the pending prescription is non-flowable, the pending prescription may be rejected. In some examples, the prescription to be reviewed rejected by second review module 1322 can be complained through a complaint module (not shown, described later).

In some examples, prescription audit platform 130 may also include a signature module (not shown). The signature module may be used to sign the electronic prescription before the electronic prescription is transferred to the dispensing end 150 and the user end 140. This makes it possible to make the electronic prescription more reliable. In some examples, the signature may be an electronic signature.

In some examples, the prescription review platform 130 may also include a complaint module. The complaint module can be used for complaint when the doctor disagrees with the examination result (for example, the first examination result or the second examination result). This can improve the fluency of the electronic prescription flow. In some examples, the electronic prescription complaining of success may continue to flow to the customer end 140 and the medication intake end 150. See the relevant description of a qualified electronic prescription flow.

As shown in fig. 3, in some examples, prescription management system 10 may include a user terminal 140. In some examples, the user terminal 140 may be configured to obtain electronic prescriptions, select a dispensing terminal 150 according to the self-location information and the inventory information and price information of the dispensing terminal 150, and create a dispensing order based on the electronic prescriptions. For example, the user terminal 140 may assist the user in comparing prices of drugs, displaying drug inventories, or displaying locations for taking drugs, etc. to assist the user in selecting a closer or more favorable medication taking terminal 150 to purchase drugs by price, distance, etc. In this case, the medication right and the option right of the user can be sufficiently secured. Therefore, the convenience and the economy of medicine application of the user can be improved.

In some examples, the fill order may include a prescription number, information of the pharmacy 150, a delivery method, a payment method, and a payment amount. Thus, more comprehensive dispensing information can be obtained. In some examples, the user end 140 may configure the delivery pattern of the medication order. For example, the distribution mode may include, but is not limited to, a network order pick-up, a network order delivery, a drug cabinet pick-up, and the like. In some examples, the medication intake terminal 150 may dispense the corresponding medication to the user in a delivery manner. This enables the user to take the medicine easily.

In some examples, the client 140 may store the medication order to the blockchain 110. In this case, the security of the electronic prescription flow can be improved based on the block chain 110. In some examples, the user end 140 may connect to the blockchain 110 via the user side C4 node and store the medication order to the blockchain 110.

In some examples, the user terminal 140 may be at least one of a desktop application, a mobile application, and a wechat applet. Thus, multiple ways of selecting a pharmacy order and creating a pharmacy order can be supported.

In some examples, the user terminal 140 may include a payment module (not shown). The payment module may pay for the drugs in the pharmacy order. This enables the cost of the medicine to be conveniently paid.

As described above, in some examples, prescription management system 10 may include a medication intake 150. In some examples, the drug terminal 150 may be configured to obtain electronic prescriptions and fill orders from the blockchain 110 according to the prescription numbers, and if the electronic prescriptions are verified to be valid, dispense corresponding drugs to the user according to the fill orders. As described above, the dispensing end 150 may dispense the corresponding medication to the user in a delivery manner.

In some examples, the fetching end 150 may store the delivery information to the blockchain 110. In this case, the security of the electronic prescription flow can be improved based on the block chain 110. In some examples, the terminal 150 may connect to the blockchain 110 via the drug delivery facility C5 node and store the delivery information to the blockchain 110. In some examples, the delivery information may include logistics information. In addition, the logistics information can be obtained by the user terminal 140 and displayed for the user to view. Therefore, the distribution progress of the medicine can be obtained in time.

In some examples, the medicine dispensing terminal 150 may check the validity of the electronic prescription based on the first or second audit result and the self-information of the electronic prescription to determine whether the electronic prescription is valid, terminate dispensing the medicine in response to the electronic prescription being invalid, and dispense the medicine in response to the electronic prescription being valid. For example, the electronic prescription expiration date may be checked to confirm whether the electronic prescription is expired. Thus, the safety of medication can be improved.

In some examples, the prescription administration system 10 may have multiple dispensing ends 150. In some examples, the plurality of medication intake ends 150 may include, but are not limited to, a medical facility pharmacy, a social pharmacy, a third party drug delivery company, or a kiosk, among others. Therefore, the convenience of taking the medicine can be improved.

FIG. 6 is another schematic block diagram illustrating a prescription management system 10 according to an example of the present disclosure.

