Intelligent sample whole-process management system and method, storage medium and electronic equipment

Abstract

This invention discloses an intelligent sample end-to-end management system, method, storage medium, and electronic device. The intelligent sample dynamic management system incorporates a resource dynamic configuration module in its data layer. It uses structured standard operating procedures to drive resource demand analysis, extracts a reagent and consumable list, recommends optimal batches based on inventory, and dynamically reserves equipment time slots. A priority-based preemptive resource allocation algorithm is employed, comprehensively considering sample timeliness, rarity, and project strategic importance for resource allocation. Real-time monitoring of inventory status enables dynamic early warning and replenishment. This invention also provides corresponding methods and storage media. This invention achieves deep coupling between standard operating procedures and resources, priority resource assurance, and a closed-loop quality control throughout the entire process, solving the problems of process-resource disconnect and insufficient critical resource assurance in existing technologies, thus ensuring the high quality of biological samples throughout their entire lifecycle.

Description

Technical Field

[0001] This invention relates to the whole-process management of biological samples, and more specifically, to an intelligent whole-process management system, method, storage medium and electronic device for samples. Background Technology

[0002] Biological sample resources (such as blood, tissues, nucleic acids, and cells) are the core foundation for life science research, precision medicine, and new drug development. The quality of biological samples directly determines the reliability, reproducibility, and success of translational applications of research results. As biomedical research develops towards high throughput and high precision, higher quality requirements are being placed on the entire process management of biological samples, from collection, processing, and storage to use.

[0003] Currently, traditional methods of managing biological samples rely heavily on manual operations and decentralized information systems, which have the following technical drawbacks: 1. Lack of priority and dynamism in resource allocation: Existing systems typically employ first-come, first-served or simple reservation mechanisms, failing to intelligently schedule resources based on factors such as sample timeliness, sample rarity, and project strategic importance. When high-value or time-sensitive sample tasks conflict with routine tasks, the lack of effective preemption and guarantee mechanisms may lead to a decline in the quality or invalidation of critical samples.

[0004] 2. Disconnect between processes and resources: Standard Operating Procedures (SOPs) are usually in text form and cannot be automatically parsed and executed by the system. The allocation of resources such as reagents, consumables, and equipment required for experiments is separated from the SOPs, relying on human experience and judgment. In scenarios requiring large-scale, distributed processing, such as large-scale multi-center clinical trials, large cohorts, and public health emergencies, resource mismatches, omissions, or conflicts are prone to occur, affecting the consistency and timeliness of sample processing.

[0005] 3. Lagging and reactive inventory management: The management of consumables and reagents relies heavily on manual inventory checks or static early warnings with low thresholds, failing to respond in real time to fluctuations in consumption and changes in task requirements. There is a lack of proactive identification and priority use mechanisms for consumables nearing their expiration date, and expired consumables pose a risk of misuse, directly threatening sample quality.

[0006] 4. Lack of a complete quality loop: Existing systems primarily monitor sample processing through post-processing recording, lacking real-time data acquisition, quality assessment, and feedback control capabilities. Quality changes at each stage of sample collection, transportation, processing, and storage are difficult to trace. Once quality issues arise, it is difficult to pinpoint the cause and implement preventative interventions.

[0007] In summary, existing technologies have significant shortcomings in achieving standardized, intelligent, and traceable management of the entire biological sample process, making it difficult to meet the stringent requirements of consistency, safety, and reproducibility for high-quality biological samples. Therefore, there is an urgent need to build a comprehensive management system that deeply integrates standard operating procedure execution, resource allocation, equipment collaboration, and quality monitoring to ensure the high quality of biological samples throughout their entire lifecycle. Summary of the Invention

[0008] To address the problems of existing technologies, this invention provides an intelligent sample end-to-end management system, method, storage medium, and electronic device.

[0009] An intelligent sample dynamic management system, applied to an intelligent sample end-to-end management system, includes a resource dynamic configuration module; the resource dynamic configuration module includes the following sub-modules: (1) The resource requirement parsing module is configured to perform resource requirement parsing, including the following steps: Step A: When the administrator configures the standard operating procedure for the project in the intelligent sample full-process management system, the resource requirements associated with each atomic step of the standard operating procedure are parsed, and the required reagent or consumable list is extracted based on the resource requirements. Step B: Combining the database of the intelligent sample end-to-end management system, the recommended batches of reagents or consumables are obtained through the algorithm configuration module; Step C: Based on the estimated single-step time and total process duration of the standard operating procedure, and combined with the corresponding equipment usage time slot, the device status is synchronized in real time through the IoT sensing and automated execution layer of the intelligent sample full-process management system to dynamically adjust the availability of the reservation. After steps D, B, and C are completed, the system locks the consumable inventory and schedules the corresponding time slot in the equipment calendar; the locking information is synchronized to the intelligent sample end-to-end management system in real time. (2) The algorithm configuration module is configured as follows: A priority-based preemptive resource allocation algorithm is used to calculate the overall priority score, as shown in the following formula: P=w1·T urgency +w2·V sample +w3·S level Where T urgency V is the sample timeliness factor. sample S is the sample rarity factor. level The project strategic level factor is P, which is the comprehensive priority score, with a normalized value range of 0 to 1. w1, w2, and w3 are the weights, satisfying ∑w i =1; The priority of obtaining reagents, consumables or equipment for the project is determined by the ranking of the comprehensive priority score, and the recommended batch of reagents or consumables is obtained. (3) The dynamic inventory early warning and replenishment module is configured as follows: The system dynamically monitors the inventory status of consumables in the inventory database of the intelligent sample full-process management system, updates the consumables status to the administrator in real time, and makes predictions, alarms, and notifications to the administrator based on the consumables status.

