Drug utilization value analysis system based on data flow

Abstract

The invention relates to the field of drug utilization value analysis, and particularly discloses a data stream-based drug utilization value analysis system, which comprises a data layer, an application layer and a user layer, according to the method, the multi-source drug data stream is constructed and stored in the multi-source drug database, and based on systematic management of drug information, drug use errors caused by drug name confusion, dosage errors and dosage form confusion can be reduced, the drug use safety is improved, and the method is beneficial to quickly and accurately retrieving and querying specific types of drug information; by integrating the elderly physiological features and the drug feature information, the drug utilization value analysis model is constructed, the elderly physiological features and the drug features are comprehensively considered, the utilization value of the drug on the elderly can be comprehensively evaluated, and the influence of the drug on the elderly patient can be predicted more accurately.

Description

Technical Field

[0001] This invention relates to the field of drug utilization value analysis, and more specifically, to a data flow-based drug utilization value analysis system. Background Technology

[0002] Due to the vast variety of medication information in hospitals, both medical staff and patients often struggle to accurately remember detailed information, leading to confusion. This confusion arises from various sources, including: similar drug names (e.g., similar letter order or pronunciation, like Celebrex and Celexa); similar dosage forms (e.g., liquid preparations, oral tablets, and injections); different dosage units (e.g., grams versus milligrams); and different frequencies and timings of use. For elderly patients recovering post-surgery, who often require multiple medications, the risk of cognitive and visual impairments at home can lead to confusion regarding drug names, dosage forms, and dosages, resulting in medication errors, potentially causing relapses or even worsening of their condition and reducing the effectiveness of the medications. Furthermore, the decline in physiological function in the elderly affects drug metabolism and excretion, making it difficult to accurately predict the effectiveness of medications in this population. To address these issues, a technical solution is proposed. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, this invention provides a drug utilization value analysis system based on data flow. By constructing a multi-source drug data flow and storing it in a multi-source drug database, and through systematic management of drug information, it can reduce drug misuse caused by confusion in drug names, dosage errors, and dosage forms, thereby improving medication safety and facilitating the rapid and accurate retrieval and querying of specific types of drug information. By integrating information on the physiological characteristics and drug characteristics of the elderly, a drug utilization value analysis model is constructed. By comprehensively considering both physiological and drug characteristics, it can comprehensively assess the utilization value of drugs in the elderly and help to more accurately predict the impact of drugs on elderly patients, thus solving the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] The data-flow-based drug utilization value analysis system comprises a data layer, an application layer, and a user layer. The data layer constructs multi-source drug data streams and stores them in a multi-source drug database. The application layer builds a drug utilization value analysis model based on the multi-source drug database. The application layer includes modules for geriatric physiological analysis, drug information screening, feature information interaction, drug utilization value analysis, drug use monitoring, and alerts. The drug utilization value analysis module establishes the drug utilization value analysis model, and the formula for the model is:

[0006] D VALUE =CS×H final +(1-CS)E fusion ;

[0007] In the formula: D VALUE The results of the drug utilization value analysis are shown, where CS represents the spatial similarity between feature nodes, and H represents the spatial similarity between feature nodes. final For the final physiological information feature representation, E fusion This is a fused representation of word-level feature vectors.

[0008] As a further embodiment of the present invention, the data layer includes a drug information input module, a drug information classification module, a drug data distribution module, a multi-source database management module, and a multi-source data storage module; the drug information input module is used to input drug information; the drug information classification module is used to classify drug information; the drug data distribution module is used to distribute the classified drug information to the corresponding drug data stream; and the multi-source database management module is used to generate a multi-source drug database and store drug information.

[0009] As a further embodiment of the present invention, the drug information entry module is connected to the drug information classification module, the drug information classification module is connected to the drug data distribution module, the drug data distribution module is connected to the multi-source database management module, and the multi-source database management module is connected to the multi-source data storage module.

[0010] As a further aspect of the present invention, the multi-source drug data stream includes a drug name stream, a drug dosage stream, a drug dosage form stream, a drug instruction stream, a drug efficacy stream, and a drug contraindication stream; the multi-source drug database includes a drug name database, a drug dosage database, a drug dosage form database, a drug instruction database, a drug efficacy database, and a drug contraindication database.

