Clinical test drug online management platform based on Internet of Things

Abstract

The invention relates to the field of online management platforms, and discloses a clinical test drug online management platform based on the Internet of Things, which comprises an input module, a classification module, a storage module, a data scheduling module and a query module, and is characterized in that the output end of the input module is connected with the input end of the classification module; the output end of the classification module is connected with the input end of the storage module, the output end of the storage module is connected with the input end of the data scheduling module, the output end of the output module of the data scheduling module is connected with the input end of the query module, and the output end of the storage module is connected with the input end of the query module. The platform thoroughly changes the traditional manual recording mode through the deep linkage of the input module and the Internet of Things equipment, and realizes the real-time association of physical drugs-system data-subjects.

Description

Technical Field

[0001] This invention relates to the field of online management platforms, specifically an online management platform for clinical trial drugs based on the Internet of Things. Background Technology

[0002] In the field of pharmaceutical research and development, clinical trials, as the core link in verifying the safety and efficacy of drugs, directly affect the reliability of trial results and the medication safety of subjects due to the standardization of drug management, the accuracy of data, and the traceability of the entire process. They must also strictly adhere to international and domestic regulations such as the Good Clinical Practice (GCP) guidelines. However, traditional clinical trial drug management models are increasingly unable to meet the complex needs of modern clinical trials in terms of technology application, process efficiency, and risk control, mainly exhibiting the following core pain points:

[0003] I. The traditional management model suffers from low efficiency in data collection and correlation.

[0004] Traditional clinical trial drug management relies heavily on manual records and paper ledgers. Drug information (such as production batch number, expiration date, and storage conditions), trial-related data (such as sponsor information and ethics approval number), and subject medication records (such as administration time and dosage) must be entered manually one by one. This is not only time-consuming and labor-intensive but also prone to data errors due to human error. More importantly, it is difficult to achieve real-time correlation between the physical flow of drugs and system data. For example, when drugs are stored, environmental temperature and humidity data must be manually read from sensors and entered into the system. When subjects take the medication, drug labels must be manually verified against subject information. This results in a lag in the "physical drug - system data - subject" correlation chain, failing to meet the real-time and consistency requirements of clinical trials. Furthermore, when drugs involve multiple levels of circulation (such as from the manufacturer to the research center warehouse, and then to the subject), the fragmentation of manual records can easily cause data gaps. Subsequent traceability requires the integration of multiple paper ledgers, which is extremely inefficient.

[0005] II. Lack of Standardization and Intelligent Support for Categorized Management

[0006] Clinical trial drugs need to be categorized across multiple dimensions, including trial phase (Phase I / II / III), dosage form (tablets / injections / cold chain drugs), and dosing regimen (experimental group / control group), to ensure accurate medication use and orderly inventory management. Traditional classification methods rely heavily on manual division based on experience or simple rules, lacking unified data preprocessing and standardized rules. On the one hand, the data to be classified (such as inconsistent drug specifications or ambiguous trial phase labeling) has not undergone systematic cleaning and feature extraction, easily leading to classification bias. On the other hand, classification rules are often fixed in human memory or static documents, making it impossible to adjust them according to the trial protocol (such as adding a "emergency backup drug" category) or optimize them through data feedback. For example, if cold chain drugs and room temperature drugs are mixed on the same shelf due to manual classification errors, it may cause drug spoilage risks; incorrect drug classification for subject groups can directly affect the validity of trial data and even invalidate trial results.

[0007] III. Storage management presents security risks and resource waste.

[0008] Clinical trial drugs have stringent storage requirements (e.g., cold chain drugs need to be maintained at 2-8℃, and special drugs need to be protected from light and moisture), and their statuses need to be differentiated into "in stock awaiting dispatch," "dispatched," and "expired and to be recalled." Traditional storage management models struggle to achieve precise control. On one hand, environmental monitoring relies on regular manual inspections, which cannot monitor warehouse temperature and humidity fluctuations in real time. If sensor alarms occur (e.g., refrigeration equipment malfunctions causing excessive temperature), delayed manual response can easily lead to drug losses. On the other hand, storage media is often limited, relying heavily on local hardware servers for data storage, without tiered management based on data lifecycle—for example, frequently accessed in-stock drug data within the last 30 days is mixed with historical archived data older than 30 days, consuming hardware storage resources and reducing data retrieval speed. Furthermore, traditional storage models lack historical version management; changes to drug information (e.g., adjustments to storage conditions, extensions of expiration dates) only record the latest status, making it impossible to trace historical data. If drug quality issues arise, it is difficult to pinpoint the change point, failing to meet GCP requirements for data traceability.