[0010] Preferably, in the algorithm configuration module: This reflects the urgency of the maximum allowable time window from sample collection to processing completion; the calculation formula is: in, For the current time, Sampling time, This is the maximum processing time limit specified in the standard operating procedures; V sample The samples are graded and quantified based on their scarcity, non-renewability, and scientific and clinical value. S level Preset by the administrator in the intelligent sample end-to-end management system; When a high-priority task is submitted, the system will check whether the resources it requires are occupied by low-priority tasks; if so, the system can automatically negotiate to release them or trigger an alert to notify the administrator to make a manual decision.

[0011] Preferably, the intelligent sample dynamic management system includes the following modules: Dynamic resource configuration module: Dynamically configures the required resources, coupling the execution of standard operating procedures with resource requirements; Sample quality assessment module: Monitors samples and sample containers in real time and collects background data, and conducts assessments or issues warnings based on changes in sample quality; Intelligent Project and Personnel Management Module: Establishes digital profiles for projects and personnel; Structured Standard Operating Procedures Management Module: Used to convert text-based standard operating procedures into structured data objects that the system can parse and execute; Process quality assessment module: Real-time monitoring of personnel operations and sample processing, collection of background data, and assessment or early warning based on comparison with atomized standard operating procedures; Preferably, the sample quality assessment module includes the following sub-modules: (1) The sample tracking module is configured to track images captured in real time by the AR terminal through an integrated AI recognition engine. The image tracking specifically includes: A. The sample receiving station with integrated RFID reader can track the tags on the sample containers inside the sample box; B. Place the sample container in a pre-defined fixed shooting area and perform image tracking on the sample collected in the sample container; (2) The quality assessment module is configured as follows: a. Retrieve the matching sample list from the intelligent sample full-process management system and verify the retrieved sample list information against the sample container label information; b. Compare and verify the collected sample images with the sample container labels; c. Evaluate the quality of the collected sample images.

[0012] An intelligent sample end-to-end management system, integrating the aforementioned intelligent sample dynamic management system, further includes: The multimodal hybrid terminal is configured as the main interactive interface for personnel training and assessment and actual operation. Through the multimodal hybrid terminal, personnel can complete the collection and processing of samples. The IoT sensing and automated execution layer is configured to achieve full-dimensional sensing and automated data collection of samples, environment, and devices during the actual operation phase; and to provide real device status simulation data for virtual-real fusion assessment during the training phase. The network and communication layer is configured to enable communication connections between field devices and cloud / local servers; The central decision-making platform is configured to integrate a data processing module, a behavior perception and intelligent evaluation module, a practical intelligent analysis module, and a knowledge graph and continuous learning module to realize functions such as personnel training, sample tracking, data processing, behavior evaluation, model learning and updating. The remote collaboration and control terminal is configured to enable authorized personnel to remotely manage and collaborate on the entire process of training and practical operation on a multimodal hybrid terminal. Specifically, it includes scenario and project configuration, sample task initiation, training system construction, real-time monitoring, dynamic event injection, proactive intervention and guidance, training assessment and authorization management, as well as review and analysis functions. The digital twin platform layer is configured to build and store digital twin device, sample, and spatial information.

[0013] An intelligent sample dynamic management method, based on the intelligent sample dynamic management system, realizes dynamic resource allocation, including: Step 1: When the administrator configures the standard operating procedure for a project in the intelligent sample full-process management system, the resource requirements associated with each atomic step of the standard operating procedure are parsed, and the required reagent or consumable list is extracted based on the resource requirements. Step 2: Combine the database of the intelligent sample end-to-end management system to configure and obtain recommended batches of reagents or consumables through algorithms; Step 3: Based on the estimated single-step time and total process duration of the standard operating procedure, and combined with the corresponding equipment usage time slot, the availability of the reservation is dynamically adjusted by synchronizing the equipment status in real time through the IoT sensing and automated execution layer of the intelligent sample full-process management system. After steps 4, 2, and 3 are completed, the system locks the consumable inventory and schedules the corresponding time slot in the equipment calendar; the locking information is synchronized to the intelligent sample end-to-end management system in real time. A priority-based preemptive resource allocation algorithm is used to calculate the overall priority score, as shown in the following formula: P=w1·T urgency +w2·V sample +w3·S level Where T urgency V is the sample timeliness factor. sample S is the sample rarity factor. level The project strategic level factor is P, which is the comprehensive priority score, with a normalized value range of 0 to 1. w1, w2, and w3 are the weights, satisfying ∑w i =1; The priority of obtaining reagents, consumables, or equipment for the project is determined by the ranking of the comprehensive priority score, and the recommended batch of reagents or consumables is obtained.

[0014] An intelligent sample end-to-end management method, implemented based on the aforementioned intelligent sample end-to-end management system, includes the following steps: Step 1: Establish the project configuration and training system; Step 2: Conduct personnel training and certification authorization; Step 3: Start the sample task; Step 4: Sample collection and initial quality screening; Step 5: Process the samples; Step 6: Process the sample data; Step 7: Perform full lifecycle tracing of the sample and generate a full lifecycle report of the sample.