As shown in fig. 6, in some examples, prescription management system 10 may also include an inventory management module 160 and a supervisor module 170.

In some examples, the inventory management module 160 may be configured to record warehousing information of the drug when a new drug is replenished, record ex-warehousing information of the drug after the drug is dispensed, and store the warehousing information and the ex-warehousing information to the blockchain 110. In this case, medication safety of the user can be further ensured by stock management. In some examples, the inventory management module 160 may connect to the blockchain 110 via the drug delivery facility C5 node to store warehousing information and ex-warehousing information to the blockchain 110.

In some examples, the administration module 170 may be used to query the electronic prescription and the flow information of the electronic prescription from the blockchain 110 for review by the regulatory authority. In this case, an effective management means can be provided for the monitoring department. Therefore, the safety of the medicine taking of the user can be further guaranteed. In some examples, the administration module 170 may connect to the blockchain 110 via the administration C3 node to query the electronic prescription.

The prescription management system 10 related to the present disclosure assists prescription issuing and generates an electronic prescription based on the drug knowledge base 131, and automatically checks the electronic prescription based on a prescription check rule created by the updatable drug knowledge base 131, the checked electronic prescription flow is subjected to desensitization processing and then transferred to the user terminal 140 and the medicine taking terminal 150, the user selects the medicine taking terminal 150 and creates a medicine dispensing order through the user terminal 140 according to personal preference, and then the medicine taking terminal 150 can dispense medicines to the user, wherein the drug knowledge base 131 is updated through artificial intelligence, natural language processing technology and big data, and the prescription check rule is updated along with the drug knowledge base 131. In addition, the various components in the prescription management system 10 may store and share electronic prescriptions and flow information for electronic prescriptions based on the blockchain 110. In this case, the drug knowledge base 131 is updated intelligently to ensure the medication safety of the user through big data, and the required amount of the prescription checking pharmacist can be reduced through automatic checking under the condition of pre-checking of the prescription, and the medication right and the selection right of the user are fully considered for taking the medicine, so that the medicine taking is convenient and fast. Therefore, the electronic prescription can be automatically checked effectively, so that the labor cost is reduced, and the medication safety, convenience and economy are improved. In addition, monitoring is performed based on the block chain 110, so that traceability of electronic prescription circulation can be achieved, and safety of electronic prescription circulation can be improved. In addition, the purchase, sale and storage data of the medicine are monitored to complete the data tracking of the medicine from the whole process of medical institutions, circulation and distribution, so that the medicine taking safety and timeliness of patients are guaranteed, and an effective management means is provided for supervision departments.

While the present disclosure has been described in detail in connection with the drawings and examples, it should be understood that the above description is not intended to limit the disclosure in any way. Those skilled in the art can make modifications and variations to the present disclosure as needed without departing from the true spirit and scope of the disclosure, which fall within the scope of the disclosure.

Claims

1. A prescription management system based on artificial intelligence, big data and a block chain is characterized by comprising the block chain, a doctor prescription end, a prescription auditing platform, a user end and a medicine taking end, wherein the prescription auditing platform comprises a medicine knowledge base and an auditing engine;

the doctor prescription end utilizes the medicine knowledge base to carry out auxiliary prescription and generate an electronic prescription, and then the electronic prescription is stored in the block chain, wherein the electronic prescription has a unique prescription number;

the medicine knowledge base is used for storing a plurality of kinds of medicine information, wherein the instruction information in the medicine information is obtained by extracting document contents in a paper medicine instruction by using an optical character recognition technology, recognizing unstructured data in the document contents by using artificial intelligence and a natural language processing technology, and performing big data processing on the unstructured data to obtain a structured field of interest;

the auditing engine comprises a first auditing module, the first auditing module acquires the electronic prescription from the block chain according to the prescription number, automatically audits the electronic prescription based on prescription auditing rules to acquire first auditing information comprising a first auditing result and a first auditing proposal, stores the first auditing information into the block chain, desensitizes the electronic prescription and transfers the electronic prescription to the medicine taking end and the user end under the condition that the electronic prescription is qualified based on the first auditing result, wherein the prescription auditing rules are established on the basis of the medicine knowledge base and are updated along with the medicine knowledge base;

the user side is used for acquiring the electronic prescription, selecting a medicine taking end according to self positioning information and inventory information and price information of the medicine taking end, creating a medicine dispensing order based on the electronic prescription and storing the medicine dispensing order to the block chain;

and the medicine taking end is used for obtaining the electronic prescription and the dispensing order from the block chain according to the prescription number, and under the condition that the electronic prescription is verified to be effective, corresponding medicines are dispensed to a user according to the dispensing order and delivery information is stored in the block chain.