[0011] As a further aspect of the present invention, the geriatric physiological analysis module is used to analyze and extract physiological characteristic information from geriatric physiological information; the drug information screening module is used to screen drug characteristic information from a multi-source drug database; the characteristic information interaction module is used to realize the interaction between geriatric physiological characteristic information and drug characteristic information; the drug utilization value analysis module is used to establish a drug utilization value analysis model through geriatric physiological characteristic information and drug characteristic information; the drug use monitoring module is used to perform a secondary calculation of the results of the drug utilization value analysis model; and the reminder module is used to issue a reminder when the deviation between the calculation results of the drug use monitoring module and the drug utilization value analysis module exceeds 5%.

[0012] As a further embodiment of the present invention, the elderly physiological analysis module is connected to the drug information screening module, the drug information screening module is connected to the feature information interaction module, the feature information interaction module is connected to the drug utilization value analysis module, the drug utilization value analysis module is connected to the drug use monitoring module, and the drug use monitoring module is connected to the reminder module.

[0013] As a further embodiment of the present invention, the elderly physiological analysis module includes a physiological information collection unit, a physiological characteristic information representation unit, and a physiological characteristic information output unit; the physiological information collection unit is used to collect elderly physiological information, including age, weight, gender, height, liver function characteristics, kidney function characteristics, and blood-brain barrier function characteristics; the physiological characteristic information representation unit is used to analyze and extract physiological characteristic information from the physiological information; and the physiological characteristic information output unit is used to output physiological characteristic information.

[0014] As a further embodiment of the present invention, the physiological information collection unit is connected to the physiological feature information representation unit, and the physiological feature information representation unit is connected to the physiological feature information output unit.

[0015] As a further aspect of the present invention, the drug information screening module is used to screen drug feature information from a multi-source drug database. The specific steps for screening drug feature information are as follows:

[0016] Step 1, Drug Information Processing: Extract drug information from multi-source drug databases and preprocess the extracted drug information, including handling missing values, outlier values, and removing duplicate values.

[0017] Step 2, Drug Feature Extraction: Extract features from the processed drug information, extracting the textual feature information for each drug;

[0018] Step 3, Drug Feature Information Segmentation: Use Chinese word segmentation tools to segment the text features of each drug into word granularities of different lengths;

[0019] Step 4, Extract Granularity Feature Vectors: Extract word granularity feature vectors of different lengths:

[0020] Wg EIGEN =Tanh(W*Y N +B);

[0021] In the formula: Wg EIGEN Y is the word-level feature vector, Tanh is the Tanh activation function, W is the learnable weight matrix, and Y is the word-level feature vector. N Input words of different lengths, where B is the bias vector and N is the total number of word granularities;

[0022] Step 5, Granular Feature Vector Fusion: Word-level feature vectors are fused using a residual connection method. The formula for fusing word-level feature vectors is as follows:

[0023]

[0024] In the formula: E fusion This represents the fusion of word-level feature vectors, where ReLU is the ReLU activation function, F(Wg EIGEN Wg represents the residual mapping of word-granularity feature vectors. EIGEN is the word-granularity feature vector, and N is the total number of word granularities;

[0025] Step 6: Obtain drug feature information: Use the fusion representation of word-level feature vectors as drug feature information.

[0026] The technical effects and advantages of the drug utilization value analysis system based on data flow of this invention are as follows: By constructing a multi-source drug data flow and storing it in a multi-source drug database, this invention can reduce drug misuse caused by confusion in drug names, dosage errors, and dosage forms through systematic management of drug information, thereby improving medication safety and facilitating the rapid and accurate retrieval and query of specific types of drug information. By integrating the physiological characteristics and drug characteristics of the elderly, a drug utilization value analysis model is constructed. This model comprehensively considers both physiological and drug characteristics of the elderly, enabling a comprehensive assessment of the utilization value of drugs in the elderly and helping to more accurately predict the impact of drugs on elderly patients. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the structure of a data flow-based drug utilization value analysis system provided in Embodiment 1 of the present invention;

[0028] Figure 2 This is a schematic diagram of the data layer for storing drug data provided in Embodiment 2 of the present invention;

[0029] Figure 3 This is a schematic diagram of the process for constructing a drug utilization value analysis model at the application layer, as provided in Embodiment 3 of the present invention. Detailed Implementation

[0030] The technical solutions of this invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described technical solutions are only a part of this invention, and not all of it. All other technical solutions obtained by those skilled in the art based on the technical solutions of this invention without inventive effort are within the scope of protection of this invention.