[0015] The present invention also provides a computer-readable storage medium having a computer program stored thereon for the intelligent sample end-to-end management method described above.

[0016] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the intelligent sample full-process management method described above.

[0017] The present invention also provides a multimodal hybrid terminal, including a general mobile smart terminal and an AR glasses terminal that is communicatively connected to the general mobile smart terminal.

[0018] Preferably, the general-purpose mobile intelligent terminal is configured as follows: In response to the user's first operation, a pure virtual 3D simulation mode is run, rendering and displaying the virtual laboratory 3D scene issued by the service layer on the display screen of the general mobile smart terminal in full screen, and receiving the user's touch interaction command on the 3D scene to execute the theoretical learning and process retraining in the training phase. In response to the user's second operation, the mobile augmented reality (AR) mode is run, the camera of the general mobile smart terminal is called to collect real-time images of the real environment, and virtual guidance information is superimposed on the real-time images according to the atomic operation steps to generate a virtual-real fusion AR screen. The user's operation is verified in real time to ensure that it conforms to the preset standard operating procedure (SOP) in order to perform a practical simulation assessment before being put on the job. The AR glasses terminal is configured to be worn on the user's head during the practical operation phase, including: The display and projection unit, including the optical waveguide display module and the microprojector, is used to overlay virtual guidance information onto the user's real field of vision in a perspective manner, providing immersive operation guidance. The sensing and positioning unit includes: The spatial positioning and mapping SLAM module includes a set of binocular cameras and a depth sensor, which is used to build a 3D map of the operating environment in real time and realize the spatial registration and tracking of virtual information in the 3D map. The environmental perception module is used to collect video streams, ambient light information, and voice commands from the operation site. The computing and communication unit includes: An edge computing chipset is used to locally process the data collected by the sensing and positioning unit, run AI models, and render AR interfaces. The communication module is used to interact with the general mobile smart terminal and the system service layer.

[0019] Preferably, the pure virtual 3D simulation mode supports offline operation; When the general-purpose mobile smart terminal runs the pure virtual 3D simulation mode, it is further configured as follows: The virtual laboratory 3D scene package is stored and loaded locally, and the rendering of the 3D scene and the response to the touch interaction commands are completed independently without a network connection.

[0020] Preferably, the SLAM module employs an improved algorithm based on ORB-SLAM3, specifically including: The feature extraction and matching unit is used to extract ORB feature points from the images captured by the binocular camera and perform feature matching with the previous frame image. The pose estimation and optimization unit is used to estimate the current pose of the AR glasses terminal based on the feature matching results and the depth data obtained by the depth sensor, and to correct the current pose through local map optimization and loop closure detection. The map persistent storage unit is used to store the constructed 3D map and the corresponding ORB feature point cloud data locally, and to quickly load and match the 3D map in response to the restart or relocation request of the AR glasses terminal, so as to achieve rapid relocation in dynamic environments.

[0021] Preferably, the environment sensing module includes: A high-definition camera is used to capture video streams of the operation site, and the video streams are used for behavior perception, sample quality identification, or operation compliance monitoring. An ambient light sensor is used to collect ambient light intensity, and the edge computing chipset dynamically adjusts the display brightness of the optical waveguide display module according to the ambient light intensity; A microphone array is used to capture the user's voice commands and support voice interaction.

[0022] Preferably, the sensing and positioning unit further includes: An eye-tracking system, including an infrared camera and an infrared sensor, is used to track the position of the user's gaze point and determine whether the gaze point is located within a preset key operation object area, generating attention assessment data for behavior evaluation.

[0023] Preferably, the edge computing chipset integrates a CPU, a GPU, and an AI acceleration core NPU; The edge computing chipset is specifically configured as follows: The NPU runs a lightweight AI model to perform real-time inference on the video stream captured by the high-definition camera, perform gesture recognition, object detection or user behavior analysis, and generate recognition results. The GPU renders and generates virtual guidance information in the AR interface based on the recognition results and the spatial location information provided by the SLAM module. The CPU uploads the recognition results and keyframe data from the AR interface rendering process to the system service layer via the communication module.

[0024] Preferably, the communication module of the AR glasses terminal supports 5G, Wi-Fi 6 and Bluetooth protocols; The AR glasses terminal uses the communication module to share its first-person view and operating interface with the control center screen in real time, and receives remote collaboration instructions from the system service layer.

[0025] Preferably, when the general-purpose mobile smart terminal is running the mobile augmented reality (AR) mode, it further includes: The step decomposition unit is used to decompose the target operation process into the atomic operation steps, and to define the corresponding operation object, operation sequence and parameter range for each atomic operation step. The virtual-real registration unit is used to determine the overlay position of the virtual guidance information in the real-time image based on the spatial registration results provided by the SLAM module. The compliance verification unit is used to determine whether the operation is compliant by comparing the user's actual operation trajectory, operation object and input parameters with the preset conditions of the atomic operation steps, and to generate instant error correction feedback information when it is non-compliant. The error correction feedback information is provided through visual prompts via the display and projection unit of the AR glasses terminal, or through voice broadcasts via the voice module of the AR glasses terminal.

[0026] Preferably, the AR glasses terminal further includes: The voice module integrates a speaker and a microphone. The speaker is used to play operation prompts, alarm information, and training guidance voice messages, and the microphone is used to receive the voice commands.

[0027] Preferably, the universal mobile smart terminal is a smartphone or a tablet computer.