[0009] IV. Query and data scheduling are difficult to adapt to the needs of multiple scenarios.

[0010] During clinical trials, researchers, sponsors, and regulatory agencies frequently need to query drug data. These queries range from "rapid retrieval of single data items" (such as a subject's medication records) to "multi-data correlation analysis" (such as the flow and usage of a batch of drugs across different research centers). Traditional query methods primarily rely on manual searching through paper ledgers or scattered Excel spreadsheets, which is not only slow but also unable to perform multi-dimensional filtering and statistical analysis. Even when simple database queries are introduced in some scenarios, there is a lack of professional query optimization mechanisms—for example, when querying "all cold chain drug entry records for a certain trial phase," a full table scan is required due to the lack of indexes for fields such as "trial phase" and "drug type," resulting in significant time consumption. Furthermore, when data needs to be retrieved across storage nodes (such as simultaneously querying recent data on local hardware storage and historical data backed up off-site), the traditional model lacks a unified data scheduling mechanism, requiring manual acquisition and integration of data separately, which fails to meet the efficiency and correlation requirements of clinical trials for data queries.

[0011] V. Compliance and traceability are difficult to fully guarantee.

[0012] Regulations such as GCP clearly require that the management of clinical trial drugs must achieve "full-process traceability," including operation logs for data entry, modification, and deletion. Every step in the drug circulation process must record the person involved, the time, and the basis for the record. Under traditional management models, paper logs are easily lost or tampered with, and electronic records often lack access control and version tracking—for example, unauthorized personnel may modify drug expiration date data, and it is impossible to trace the person who made the modification or the content before and after the modification. Compliance verification upon drug entry (such as warnings for drugs less than 3 months from expiration and temperature matching verification for cold chain drugs) relies on manual checking, which is prone to overlooking risk points due to negligence, leading to non-compliant drugs entering the trial process and facing regulatory penalties. At the same time, the traditional model is difficult to connect to external regulatory platforms, requiring manual data compilation and periodic reporting, which cannot achieve real-time synchronization of compliance data and increases the lag in regulatory response.

[0013] VI. The development of Internet of Things (IoT) technology provides an opportunity for management upgrades.

[0014] With the maturity of IoT technology, devices such as RFID tags, smart temperature and humidity sensors, and IoT gateways have achieved low-cost and high-stability application conditions, enabling real-time data interaction between "items, devices, and systems." RFID tags can assign a unique identifier to each drug and support contactless rapid scanning; smart sensors can collect and store environmental temperature and humidity and drug location information in real time, and synchronize it to the cloud via IoT gateways; handheld RFID devices can achieve instant binding of "drug-subject ID" when a subject takes medication. The application of these technologies provides a possibility to solve the pain points of traditional management models such as real-time data association, intelligent classification, and dynamic monitoring. At the same time, the development of database technology (such as SQL parsing and query optimization) and cloud storage technology also provides technical support for efficient querying, hierarchical storage, and data security of clinical trial drugs. Against this backdrop, building an online management platform for clinical trial drugs that integrates IoT technology and information management, integrating core modules such as data entry, classification, storage, data scheduling, and querying, to achieve digital, intelligent, and compliant management of the entire drug lifecycle has become an inevitable trend in the development of modern clinical trials. To this end, we propose an IoT-based online management platform for clinical trial drugs. Summary of the Invention

[0015] To address the shortcomings of existing technologies, this invention provides an online management platform for clinical trial drugs based on the Internet of Things, which solves the aforementioned problems.