[0031] Example 1

[0032] Figure 1 A schematic diagram of the structure of a data flow-based drug utilization value analysis system provided in Embodiment 1 of the present invention is shown. Figure 1 As shown, the drug utilization value analysis system based on data flow in this embodiment includes a data layer, an application layer, and a user layer. The data layer is connected to the application layer, the application layer is connected to the user layer, and the data layer is connected to the user layer.

[0033] The data layer is used to construct multi-source drug data streams and store them in a multi-source drug database;

[0034] The application layer is used to construct drug utilization value analysis models based on multi-source drug databases;

[0035] The user layer is used to provide information services and information access interfaces.

[0036] In this embodiment of the invention, the data layer includes a drug information entry module, a drug information classification module, a drug data distribution module, a multi-source database management module, and a multi-source data storage module; the drug information entry module is connected to the drug information classification module, the drug information classification module is connected to the drug data distribution module, the drug data distribution module is connected to the multi-source database management module, and the multi-source database management module is connected to the multi-source data storage module.

[0037] The drug information entry module is used to enter drug information; the drug information classification module is used to classify drug information; the drug data distribution module is used to distribute the classified drug information to the corresponding drug data stream; and the multi-source database management module is used to generate a multi-source drug database and store drug information.

[0038] Multi-source drug data streams include drug name stream, drug dosage stream, drug dosage form stream, drug instructions for use stream, drug efficacy stream, and drug contraindication stream. The drug name stream is used to uniquely identify and recognize specific drugs, including the drug's official name, generic name, and brand name; the drug dosage stream provides dosage guidance for drug use, including recommended or actual dosage information; the drug dosage form stream guides patients in the correct use of drugs, ensuring they receive the appropriate form, including dosage forms such as tablets, capsules, and injections; the drug instructions for use stream provides patients and healthcare professionals with detailed information on drug use, including methods of use, purpose, frequency of administration, and precautions; the drug efficacy stream provides information on the therapeutic effects and mechanisms of action of drugs, including efficacy, mechanism of action, and disease treatment outcomes; and the drug contraindication stream helps physicians and patients understand the potential risks of drugs and prevent their use in specific patient groups or for patients interacting with other drugs, including contraindications, adverse reactions, and existing risks.

[0039] The drug information entry module enables rapid input of drug information into the system, while the multi-source database management module helps to centrally manage drug information from different sources, improving information integration and management efficiency. The drug information classification module effectively categorizes drug information, while the drug data distribution module distributes the categorized information to the corresponding data streams, making drug information more organized and easier to retrieve and use. The drug name stream is used to uniquely identify and recognize specific drugs, including official names, generic names, and brand names, helping to avoid drug misuse due to naming confusion. The drug dosage stream, drug formulation stream, drug instructions for use stream, and drug contraindications stream provide detailed medication guidance and safety information for patients and healthcare professionals, helping to ensure that patients receive the appropriate drug form and use the drug correctly. The drug efficacy stream provides information on the therapeutic effects and mechanisms of drug treatment, helping doctors and patients understand the efficacy of drugs.

[0040] The multi-source drug database includes a drug name database, a drug dosage database, a drug dosage form database, a drug instructions for use database, a drug efficacy database, and a drug contraindication database. The drug name database stores data related to drug names; the drug dosage database stores data related to drug dosage forms; the drug instructions for use database stores data related to drug instructions for use; the drug efficacy database stores data related to drug efficacy; and the drug contraindication database stores data related to drug contraindications.