[0028] The present invention also provides an Internet of Things (IoT) sensing and automation execution layer, comprising a sample identification and tracking unit, an environmental monitoring unit, an intelligent experimental equipment unit, and an IoT gateway.

[0029] Preferably, the sample identification and tracking unit includes an RFID tag attached to the sample container and a fixed RFID reader deployed at key operation nodes for non-contact identification and location tracking of the sample container. The environmental monitoring unit includes a variety of environmental sensors deployed within the experimental area for real-time collection of environmental parameters; The intelligent experimental equipment unit includes multiple experimental instruments with data communication interfaces, which are used to automatically collect and upload their respective equipment operating parameters. The sample identification and tracking unit, the environmental monitoring unit, and the intelligent experimental equipment unit are respectively connected to the Internet of Things gateway via wired or wireless means; The IoT gateway is configured with a heterogeneous network adaptation module, which is used to aggregate the data uploaded by each unit and send the data to the system service layer through a remote communication protocol. The system service layer is configured to determine anomalies in the data based on a preset threshold, and generate an alarm command when an anomaly is determined. The alarm command is then sent to the corresponding execution unit through the IoT gateway.

[0030] Preferably, the RFID tag is a low-temperature resistant UHF RFID tag, which is attached to the outer wall of the sample container; The sample containers include vacuum blood collection tubes, urine cups, tissue boxes, cryopreservation tubes, or cryopreservation boxes.

[0031] Preferably, the sample identification and tracking unit further includes: Handheld RFID reader / writer, integrated into mobile smart terminals or AR glasses terminals; The handheld RFID reader supports a close-range precision reading mode, with a reading distance configured to not exceed 10 centimeters. It is used to perform point-to-point confirmation of the RFID tags to verify whether the sample being handled by the operator is correct.

[0032] Preferably, the fixed RFID reader is deployed at the sample receiving station, next to the centrifuge, inside the biosafety cabinet, inside the ultra-low temperature freezer, or at the inlet of the liquid nitrogen container; The fixed RFID reader is used to automatically read the RFID tags in batches when the samples pass through the key operation nodes, so as to update the location information of the sample containers in the system service layer.

[0033] Preferably, the environmental monitoring unit includes the following environmental sensors: Temperature and humidity sensors are deployed in laboratory environments, cold storage facilities, or refrigerators to collect air temperature and relative humidity. A liquid level sensor, deployed inside a liquid nitrogen container, is used to collect the liquid nitrogen level. Door magnetic sensors are deployed on the doors of refrigerators or cold storage facilities to detect the open or closed status of the doors.

[0034] Preferably, the IoT gateway has a built-in narrowband IoT (NB-IoT) communication module; The environmental sensor is connected to the IoT gateway through the NB-IoT communication module and uploads the environmental parameters in a low-frequency, small-data-packet manner to achieve low-power wide-area coverage.

[0035] Preferably, the experimental instruments in the intelligent experimental equipment unit include an intelligent pipette, a low-temperature high-speed centrifuge, or a fully automated blood component separation system; The experimental instrument is connected to the IoT gateway via an RS-232, USB, or Ethernet interface. The operating parameters of the device include at least the operating time, operating temperature, rotation speed, and processing volume.

[0036] Preferably, the IoT gateway includes: The protocol conversion engine is used to convert the different fieldbus protocols or short-range wireless protocols used by the sample identification and tracking unit, the environmental monitoring unit, and the intelligent experimental equipment unit into standardized Internet protocol data packets. The edge computing module is used to perform local preprocessing and filtering of the data packets, and upload the filtered valid data to the system service layer.

[0037] Preferably, the system service layer includes: A digital twin model library stores digital twin models corresponding to each experimental instrument in the intelligent experimental equipment unit; During the training phase, the system service layer calls the digital twin model to simulate the operation feedback and parameter changes of the experimental instrument, and sends the simulation data to the general mobile smart terminal or the AR glasses terminal, so that trainees can have an interactive experience consistent with operating real equipment.

[0038] Preferably, the system service layer further includes: The abnormal scenario simulation module is used to send simulated abnormal environmental data to the environmental monitoring unit during the training phase to trigger alarm commands from the system service layer, thereby assessing the emergency response capabilities of the trainees.

[0039] Preferably, the system service layer further includes: The multi-source data fusion analysis module is used to spatiotemporally correlate the sample location information provided by the sample identification and tracking unit, the environmental parameters provided by the environmental monitoring unit, and the equipment operation parameters provided by the intelligent experimental equipment unit to generate a complete operation traceability chain for a single sample.

[0040] Preferably, the alarm commands include audible and visual alarm commands and remote notification commands; The audible and visual alarm command is sent to the audible and visual alarm devices deployed in the experimental area through the Internet of Things gateway; The remote notification command is sent to the designated mobile terminal or control center screen through the message push interface of the system service layer.

[0041] This invention also provides a network and communication layer. Applied to training and practical scenarios, it includes: A 5G industrial gateway / customer premises equipment (CPE) integrates 5G modem, Wi-Fi router and industrial gateway functions; Preferably, the 5G industrial gateway / CPE is communicatively connected to the field device layer, the platform layer, and the cloud resources, respectively, to enable full-level data interoperability between the field device layer, the platform layer, and the cloud resources; The field device layer includes AR glasses, general mobile devices, RFID readers, IoT sensors, and smart devices; The platform layer includes a central decision-making server and a local intelligent sample full-process management system server. The cloud resources include a cloud-based fault knowledge base, an AI model training platform, and a teaching management server.