[0016] To achieve the above-mentioned objectives, the present invention provides the following technical solution: an online management platform for clinical trial drugs based on the Internet of Things, comprising an input module, a classification module, a storage module, a data scheduling module, and a query module. The output end of the input module is connected to the input end of the classification module, the output end of the classification module is connected to the input end of the storage module, the output end of the storage module is connected to the input end of the data scheduling module, the output end of the data scheduling module is connected to the input end of the query module, and the output end of the storage module is connected to the input end of the query module.

[0017] Preferably, the data entry module is responsible for accurately entering drug information and test-related data into the system and associating it with IoT devices.

[0018] Preferably, the data entry module includes a basic information entry submodule, an inventory information entry submodule, a subject medication information entry submodule, a data verification and cleaning submodule, an entry log and version management submodule, and an interface and integration submodule.

[0019] The basic information entry submodule is responsible for collecting core static information about drugs and experiments;

[0020] The inventory entry submodule connects to IoT devices to record dynamic data on medicines entering the warehouse / research center;

[0021] The subject medication information entry submodule records the entire process data from drug distribution from the warehouse to the subject, and is linked to the execution process of the clinical trial;

[0022] The data verification and cleaning submodule ensures the accuracy and compliance of the entered data;

[0023] The log entry and version management submodule meets the traceability and auditing needs of clinical trials, recording all data entry and change operations;

[0024] The interface and integration sub-module enable the data entry module to work in conjunction with other modules on the platform and external systems.

[0025] Preferably, the storage module includes an input module, a storage terminal, a history module, and an output module. The output terminal of the input module is connected to the input terminal of the storage terminal, the output terminal of the storage terminal is connected to the input terminal of the history module, and the output terminal of the history module is connected to the input terminal of the output module.

[0026] Preferably, the storage terminal includes a hardware storage module and a cloud disk. The hardware storage module is used to store content within the past thirty days, and the cloud disk is used to store data beyond the past thirty days.

[0027] Preferably, the query module includes an SQL parser, a query optimizer, an executor, and a result processing and return module;

[0028] The query optimizer analyzes multiple possible execution paths and selects the optimal solution;

[0029] The SQL parser performs comprehensive parsing and validation of the SQL query statement entered by the user. First, it performs syntax validation to determine if the statement conforms to SQL syntax standards. Then, it conducts semantic analysis to verify whether the tables and fields involved in the query actually exist and whether the data types match.

[0030] The executor is responsible for invoking the database storage engine to perform specific data operations;

[0031] The results processing and return module processes the raw data acquired by the executor.

[0032] Preferably, the classification module includes a classification object input and preprocessing module, a classification rule model module, a classification decision execution module, a classification result verification and correction module, and a classification result output module;

[0033] The object input and preprocessing module receives the raw objects to be classified and completes the "cleaning-standardization-feature extraction" process.

[0034] The classification rule model module has rule-based storage and model storage;

[0035] The classification decision execution module calls the "preprocessed object features" and matches them with the "stored rules / models" to perform classification judgment;

[0036] The classification result verification and correction module verifies the accuracy of the classification results and reduces misclassification.

[0037] The classification result output module outputs the final classification result in the required format and records the classification log.

[0038] Compared with existing technologies, this invention provides an online management platform for clinical trial drugs based on the Internet of Things, which has the following beneficial effects:

[0039] I. Improve data collection and correlation efficiency to ensure data real-time performance and consistency.

[0040] The platform, through deep integration of its data entry module and IoT devices, completely transforms the traditional manual record-keeping model, achieving real-time correlation between "physical drugs - system data - subjects." In the data collection phase, inherent drug information (production batch number, expiration date) can be entered via barcode scanning. Temperature and humidity data upon warehousing are automatically synchronized to the system by intelligent sensors. When subjects administer medication, they instantly bind the "drug label - subject ID" using a handheld RFID device. This not only eliminates the tedious manual entry process but also minimizes data error rates, meeting the core requirements of clinical trials for data accuracy. Simultaneously, during the multi-level drug circulation process (manufacturer → research center → subject), each step is synchronized to the system in real time, forming a complete data chain. Subsequent traceability eliminates the need to integrate scattered ledgers; the entire process data can be obtained simply by querying the system, improving traceability efficiency by over 80% and effectively avoiding data breakpoint issues.