[0041] In this embodiment of the invention, the application layer includes an elderly physiological analysis module, a drug information screening module, a feature information interaction module, a drug utilization value analysis module, a drug use monitoring module, and a reminder module; the elderly physiological analysis module is connected to the drug information screening module, the drug information screening module is connected to the feature information interaction module, the feature information interaction module is connected to the drug utilization value analysis module, the drug utilization value analysis module is connected to the drug use monitoring module, and the drug use monitoring module is connected to the reminder module.

[0042] The geriatric physiological analysis module is used to analyze and extract physiological characteristic information from geriatric physiological information; the drug information screening module is used to screen drug characteristic information from a multi-source drug database; the characteristic information interaction module is used to realize the interaction between geriatric physiological characteristic information and drug characteristic information; the drug utilization value analysis module is used to establish a drug utilization value analysis model through geriatric physiological characteristic information and drug characteristic information; the drug use monitoring module is used to perform a secondary calculation of the results of the drug utilization value analysis model; and the reminder module is used to issue a reminder when the deviation between the calculation results of the drug use monitoring module and the drug utilization value analysis module exceeds 5%.

[0043] The elderly physiological analysis module includes a physiological information collection unit, a physiological characteristic information representation unit, and a physiological characteristic information output unit. The physiological information collection unit is connected to the physiological characteristic information representation unit, and the physiological characteristic information representation unit is connected to the physiological characteristic information output unit. The physiological information collection unit is used to collect elderly physiological information, including age, weight, sex, height, liver function characteristics, kidney function characteristics, and blood-brain barrier function characteristics. The physiological characteristic information representation unit is used to analyze and extract physiological characteristic information from the physiological information. The physiological characteristic information output unit is used to output the physiological characteristic information.

[0044] The physiological feature information representation unit is used to analyze and extract physiological feature information from physiological information. The steps for extracting physiological feature information from physiological information are as follows:

[0045] Step B1: Numericalize the physiological information to ensure that each physiological information is represented in digital form; gender is represented in binary form, with male represented as 1 and female represented as 0.

[0046] Step B2: The physiological information is organized into a physiological information input matrix, where each row corresponds to an elderly sample and each column corresponds to a physiological feature.

[0047] Step B3: Extract the features of each physiological information:

[0048] S eigen =RELU(W e *X m +b);

[0049] In the formula: S eigen For physiological information feature representation, ReLU is the ReLU activation function, W e Let X be the weight matrix. m Here, b is the physiological information input matrix, and b is the bias term.

[0050] Step B4: Introduce multi-granularity residual blocks to process the physiological information feature representation.

[0051]

[0052] Where: H i This represents the processed physiological information features. This represents the physiological information features after processing the (i-1)th multi-granularity residual block.

[0053] Step B5: Establish the connections between physiological information feature representations:

[0054]

[0055] Where: H final For the final physiological information feature representation, g is the number of multi-granularity residual blocks, and H is the number of residual blocks. i The processed physiological information feature representations are shown below. H0 represents the physiological information feature representation output by the first multi-granularity residual block, and H1 represents the physiological information feature representation output by the second multi-granularity residual block. g The physiological information feature representation output by the (g-1)th multi-granularity residual block;

[0056] Step B6: The final physiological information feature representation is used as physiological feature information.

[0057] Numericalizing physiological information facilitates processing and analysis in computer systems. Using binary representation for gender allows the model to understand and utilize gender information. Combining physiological information into an input matrix provides a unified data structure for subsequent feature extraction and processing. Through the feature extraction step in the neural network (step B3), physiological information is abstracted and represented, extracting important features to capture key patterns and changes, making the final physiological feature information more representative. Introducing multi-granularity residual blocks for multi-level feature processing of physiological information helps extract and integrate information from different granularities, enhancing the network's ability to express physiological information. Establishing connections between physiological information feature representations (step B5) adds feature representations of different granularities through residual connections, forming a richer and deeper representation of physiological information. The final physiological information feature representation (step B6) serves as the extracted physiological feature information, representing the model's abstract and comprehensive representation of the physiological state of the elderly.