2. The prescription management system of claim 1, wherein:

the auditing engine further comprises a second auditing module, wherein the second auditing module is used for receiving the operation of auditing the prescription to be reviewed including the first prescription to be reviewed and/or the updated rejected prescription by a prescription pharmacist based on the first auditing information so as to generate second auditing information including a second auditing result and a second auditing suggestion, storing the second auditing information into the block chain, and under the condition that the prescription to be reviewed is determined to be qualified based on the second auditing result, desensitizing the prescription to be reviewed and transferring the prescription to the medicine taking end and the user end, wherein the first prescription to be reviewed is an intercepted electronic prescription, and the rejected prescription is an electronic prescription which can not be transferred.

3. The prescription management system of claim 1, wherein:

the prescription management system further comprises a supervision module, and the supervision module is used for inquiring the electronic prescription and circulation information of the electronic prescription from the block chain so as to be examined by a supervision department.

4. The prescription management system of claim 1, wherein identifying unstructured data in the document content through artificial intelligence and natural language processing techniques is:

inputting target texts corresponding to the document contents into an artificial intelligence-based target model to identify unstructured data in the document contents, wherein the training process of the target model comprises the following steps:

after filtering invalid characters in document content to be trained, segmenting a long text in the document content to be trained to obtain a target text corresponding to the document content to be trained, and labeling the target text to construct a domain prior dictionary as a training sample, wherein if the text length of the target text is smaller than a first preset text length, the target text is merged to enable the merged text length to be not smaller than the first preset text length and larger than a second preset text length; and is

Training an artificial intelligence based model with the training samples to obtain the target model.

5. The prescription management system of claim 4, wherein:

when the model based on the artificial intelligence is trained, a basic model is trained by utilizing the domain prior dictionary, the basic model is utilized to predict the data which are not labeled to obtain pseudo-labeled data, and the pseudo-labeled data is added into the training sample to jointly train the model based on the artificial intelligence to obtain the target model.

6. The prescription management system of claim 1, wherein:

the prescription management system further comprises an inventory management module, wherein the inventory management module is used for recording warehousing information of the medicines when new medicines are supplemented, recording ex-warehouse information of the medicines after the medicines are issued, and storing the warehousing information and the ex-warehouse information to the block chain.

7. The prescription management system of claim 1, wherein:

the medicine information further comprises medicine codes, medicine names, medicine classifications, medicine toxicology classifications and disease information, wherein the instruction information comprises the document content and the interested field, and the interested field comprises the medicine names, medicine components, indication content, contraindication content, adverse reactions and usage information; the auditing categories of the prescription auditing rules comprise indication medication, contraindication medication, administration, dosage, extreme amount, total amount, frequency, special population, pathophysiological state, gestational week, compatibility of medicines in the same group and incompatibility of medicines; the dispensing order comprises the prescription number, information of the dispensing end, a delivery mode, a payment mode and payment amount, and the dispensing end dispenses corresponding medicines to a user in the delivery mode.

8. The prescription management system of claim 1, wherein:

the electronic prescription further comprises user information, clinic information, department information, medical insurance information, diagnosis information, prescription time, validity period, medication information, prescription information, trial information, prescription information and dispensing information.

9. The prescription management system of claim 2, wherein:

the medicine taking terminal checks the validity of the electronic prescription based on the first or second checking result and the self-information of the electronic prescription to determine whether the electronic prescription is valid, terminates the dispensing of the medicine in response to the electronic prescription being invalid, and dispenses the medicine in response to the electronic prescription being valid.

10. The prescription management system according to claim 1 or 2, wherein:

the prescription auditing platform also comprises a signature module and a complaint module; the signature module is used for signing the electronic prescription before the electronic prescription is transferred to the medicine taking end and the user end; and the complaint module is used for complaint when a doctor disagrees the examination result.