[0041] II. Achieve standardization and intelligentization of classification management to reduce the risk of classification deviation.

[0042] The platform's classification module provides standardized and intelligent support for clinical trial drug classification through a complete workflow design encompassing "preprocessing - rule modeling - decision execution." In the data preprocessing stage, the system automatically cleans invalid information (such as standardizing drug specification descriptions and correcting trial stage annotations) and extracts key features (such as drug dosage forms and trial groups) to ensure consistency of the data to be classified. The classification rule model supports both manually preset explicit rules (such as "cold chain drugs are classified into special storage categories" and "trial group drugs are associated with corresponding subject groups") and dynamic optimization through machine learning models. For example, it can automatically supplement "emergency backup drug" classification rules based on historical classification errors. This intelligent classification mode not only avoids biases caused by reliance on human experience but also allows for rapid response to adjustments in the trial protocol, improving classification accuracy to over 99%. It effectively prevents problems such as mixing cold chain drugs with room temperature drugs and incorrect drug classification for subjects, ensuring the validity of trial data and the safety of drug storage.

[0043] III. Optimize storage management to reduce security risks and resource waste.

[0044] To address the refined needs of clinical trial drug storage, the platform's storage module achieves a dual improvement in safety and efficiency through hierarchical management and real-time monitoring. Regarding environmental monitoring, intelligent temperature and humidity sensors within the warehouse collect data in real time and synchronize it to the system. If the temperature exceeds preset ranges (e.g., cold chain drugs exceeding 8°C), the system immediately triggers an alarm and pushes it to management personnel, reducing response time to the second level. This avoids the risk of drug spoilage caused by delayed manual inspections, reducing drug loss rates by over 60%. At the data storage level, the platform adopts a hierarchical model of "local hardware + cloud disk." Frequently accessed data within the past 30 days is stored on the hardware to ensure fast retrieval; historical data older than 30 days is archived to the cloud disk, reducing hardware resource consumption and lowering storage costs by 40%. Simultaneously, the historical module records all drug information changes, supporting retrospective queries. If quality issues arise, the change point can be quickly located, fully complying with GCP requirements for data traceability.

[0045] IV. Adapt to various query and data scheduling needs, and improve data utilization efficiency.

[0046] The collaborative design of the platform's query module and data scheduling module completely solves the inefficiency problem of traditional query modes. For scenarios involving "rapid retrieval of single data" (such as a subject's medication record), the query module can directly retrieve data from the storage module. Combined with the SQL parser's syntax validation and the query optimizer's index retrieval (such as creating indexes for "subject ID" and "drug batch number"), the query response time is shortened to milliseconds. For scenarios involving "multiple data correlation analysis" (such as the cross-research center flow of a batch of drugs), the data scheduling module can automatically retrieve relevant data from the hardware and cloud disk without manual integration. At the same time, the executor executes queries according to the optimal plan, avoiding full table scans, improving the efficiency of multiple data queries by more than 70%. In addition, the system supports multi-dimensional filtering and statistical analysis (such as counting the number of entries by trial stage and drug type). Researchers and sponsors can quickly obtain the data they need, providing data support for trial progress management and decision-making.

[0047] V. Strengthen compliance and traceability to meet regulatory requirements.

[0048] The platform fully ensures the compliance of clinical trial drug management through end-to-end logging and access control. The log and version management sub-modules of the data entry module record every data entry, modification, and deletion operation in detail, including the operator, time, content, and versions before and after the modification. Only authorized personnel (such as warehouse managers and researchers) can perform the corresponding operations, effectively preventing unauthorized tampering. The compliance verification function automatically checks the expiration date of drugs entering the warehouse (warning for drugs less than 3 months from expiration) and the matching degree of storage conditions, preventing non-compliant drugs from entering the trial process and reducing the risk of regulatory penalties. Meanwhile, the interface and integration sub-modules support connection to external regulatory platforms, enabling real-time synchronization of compliance data without manual compilation and reporting. This completely solves the problem of regulatory response delays and fully meets the "end-to-end traceability" requirements of international and domestic regulations such as GCP and FDA, enhancing the company's compliance image.