[0042] Preferably, the 5G industrial gateway / CPE transmits the video stream of the AR glasses, the sensor data of the IoT sensors, training operation data, and practical guidance instructions through a communication method with high bandwidth, low latency, and high connection density.

[0043] Preferably, the 5G industrial gateway / CPE is communicatively connected to the cloud-based fault knowledge base and the teaching management server, for the following purposes: During the training phase, remote teaching and resource distribution are supported; During the practical phase, remote diagnostics and emergency guidance are supported.

[0044] Preferably, the 5G industrial gateway / CPE accesses the 5G network through the 5G modem to achieve communication with the cloud resources.

[0045] Preferably, the 5G industrial gateway / CPE establishes a wireless communication connection with the AR glasses and the general mobile device through the Wi-Fi router.

[0046] Preferably, the 5G industrial gateway / CPE establishes communication connections with the RFID reader, the IoT sensor, and the smart device through the industrial gateway function.

[0047] Preferably, the 5G industrial gateway / CPE is also communicatively connected to the central decision server and the local intelligent sample full-process management system server to enable data interoperability between the field equipment layer and the platform layer.

[0048] Preferably, the 5G industrial gateway / CPE integrates the functions of the 5G modem, the Wi-Fi router, and the industrial gateway into one unit.

[0049] The present invention also provides a central decision-making platform, comprising: Data processing module, pre-job intelligent training engine, practical intelligent analysis module, and knowledge graph and continuous learning module; Preferably, the data processing module is used to preprocess multiple audio, video, sensor timing data, touch screen operation data, training operation data and practical operation data from AR terminals and general mobile devices to provide standardized data for training evaluation and practical analysis. The pre-job intelligent training engine is responsible for the logical processing of pre-job training business, including adaptive virtual scene generation, multimodal training scheduling and allocation, and multimodal behavior perception and intelligent evaluation. The practical intelligent analysis module is used to integrate and analyze multi-dimensional data from the AR terminal during the practical stage, so as to realize sample quality identification, operation compliance monitoring and operation status judgment. The knowledge graph and continuous learning module is used to construct a full-process knowledge graph based on historical data and learn from newly generated data to achieve self-iterative upgrades of the system.

[0050] Preferably, the adaptive virtual scene generation automatically generates a 3D virtual laboratory scene based on the pre-set spatial layout units, digital twin resource units, and structured standard operating procedures of the intelligent sample full-process management system; the adaptive virtual scene generation includes: Template matching based on SOP semantics extracts keywords from standard operating procedures, calculates the cosine similarity between the keywords and the description vectors of templates for each scenario, and selects the template with the highest similarity as the basis. Automatic layout based on bounding box collision detection allocates equipment resources to predefined functional areas of the virtual laboratory. Core equipment is placed at template anchor points, and subsequent equipment uses gradient descent to find the optimal collision-free placement position. Consumables are placed in easily accessible locations according to usage order and frequency.

[0051] Preferably, the multimodal training scheduling and allocation allocates and schedules training modes and content based on the trainees' training stage, learning progress, and current environment; the scheduling rules include: In the first phase, new recruits or those learning theory are assigned to a pure virtual 3D mode, which generates a complete virtual laboratory panoramic scene for trainees to freely explore via touch screen interaction on general mobile devices, and supports offline operation. The second stage of pre-job practical assessment involves being assigned to the AR virtual-real fusion mode, where virtual operation instructions and verification information are overlaid on real equipment and consumables in a real laboratory using an AR terminal. For the third stage, regular refresher training or process updates, it is recommended to use the pure virtual 3D mode for review and practice, or choose the AR mode for targeted reinforcement.

[0052] Preferably, the multimodal behavior perception and intelligent assessment collects the operational data of trainees in the training mode, and uses a preset rule engine and machine learning model to quantitatively score the fidelity, accuracy, standardization and efficiency of the operation process in multiple dimensions, and generates a training assessment report to be stored in the personnel skill file; the assessment results serve as the basis for personnel practical authorization.

[0053] Preferably, the practical intelligent analysis module includes: The multimodal operation tracking submodule integrates and analyzes the operator's gaze focus, gestures, voice command sequences, and decision response time to construct an attention-action timeline and determine the current operation step and state. The evaluation and identification submodule runs a deep learning model to analyze sample images captured by the AR camera in real time, enabling sample quality identification, operational compliance identification, and result verification. The analysis instruction submodule makes decisions based on standard operating procedures, personnel operation tracking results, and evaluation and identification results. When normal, it generates the next AR guidance instruction; when abnormal, it generates the corresponding alarm and intervention instructions.

[0054] Preferably, the sample quality identification includes hemolysis identification, lipemia identification, clot identification, and label ambiguity identification; the operation compliance identification includes glove wearing identification, mask wearing identification, lab coat wearing identification, operation inside a biosafety cabinet identification, and standardized sample addition identification; the result verification includes post-centrifugation stratification clarity verification and liquid level height compliance verification.

[0055] Preferably, the preprocessing includes decoding, cleaning, and feature extraction.