[0058] In this embodiment of the invention, the drug information screening module is used to screen drug feature information from a multi-source drug database. The specific steps for screening drug feature information are as follows:

[0059] Step 1, Drug Information Processing: Extract drug information from multi-source drug databases and preprocess the extracted drug information, including handling missing values, outlier values, and removing duplicate values.

[0060] Step 2, Drug Feature Extraction: Extract features from the processed drug information, extracting the textual feature information for each drug;

[0061] Step 3, Drug Feature Information Segmentation: Use Chinese word segmentation tools to segment the text features of each drug into word granularities of different lengths; for example, segmenting "this drug is very effective in treating pain" results in "this", "drug", "to", "treat", "pain", "very", "effective";

[0062] Step 4, Extract Granularity Feature Vectors: Extract word granularity feature vectors of different lengths:

[0063] Wg EIGEN =Tanh(W*Y N +B);

[0064] In the formula: Wg EIGEN Y is the word-level feature vector, Tanh is the Tanh activation function, W is the learnable weight matrix, and Y is the word-level feature vector. N Input words of different lengths, where B is the bias vector and N is the total number of word granularities;

[0065] Step 5, Granular Feature Vector Fusion: Word-level feature vectors are fused using a residual connection method. The formula for fusing word-level feature vectors is as follows:

[0066]

[0067] In the formula: E fusion This represents the fusion of word-level feature vectors, where ReLU is the ReLU activation function, F(Wg EIGEN Wg represents the residual mapping of word-granularity feature vectors. EIGEN is the word-granularity feature vector, and N is the total number of word granularities;

[0068] Step 6: Obtain drug feature information: Use the fusion representation of word-level feature vectors as drug feature information.

[0069] The drug information screening module processes, extracts, segments, and fuses data from multi-source drug databases, ultimately generating more expressive and distinguishable drug feature information. This helps improve the understanding of drug semantic information, provides a stronger foundation for subsequent drug analysis, mining, and application, and can improve data quality, enhance feature representation, and fuse multi-level information, providing a powerful tool and support for drug information processing.

[0070] In this embodiment of the invention, the feature information interaction module is used to realize the interaction of elderly physiological feature information and drug feature information. The specific steps for the interaction are as follows:

[0071] Step A1, Feature graph structure establishment: Use the physiological characteristics of the elderly and the drug characteristics as the feature nodes of the feature graph structure, and use the interaction relationship between the physiological characteristics of the elderly and the drug characteristics as the edges of the feature graph structure.

[0072] Step A2, the interaction between elderly physiological characteristics and drug characteristics is reflected by the spatial similarity between feature nodes. The formula for the spatial similarity between feature nodes is:

[0073]

[0074] In the formula: CS is the spatial similarity between feature nodes, X m H is the physiological information input matrix. final This is a fused representation of word-level feature vectors.

[0075] By establishing a feature map structure, it is helpful to grasp the relationship between the physiological characteristics and drug characteristics of the elderly in a global way, forming a more comprehensive and complete interactive view. The calculation of spatial similarity provides a quantitative way to measure the degree of similarity between physiological characteristics and drug characteristics, and can discover common patterns and associations in the physiological characteristics and drug characteristics of the elderly. Understanding the relationship between the physiological characteristics and drug characteristics of the elderly can help provide more personalized and precise medical services. Through the interactive capabilities of the model, it is expected to promote more accurate prediction of the health status of the elderly.

[0076] As a further aspect of the present invention, the drug utilization value analysis module is used to establish a drug utilization value analysis model based on elderly physiological characteristic information and drug characteristic information. The formula for the drug utilization value analysis model is:

[0077] D VALUE =CS×H final +(1-CS)E fusion ;

[0078] In the formula: D VALUEThe results of the drug utilization value analysis are shown, where CS represents the spatial similarity between feature nodes, and H represents the spatial similarity between feature nodes. final For the final physiological information feature representation, E fusion This is a fused representation of word-level feature vectors.

[0079] By comprehensively considering the physiological and drug characteristics of the elderly, the model more comprehensively assesses the utilization value of drugs, which helps to more accurately predict the impact of drugs on elderly patients. The introduction of spatial similarity considers the interrelationships between different characteristics, making the model more flexible and adaptable, and better able to adapt to different situations of elderly patients and drugs. The consideration of the physiological characteristics of elderly patients makes drug assessment more personalized. The model formula clearly expresses the calculation method of comprehensive characteristics, making the model results more interpretable and helping to understand the formation process of drug utilization value.