[0049] VI. Promote the digital transformation of clinical trial management and improve overall operational efficiency.

[0050] As a comprehensive management tool integrating the Internet of Things (IoT) and information technology, the platform not only optimizes single aspects of drug management but also promotes the digital upgrade of the entire clinical trial management process. By connecting with hospital HIS systems and manufacturer systems, it achieves automatic synchronization of subject information and original drug data, reducing cross-system data entry work. Linked with the trial progress management module, it can automatically remind subjects who have not taken their medication on time, ensuring the trial proceeds as planned. This end-to-end digital management model significantly reduces manual intervention, improving the work efficiency of research center managers by more than 50%. Sponsors can monitor drug circulation and trial progress in real time, shortening the overall clinical trial cycle by 15%-20%, and providing strong support for accelerating pharmaceutical R&D. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the structure of the present invention;

[0052] Figure 2 This is a schematic diagram of the storage module of the present invention;

[0053] Figure 3 This is a schematic diagram of the query module of the present invention;

[0054] Figure 4 This is a schematic diagram of the classification module of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Please see Figure 1-4 An online management platform for clinical trial drugs based on the Internet of Things (IoT) includes an input module, a classification module, a storage module, a data scheduling module, and a query module. The output of the input module is connected to the input of the classification module, the output of the classification module is connected to the input of the storage module, the output of the storage module is connected to the input of the data scheduling module, the output of the data scheduling module is connected to the input of the query module, and the output of the storage module is connected to the input of the query module. During operation, the input module inputs data, the classification module classifies the data, and then the storage module stores the data. When multiple queries are required, the query module retrieves multiple data queries from the storage module through the data scheduling module. When querying a single piece of data, the query module directly queries from the storage module.

[0057] The data entry module is responsible for accurately inputting drug information and trial-related data into the system and linking it with IoT devices (such as smart temperature and humidity sensors and RFID tags) to lay the foundation for subsequent tracking, monitoring, and management. This module must balance data accuracy, compliance (meeting GCP and other clinical trial guidelines), and ease of use.

[0058] Basic Information Input Submodule

[0059] The task of collecting core static information about drugs and trials is the foundation for subsequent management and includes:

[0060] Drug-specific information entry:

[0061] Basic attributes of a drug: name, generic name, brand name, dosage form (tablets / injection, etc.), specifications (dosage / packaging), batch number, expiration date, manufacturer, storage conditions (temperature / humidity requirements), etc.

[0062] IoT Identification Association: Binding a unique RFID tag ID or QR code to a medicine to achieve a one-to-one correspondence between the physical medicine and the system data.

[0063] Clinical trial related information entry:

[0064] Trial project information: Trial number, trial name, sponsor, research institution, ethics approval number, etc.;

[0065] Drug usage information: applicable trial phase (Phase I / II / III), dosing regimen (dosage / frequency), subject grouping (experimental group / control group), etc.

[0066] Design highlights: Reduce manual input errors through drop-down selection and data dictionary (such as preset "dosage form" and "trial stage" options), and support barcode scanning for key fields (such as batch number and expiration date).

[0067] Warehouse entry information input submodule

[0068] Connect to IoT devices to record dynamic data of medicines entering the warehouse / research center, achieving "connection to the network as soon as medicines enter the warehouse":

[0069] Basic information for receiving goods: receiving time, quantity received, person in charge, supplier, tracking number, etc.;

[0070] Automatic entry of IoT sensing data:

[0071] Temperature and humidity data: The ambient temperature and humidity of the medicines are automatically collected when they are put into storage through IoT temperature and humidity sensors in the warehouse (which must meet storage requirements);

[0072] Location information: The initial storage location of medicines is recorded via RFID readers (e.g., "3rd shelf in Cold Storage A Zone");

[0073] Compliance verification: The system automatically verifies the expiration date of incoming medicines (automatic warning if less than 3 months from expiration) and the matching degree of storage conditions (such as whether the ambient temperature of cold chain medicines is within the range of 2-8℃ when they are put into storage).