[0056] This invention also provides a remote collaboration and control terminal, which is a web-based cross-platform application that runs on computers, tablets, mobile phones, and control center large screens, including: The system includes a scenario and project configuration module, a real-time monitoring module, a dynamic event injection module, a proactive intervention and guidance module, a training assessment and authorization management module, and a review and analysis module. The scenario and project configuration module is used to configure SOP processes, associate resources, set quality and alarm thresholds for different projects, and create exclusive training projects for different projects, associate electronic SOPs and training resources, and set training assessment standards and quantitative indicators. The real-time monitoring module is used to view the training or practical shared perspective and system status of AR terminals or general mobile devices in real time, and to keep track of the training progress and the current operation status of the operators. The dynamic event injection module is used to remotely simulate equipment failure, sample anomaly, and environmental anomaly scenarios during training or drills. The active intervention and guidance module is used to remotely adjust air conditioning, ventilation and lighting facilities when the system alarms during the practical operation phase, and to provide real-time guidance through AR interface annotation or two-way voice and video calls during the training and practical operation phase. The training assessment and authorization management module is used to remotely view the training records and assessment results of trainees, authorize online practical operations for those who pass the assessment, and adjust personalized training paths based on the assessment results. The review and analysis module is used to replay historical training operation videos, practical operation videos, and problem reviews, and to incorporate typical cases into the training resource library.

[0057] Preferably, the system alarms include temperature exceedance alarms and operational violation alarms.

[0058] Preferably, the dynamic event injection module remotely simulates equipment failure, sample anomaly, and environmental anomaly scenarios during training or drills to test the operator's responsiveness and emergency response capabilities.

[0059] Preferably, the remote collaboration and control terminal is accessed and operated by authorized professionals, including administrators, project leaders, and trainers.

[0060] The present invention also provides a digital twin platform layer, comprising: a spatial layout unit, a digital twin resource unit, and a digital twin archive unit; The spatial layout unit includes a basic laboratory function template and a scene template library. The basic laboratory function template is a pre-set basic laboratory layout template based on the characteristics of experimental functions. The scene template library is generated based on the basic laboratory function template, and each scene template defines the division of core functional areas and their relative positional relationships. The digital twin resource unit includes a device digital twin model and a consumable digital twin model; the model supports multi-level details and is a structured, searchable, and renderable model; when the model is applied to the training module, it only contains static modeling data and interaction logic definitions, and does not connect to real devices, collect real-time status, or perform physical control. The digital twin archive unit creates a unique digital twin record for each entity sample, using the sample's unique identifier as the primary key, and aggregates the data generated throughout its entire lifecycle. The digital twin record uses blockchain hash chain technology to ensure that the data is tamper-proof, and each record contains the hash value of the previous record, forming an audit trail.

[0061] Preferably, the basic laboratory function template includes a sample receiving area, a centrifugation area, a biosafety cabinet operation area, an automatic cryopreservation tube cap opening and closing area, a cryopreservation tube bottom QR code scanning area, a 2℃-8℃ low-temperature temporary storage area, a -80℃ ultra-low temperature storage area, and a -196℃ deep low-temperature storage area.

[0062] Preferably, the scene template includes: Scene template A is a biosafety cabinet operation scene. Its core area is a large biosafety cabinet occupying the center of the screen, and the auxiliary areas are workbenches on the left and right sides for placing samples and consumables. Scene template B is a centrifugation scene in the centrifugation area. Its core area is a floor-standing low-temperature high-speed centrifuge occupying the center of the screen, and the auxiliary areas are a workbench next to the centrifuge where specimen tubes are placed and a balance is placed.

[0063] Preferably, the data aggregated by the digital twin record includes: Original information, including donor informed consent, demographic characteristics, clinical diagnosis, and collection time and location; Process data, including standard operating procedure versions, operation logs, equipment operating parameters, reagent batches, and operator digital signatures; Quality data includes AI-screened images, quality control levels, anomaly handling records, and retest results; Location trajectory, including the complete spatiotemporal trajectory from the collection point to the final storage, containing the timestamp and the person in charge of each transfer.

[0064] Preferably, the operating parameters of the equipment include rotational speed, temperature, and time.

[0065] Preferably, the model in the digital twin resource unit is dynamically linked to the status of physical assets in the intelligent sample full-process management system.

[0066] The technical solution of the invention has achieved the following beneficial technical effects: Compared with existing technologies, the intelligent sample end-to-end management system and method provided by this invention, through the construction of a multi-layer architecture with an intelligent dynamic sample management system as the core and IoT sensing and automated execution layers, achieves refined and intelligent management of the entire life cycle of biological samples, improves the quality of acquired biological samples, and forms a full-process core capability of "intelligent training and competency certification - intelligent guidance and standard operating procedure execution - real-time quality monitoring and proactive intervention - automated metadata capture and inventory management - quality-driven outbound and full-process traceability", specifically reflected in the following aspects: 1. Achieve deep integration of standard operating procedures and resources to improve the consistency and accuracy of process execution. This invention transforms traditional text-based standard operating procedures (SOPs) into structured data objects that the system can parse and execute by setting up a dynamic resource allocation module and introducing a resource requirement parsing function driven by standard operating procedures. The system can automatically extract the reagent and consumable list required for each atomization step, recommend the optimal batch based on the inventory database, and dynamically reserve equipment time slots according to the time consumption of the SOP and the equipment status. This mechanism changes the existing situation where SOP execution and resource allocation are separated, avoiding deviations and omissions caused by manual operation, ensuring that each sample is processed strictly according to standardized procedures, and guaranteeing the consistency of sample quality from the source.