[0080] This invention, through the construction of a multi-source drug data stream and its storage in a multi-source drug database, systematically manages drug information, reducing drug misuse caused by confusion in drug names, dosages, and formulations, thus improving medication safety and facilitating the rapid and accurate retrieval and querying of specific types of drug information. Furthermore, by integrating information on the physiological and drug characteristics of the elderly, a drug utilization value analysis model is constructed. This model comprehensively considers both physiological and drug characteristics, enabling a comprehensive assessment of the drug's utilization value in the elderly and helping to more accurately predict the drug's impact on elderly patients.

[0081] Example 2

[0082] Figure 2 This diagram illustrates the data storage process for drug data in the data layer according to Embodiment 2 of the present invention. Figure 2 As shown, the steps for storing drug data in this embodiment are as follows: first, drug information is entered; then, the drug information is classified; and the classified drug information is distributed to drug name stream, drug dosage stream, drug dosage form stream, drug instructions for use stream, drug efficacy stream, and drug contraindication stream, respectively. The drug name stream is stored in the drug name database, the drug dosage stream is stored in the drug dosage database, the drug dosage form stream is stored in the drug dosage form database, the drug instructions for use stream is stored in the drug instructions for use database, the drug efficacy stream is stored in the drug efficacy database, and the drug contraindication stream is stored in the drug contraindication database.

[0083] This invention integrates drug information from different sources through input, classification, and storage, constructing a complete database. Classifying and distributing drug information makes it more structured, facilitating subsequent management and analysis. Storing information in a dedicated library helps in the quick and accurate retrieval and querying of specific types of drug information. The naming and storage of the library improves system maintainability and facilitates future system updates and expansions. The classification information in the database contributes to the development and application of applications such as drug utilization value analysis and drug recommendation systems.

[0084] Example 3

[0085] Figure 3 This diagram illustrates the process of constructing a drug utilization value analysis model at the application layer, as provided in Embodiment 3 of the present invention. Figure 3 As shown, the specific steps for processing the spine image in this embodiment are as follows:

[0086] Step C1, Begin.

[0087] Step C2: Analyze and extract physiological characteristic information from the physiological information of the elderly.

[0088] Step C3: Filter out drug characteristic information from the multi-source drug database.

[0089] Step C4 enables the interaction of physiological characteristics information of the elderly and drug characteristics information.

[0090] Step C5: Establish a drug utilization value analysis model based on information on the physiological characteristics of the elderly and the characteristics of the drugs.

[0091] Step C6: Secondary calculation of the results of the drug utilization value analysis model.

[0092] Step C7: Determine whether the deviation between the secondary calculation result and the output result of the drug utilization value analysis model exceeds 5%.

[0093] If the deviation between the secondary calculation result and the output result of the drug utilization value analysis model exceeds 5%, a reminder is issued and the process returns to step C2; if the deviation between the secondary calculation result and the output result of the drug utilization value analysis model does not exceed 5%, then proceed to step C8.

[0094] Step C8, output the result.

[0095] This invention establishes a model by integrating the physiological and drug characteristics of the elderly, enabling a comprehensive assessment of the drug's value in the elderly, including efficacy and potential risks. Secondary calculations and bias assessments verify the model's stability and output reliability, improving its practicality. A bias assessment alert mechanism promptly identifies and corrects potential model problems, enhancing application safety. By analyzing the physiological and drug characteristics of the elderly, the model can provide more personalized and accurate recommendations. The feedback mechanism, by alerting and returning to step C2, helps continuously optimize the model, adapting it to the needs of different patient groups and drug characteristics.