[0074] Design highlights: Interact with IoT gateways to achieve real-time synchronization of sensor data, avoiding the lag and errors of manually entered environmental data.

[0075] Subject Medication Information Entry Submodule

[0076] Record data throughout the entire process of drug distribution from the warehouse to the subjects, and link it to the execution of clinical trials:

[0077] Distribution information: distribution time, quantity distributed, recipient (investigator), corresponding subject number, dosing cycle, etc.;

[0078] Medication records: actual medication time, dosage, and post-medication response of the subject (to be entered by the researcher or linked to the electronic medical record system);

[0079] IoT tracking: By scanning drug labels and subject IDs (such as wristbands) with handheld RFID devices, the "drug-subject" binding record is achieved, avoiding misdelivery or omission.

[0080] Design highlights: Supports mobile input (e.g., researchers scan a code using a tablet) and is linked with the trial progress management module to automatically remind subjects who have not taken their medication on time.

[0081] Data verification and cleaning submodule

[0082] Ensuring the accuracy and compliance of entered data is a core requirement of GCP standards.

[0083] Format validation: Check the field format (e.g., whether the date format is correct, whether the quantity is positive);

[0084] Logical validation: Verify the correlation between data (e.g., "the quantity of a certain batch of drugs entering the warehouse" cannot be greater than "the total production quantity of this batch");

[0085] Compliance verification: Check against the trial protocol to see if the "dosage" is within the allowable range and if the "subject grouping" is consistent with the trial design;

[0086] Anomaly handling: Data that fails validation is marked as "pending review", prompting the data entry user to make corrections and recording the correction log (retaining the content before and after the correction to meet traceability requirements).

[0087] Log entry and version management submodule

[0088] To meet the traceability and auditing requirements of clinical trials, all data entry and modification operations must be recorded:

[0089] Operation log: Records the person who entered the data, the time of entry, the content entered, and subsequent modification and deletion operations;

[0090] Version management: Retain historical versions of major changes to drug information (such as adjustments to storage conditions or extensions of expiration dates) and support retrospective queries;

[0091] Permission Association: In combination with the platform permission management module, ensure that only authorized personnel (such as warehouse keepers, researchers) can perform input operations, and synchronously record the operator's permission level in the log.

[0092] Interface and Integration Sub-module

[0093] Implement the linkage between the input module and other modules of the platform and external systems:

[0094] Interface with Internet of Things devices: Connect to RFID readers, temperature and humidity sensors, intelligent shelves, etc., and automatically obtain device data;

[0095] Interface with the inventory management module: After entering the inbound / outbound data, synchronize it to the inventory module in real time to update the inventory quantity;

[0096] Interface with external systems: Connect to the hospital HIS system (to obtain subject information), the production enterprise system (to obtain original drug data), and the supervision platform (to report the input data as required).

[0097] The storage module includes an input module, a storage end, a history module, and an output module. The output end of the input module is connected to the input end of the storage end, the output end of the storage end is connected to the input end of the history module, and the output end of the history module is connected to the input end of the output module. The input module stores information in the storage end, the history module queries history and data from the storage end, and the output module is used to transmit the data.

[0098] The storage end includes a hardware storage module and a cloud disk. The hardware storage module is used to store content within 30 days, and the cloud disk is used to store data beyond 30 days.

[0099] The query module includes an SQL parser, a query optimizer, an executor, and a result processing and return module.

[0100] SQL Parser

[0101] Function: Receive the SQL query statement input by the user (such as SELECT * FROM user WHERE age > 18), complete syntax verification (judge whether the SQL conforms to the syntax rules), and semantic analysis (confirm whether the "user table" and "age field" exist).

[0102] Query Optimizer

[0103] Function: Analyze multiple possible execution paths, select the optimal solution (such as "full table scan" or "using index"), reduce database resource consumption, and improve query speed.