[0067] 2. Establish a priority-based intelligent resource allocation mechanism to ensure the quality and safety of critical samples. This invention employs a priority-based preemptive resource allocation algorithm, comprehensively considering multiple factors such as sample timeliness, sample rarity, and project strategic importance to construct a dynamic weighted priority model. When resource conflicts occur between high-priority and low-priority tasks, the system can automatically negotiate resource release or trigger an early warning, supporting manual intervention in decision-making. This mechanism helps ensure priority resource guarantees for sample tasks with high timeliness requirements, high rarity, and strategic significance, reducing the risk of quality degradation of critical samples due to resource waiting or conflicts.

[0068] 3. The system establishes dynamic inventory early warning and proactive management capabilities, reducing potential quality risks related to consumables. This invention monitors consumable inventory status in real time, combining safety stock thresholds with historical consumption rates and future task demand forecasts to automatically trigger replenishment notifications and intelligently recommend order quantities. Simultaneously, the system marks consumables nearing their expiration date and prioritizes their use in resource allocation, while automatically disabling expired consumables. This proactive inventory management approach helps reduce the impact on sample quality caused by consumable shortages or the use of expired consumables, providing a stable and reliable material guarantee for the entire sample processing process.

[0069] 4. Construct a full-process digital twin and quality closed loop to achieve traceability and predictability of sample quality. This invention utilizes the collaborative operation of various levels within an intelligent sample end-to-end management system to create a unique digital twin record for each physical sample, and to perform real-time monitoring and data collection throughout the entire sample processing process. The system can assess or issue warnings based on changes in sample quality, achieving closed-loop control of sample quality. When sample quality anomalies occur, the system can quickly trace back to specific stages, equipment, personnel, and consumable batches, providing data support for locating and improving quality issues. Simultaneously, it supports quality trend prediction based on historical data, shifting from passive response to proactive prevention.

[0070] 5. Supports multi-level collaboration and remote control, improving system applicability and response efficiency. This invention achieves a complete chain from local automated execution to central intelligent decision-making and then to remote collaboration through a collaborative architecture comprising a multimodal hybrid terminal, an IoT sensing and automated execution layer, a network and communication layer, a central decision-making platform, a remote collaboration and control terminal, and a digital twin platform layer. Administrators can remotely monitor sample processing status in real time, receive early warning information, and perform resource scheduling and intervention, significantly improving the system's applicability and emergency response efficiency in complex experimental environments, and providing technical support for the full-process management of biological samples.

[0071] The following detailed embodiments further illustrate the above-described content of the present invention. However, this should not be construed as limiting the scope of the present invention to the following examples. All technologies implemented based on the above-described content of the present invention fall within the scope of the present invention. Attached Figure Description

[0072] Figure 1 This is a diagram of an intelligent sample end-to-end management system; Figure 2 This is a flowchart of the intelligent sample full-process management process; Figure 3 This is a module diagram of the Internet of Things (IoT) sensing and automated execution layer; Figure 4 This is a module diagram of the digital twin platform layer. Detailed Implementation

[0073] definition: In this invention, the "intelligent sample full-process management system" refers to a system used for the full-process management of biological samples.

[0074] The samples used in this invention are biological samples.

[0075] It should be noted that the algorithms for data acquisition, transmission, storage and processing steps not specifically described in the embodiments, as well as the hardware structures and circuit connections not specifically described, can all be implemented using content already disclosed in the prior art.

[0076] Example 1: A dynamic resource configuration module This embodiment provides a resource dynamic configuration module applied to an intelligent sample dynamic management system. The resource dynamic configuration module includes: (1) The resource requirement parsing module is configured to perform resource requirement parsing, including the following steps: Step A: When the administrator configures the standard operating procedures for a project in the intelligent sample full-process management system, the resource requirements associated with each atomic step of the standard operating procedures are parsed. Step B: Extract the required reagent or consumable list, and combine it with the database of the intelligent sample full-process management system to recommend the catalog number, batch and quantity of reagents or consumables according to different standard operating procedures; Step C: Based on the estimated single-step time and total process duration of the standard operating procedure, and combined with the corresponding equipment usage time slot, the device status is synchronized in real time through the IoT sensing and automated execution layer of the intelligent sample full-process management system to dynamically adjust the availability of the reservation. After steps D, B, and C are completed, the system locks the consumable inventory and schedules the corresponding time slot in the equipment calendar; the locking information is synchronized to the intelligent sample end-to-end management system in real time. (2) The algorithm configuration module is configured as follows: A priority-based preemptive resource allocation algorithm is used for resource allocation. The algorithm flow is as follows: Calculate the overall priority score: P=w1·T urgency +w2·V sample +w3·S level Where T urgency V is the sample timeliness factor. sample S is the sample rarity factor. level The project strategic level factor is P, which is the comprehensive priority score, with a normalized value range of 0 to 1. w1, w2, and w3 are weights, and their distribution is dynamically adjusted according to the current state of the laboratory, satisfying ∑w i =1; In a preferred embodiment, the weights can be dynamically adjusted based on the current state of the laboratory, as shown below: Normal mode: Balancing value and timeliness.

[0077] Weighting: w1=0.4, w2=0.4, w3=0.2.

[0078] Emergency mode: When the system detects more than 20% of the samples T urgency =0.9 (about to expire), then timeliness takes priority, project level is temporarily ignored, and all efforts are made to salvage the samples that are about to expire.

[0079] Weighting: w1=0.7, w2=0.2, w3=0.1.