[0096] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0097] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A drug utilization value analysis system based on data flow, comprising a data layer, an application layer, and a user layer, characterized in that, The data layer is used to construct multi-source drug data streams and store them in a multi-source drug database. The application layer is used to construct a drug utilization value analysis model based on the multi-source drug database. The application layer includes a geriatric physiological analysis module, a drug information screening module, a feature information interaction module, a drug utilization value analysis module, a drug use monitoring module, and a reminder module. The drug utilization value analysis module is used to establish a drug utilization value analysis model, and the formula for the drug utilization value analysis model is: D VALUE =CS×H final +(1-CS)E fusion ; In the formula: D VALUE The results of the drug utilization value analysis are shown, where CS represents the spatial similarity between feature nodes, and H represents the spatial similarity between feature nodes. final For the final physiological information feature representation, E fusion This is a fused representation of word-level feature vectors.

2. The drug utilization value analysis system based on data flow according to claim 1, characterized in that, The data layer includes a drug information entry module, a drug information classification module, a drug data distribution module, a multi-source database management module, and a multi-source data storage module; the drug information entry module is used to enter drug information. The drug information classification module is used to classify drug information; the drug data distribution module is used to distribute the classified drug information to the corresponding drug data stream. The multi-source database management module is used to generate a multi-source drug database and store drug information.

3. The drug utilization value analysis system based on data flow according to claim 1, characterized in that, Multi-source drug data streams include drug name streams, drug dosage streams, drug dosage form streams, drug instructions for use streams, drug efficacy streams, and drug contraindications streams; multi-source drug databases include drug name databases, drug dosage databases, drug dosage form databases, drug instructions for use databases, drug efficacy databases, and drug contraindications databases.

4. The drug utilization value analysis system based on data flow according to claim 1, characterized in that, The geriatric physiological analysis module is used to analyze and extract physiological characteristic information from the physiological information of the elderly; The drug information filtering module is used to filter drug characteristic information from a multi-source drug database; The feature information interaction module is used to realize the interaction of elderly physiological feature information and drug feature information; the drug utilization value analysis module is used to establish a drug utilization value analysis model through elderly physiological feature information and drug feature information; the drug use monitoring module is used to recalculate the results of the drug utilization value analysis model; the reminder module is used to issue a reminder when the deviation between the calculation results of the drug use monitoring module and the drug utilization value analysis module exceeds 5%.

5. The drug utilization value analysis system based on data flow according to claim 4, characterized in that, The drug information filtering module is used to filter drug feature information from a multi-source drug database. The specific steps for filtering drug feature information are as follows: Step 1, Drug Information Processing: Extract drug information from multi-source drug databases and preprocess the extracted drug information, including handling missing values, outlier values, and removing duplicate values. Step 2, Drug Feature Extraction: Extract features from the processed drug information, extracting the textual feature information for each drug; Step 3, Drug Feature Information Segmentation: Use Chinese word segmentation tools to segment the text features of each drug into word granularities of different lengths; Step 4, Extract Granularity Feature Vectors: Extract word granularity feature vectors of different lengths: Wg EIGEN = Tanh(W*Y) N +B); In the formula: Wg EIGEN Y is the word-level feature vector, Tanh is the Tanh activation function, W is the learnable weight matrix, and Y is the word-level feature vector. N Input words of different lengths, where B is the bias vector and N is the total number of word granularities; Step 5, Granular Feature Vector Fusion: This involves fusing word granular features using residual connections. The formula for fusing feature vectors at the feature vector level and word-level feature vector level is as follows: In the formula: E fusion This represents the fusion of word-level feature vectors, where ReLU is the ReLU activation function, F(Wg EIGEN Wg represents the residual mapping of word-granularity feature vectors. EIGEN is the word-granularity feature vector, and N is the total number of word granularities; Step 6: Obtain drug feature information: Use the fusion representation of word-level feature vectors as drug feature information.

6. The drug utilization value analysis system based on data flow according to claim 1, characterized in that, The elderly physiological analysis module includes a physiological information collection unit, a physiological characteristic information representation unit, and a physiological characteristic information output unit. The physiological information collection unit is used to collect elderly physiological information, including age, weight, sex, height, liver function characteristics, kidney function characteristics, and blood-brain barrier function characteristics. The physiological feature information representation unit is used to analyze and extract physiological feature information from physiological information; The physiological characteristic information output unit is used to output physiological characteristic information.

Images