[0104] Example: When querying "users with age > 18", if the "age" field has an index, the optimizer will preferentially select to retrieve through the index rather than traversing the entire table.

[0105] Actuator

[0106] Function: According to the execution plan determined by the optimizer, call the database storage engine (such as InnoDB) to perform actual data query operations (such as reading disk data and filtering data).

[0107] Result Processing and Return Module

[0108] Function: Formats (e.g., arranges according to the field order specified in SQL), removes duplicates, performs statistics (e.g., COUNT() calculation) on the raw data obtained by the executor, and finally returns it to the user or application.

[0109] SQL parser

[0110] As the entry point for query processing, its core function is to comprehensively parse and validate the SQL query statement entered by the user (such as SELECT * FROM user WHERE age > 18). First, it performs syntax validation to determine whether the statement conforms to SQL syntax specifications; then, it conducts semantic analysis to verify whether the tables (such as the "user table") and fields (such as the "age field") involved in the query actually exist, and whether the data types match, ensuring that the query command is logically executable.

[0111] Actuator

[0112] Based on the execution plan determined by the query optimizer, it is responsible for invoking the database storage engine (such as InnoDB) to perform specific data operations. This process includes actual execution steps such as reading target data from disk and filtering according to query conditions, and is a key step in translating the optimization plan into actual data retrieval actions.

[0113] Result Processing and Return Module

[0114] The system processes the raw data acquired by the executor to meet the user's query requirements. This includes organizing the data according to the field order specified in the SQL query, removing duplicate records, and performing statistical calculations such as COUNT(). Finally, the processed structured results are returned to the user or application, completing the entire query process.

[0115] The classification module includes a classification object input and preprocessing module, a classification rule model module, a classification decision execution module, a classification result verification and correction module, and a classification result output module.

[0116] Object Input and Preprocessing Module

[0117] Function: Receives raw objects to be classified, completes "cleaning-standardization-feature extraction", and provides qualified "processing objects" for subsequent classification.

[0118] Cleaning: Remove invalid / interference information (such as garbled characters in text or empty values ​​in product data);

[0119] Standardization: Unify the format (e.g., convert text to lowercase, standardize product weight to "grams");

[0120] Feature extraction: Extracting key attributes of an object (such as keywords in text, "category-brand-price" of a product).

[0121] Example: When processing news text to be categorized, first remove advertising characters, then extract keywords related to "politics / sports / entertainment".

[0122] Classification rule model module

[0123] Function: Stores the "judgment criteria" for classification, serving as the "decision dictionary" for the classification module, mainly divided into two categories:

[0124] Rule-based storage: Manually defined, explicit rules (e.g., "products priced ≥ 1000 yuan are classified as 'high-end,' and those priced < 1000 yuan are classified as 'affordable,' and texts containing the keyword 'epidemic' are classified as 'health'");

[0125] Model storage: Classification models trained by machine learning (such as BERT models for text classification and CNN models for image classification, storing model parameters and classification logic).

[0126] Example: In the product classification rule base of an e-commerce platform, a mapping relationship is stored such that "'mobile phone' is classified as '3C digital' and 'dress' is classified as 'women's clothing'".

[0127] Classification Decision Execution Module

[0128] Function: It calls the "preprocessed object features" to match with the "stored rules / models" and performs classification judgment. It is the "core brain" of the module.

[0129] Rule-based execution: Matching one by one according to preset conditions (e.g., if the product's "brand = Apple" and "category = mobile phone", then classify it as "3C digital - Apple mobile phone");

[0130] Model execution: Input the object features into the trained model, and the model calculates the probability and outputs the category (e.g., input text features into the BERT model and output "Sports (probability 92%)").

[0131] Classification result verification and correction module

[0132] Function: Verifies the accuracy of classification results and reduces misclassifications, comprising two core actions:

[0133] Verification: Compare the results with "standard samples" (such as correctly categorized manually) to determine whether they meet expectations (e.g., if "football news" is mistakenly categorized as "entertainment", it is marked as abnormal).

[0134] Correction: Adjustments are made to abnormal results (new rules can be added for rule-based scenarios, and erroneous samples can be sent back to retrain the model for model-based scenarios).

[0135] Example: After a content platform categorizes news, it manually checks 10% of the results. If it finds that "financial news" has been miscategorized, it will add a rule that "news containing the keywords 'stock market / GDP' should be categorized as financial news".

[0136] Classification result output module

[0137] Function: Output the final classification results in the required format and record the classification log (for easy traceability).

[0138] Results output: Output to the front-end display (e.g., displayed in the "Women's Clothing" section of the APP after product categorization) or downstream systems (e.g., synchronized to the data analysis module after data categorization);

[0139] Log storage: Records "Category Object ID - Category Time - Category - Category Basis" (e.g., "News ID123-20240520-Sports-Contains the keyword 'World Cup'").

[0140] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An online management platform for clinical trial drugs based on the Internet of Things, characterized in that, It includes an input module, a classification module, a storage module, a data scheduling module, and a query module. The output of the input module is connected to the input of the classification module, the output of the classification module is connected to the input of the storage module, the output of the storage module is connected to the input of the data scheduling module, the output of the data scheduling module is connected to the input of the query module, and the output of the storage module is connected to the input of the query module. 2.The online management platform for clinical trial drugs based on the Internet of Things according to claim 1, characterized in that: The data entry module is responsible for accurately entering drug information and test-related data into the system and connecting it with IoT devices. 3.The online management platform for clinical trial drugs based on the Internet of Things according to claim 1, characterized in that: The data entry module includes a basic information entry submodule, an inventory information entry submodule, a subject medication information entry submodule, a data verification and cleaning submodule, an entry log and version management submodule, and an interface and integration submodule. The basic information entry submodule is responsible for collecting core static information about drugs and experiments; The inventory entry submodule connects to IoT devices to record dynamic data on medicines entering the warehouse / research center; The subject medication information entry submodule records the entire process data from drug distribution from the warehouse to the subject, and is linked to the execution process of the clinical trial; The data verification and cleaning submodule ensures the accuracy and compliance of the entered data; The log entry and version management submodule meets the traceability and auditing needs of clinical trials, recording all data entry and change operations; The interface and integration sub-module enable the data entry module to work in conjunction with other modules on the platform and external systems. 4.The online management platform for clinical trial drugs based on the Internet of Things according to claim 1, characterized in that: The storage module includes an input module, a storage terminal, a history module, and an output module. The output terminal of the input module is connected to the input terminal of the storage terminal, the output terminal of the storage terminal is connected to the input terminal of the history module, and the output terminal of the history module is connected to the input terminal of the output module.

5. The online management platform for clinical trial drugs based on the Internet of Things according to claim 4, characterized in that: The storage end includes a hardware storage module and a cloud disk. The hardware storage module is used to store content within the last 30 days, and the cloud disk is used to store data beyond the last 30 days. 6.The online management platform for clinical trial drugs based on the Internet of Things according to claim 1, characterized in that: The query module includes an SQL parser, a query optimizer, an executor, and a result processing and return module; The query optimizer analyzes multiple possible execution paths and selects the optimal solution; The SQL parser performs comprehensive parsing and validation of the SQL query statement entered by the user. First, it performs syntax validation to determine if the statement conforms to SQL syntax standards. Then, it conducts semantic analysis to verify whether the tables and fields involved in the query actually exist and whether the data types match. The executor is responsible for invoking the database storage engine to perform specific data operations; The results processing and return module processes the raw data acquired by the executor. 7.The online management platform for clinical trial drugs based on the Internet of Things according to claim 1, characterized in that: The classification module includes a classification object input and preprocessing module, a classification rule model module, a classification decision execution module, a classification result verification and correction module, and a classification result output module; The object input and preprocessing module receives the raw objects to be classified and completes "cleaning-standardization-feature extraction"; The classification rule model module has rule-based storage and model storage; The classification decision execution module calls the "preprocessed object features" to match with the "stored rules / models" and performs classification judgment; The classification result verification and correction module verifies the accuracy of the classification results and reduces misclassification. The classification result output module outputs the final classification result in a required format and records a classification log.

Images