[0080] VIP Protection Mode: For example, during the launch phase of specific high-strategy projects, strategic priority is given to ensure exclusive access to key project resources.

[0081] Weighting: w1=0.3, w2=0.3, w3=0.4.

[0082] The priority of obtaining reagents, consumables, or equipment for the project is determined by the ranking of the comprehensive priority score. (3) The dynamic inventory early warning and replenishment module is configured as follows: The system dynamically monitors the inventory status of consumables in the inventory database of the intelligent sample full-process management system, updates the consumables status to the administrator in real time, and makes predictions, alarms, and notifications to the administrator based on the consumables status.

[0083] The algorithm configuration module specifically includes: P is the core quantitative indicator used by the system to determine "which sample to process first" or "which device to allocate first". The system calculates the P value of all pending tasks and arranges and schedules them in descending order of P value. That is, the higher the P value, the more urgent and important the task (such as centrifuge, pipette, or operator time), and the higher the priority it will receive resources.

[0084] This is the sample timeliness factor, reflecting the urgency of the maximum allowable time window from sample collection to processing completion; the calculation formula is: in, For the current time, Sampling time, This refers to the maximum processing time specified in the standard operating procedure (e.g., serum separation must be completed within 2 hours). The closer to the timeout threshold, the more... The closer a value is to 1, the higher its priority.

[0085] V sample The sample rarity factor is a tiered and quantified system based on the sample's scarcity, non-renewability, and research and clinical value, as shown in Table 1: Table 1 If a sample is identified as "high-quality" or has a "special phenotype" in the initial screening by AI, the system will automatically assign it to V. sample The value has been increased by one level.

[0086] S level This is a strategic level factor for projects, reflecting the urgency, level, and funding scale of the projects. level The sample bank administrator presets the values ​​in the intelligent sample end-to-end management system and supports dynamic adjustment. For example: national-level public health emergency projects are set to 1.0; strategic R&D projects are set to 0.7; routine national-level projects are set to 0.6; provincial and ministerial-level projects are set to 0.5; and departmental-level projects are set to 0.4.

[0087] The weights w1, w2, and w3 are dynamically adjusted according to the current state of the laboratory, satisfying ∑w i =1.

[0088] When a high-priority task is submitted, the system checks whether the resources it requires are being used by a low-priority task. If so, the system can automatically negotiate to release the resources (re-queueing shared devices) or trigger an alert (an alert for resources being used by low-priority tasks), notifying the administrator to make a manual decision and providing a high level of resource protection for critical projects.

[0089] Example 2 Intelligent Sample Dynamic Management System An intelligent sample dynamic management system, integrating the resource dynamic configuration module described in Embodiment 1, further includes the following modules: Sample quality assessment module: Monitors samples and sample containers in real time and collects background data, and conducts assessments or issues warnings based on changes in sample quality; Intelligent Project and Personnel Management Module: Establishes digital profiles for projects and personnel; Structured Standard Operating Procedures Management Module: Used to convert text-based standard operating procedures into structured data objects that the system can parse and execute; Process quality assessment module: Real-time monitoring of personnel operations and sample processing, collection of background data, and assessment or early warning based on comparison with atomized standard operating procedures; The specific functions of each module are as follows: Module 1, Dynamic Resource Configuration Module The module described in Example 1 is used.

[0090] Module 2, Sample Quality Assessment Module At each critical juncture of sample processing, the integrated AI recognition engine analyzes images captured in real-time by the AR terminal, instantly assesses sample quality, and simultaneously feeds the evaluation results and operational instructions back to the operator's AR view, achieving closed-loop control of "operation, quality inspection, and guidance simultaneously." The module has a built-in standardized sample quality AI recognition engine interface, enabling connection to the central decision-making platform. It receives raw sample quality screening data and analysis results uploaded from the AR interactive terminal in real-time, including sample quality inspection images and AI recognition ratings. These include hemolysis level (0-4), lipemia level (none / mild / moderate / severe), clot detection (none / punctate / lumpy / severe), and label integrity (complete / blurred / damaged / missing).

[0091] When the operator places the sample in the AR camera's field of view according to the standard operating procedure, the system automatically triggers image capture and quality analysis. The analysis results are superimposed on the operator's real field of view in a non-invasive manner. The system provides real-time feedback based on the sample's quality status and guides the next step of the operation, including continuing the operation when the sample is qualified, continuing the operation with a note on minor quality issues, and suspending the processing of the sample when there are serious quality issues. For consecutive serious quality issues (such as hemolysis in multiple samples from the same batch), the system can trigger global intervention, suspend the processing of the entire batch, and notify the administrator for remote review.

[0092] Meanwhile, the backend of this module has the function of quality data aggregation and analysis, including: automatic collection of real-time quality records for each sample to form a complete archive containing images, timestamps, and operator information; statistical analysis of the quality distribution of samples in the same collection or processing batch, generating histograms and control charts; identification of personnel requiring key supervision and training by statistically analyzing the quality pass rate and error type distribution of each operator, and the system can automatically push targeted practice tasks to them; real-time monitoring of the dynamic changes in quality indicators, and when a trend is found such as an increase in the failure rate of multiple batches of samples or an increase in abnormalities in samples processed by a certain equipment, an early warning is immediately issued to the administrator, and it is recommended to check the reagent batch, equipment calibration status, or personnel operating procedures.

[0093] Module 3, Intelligent Management of Projects and Personnel The intelligent project and personnel management module is used for project setup and personnel creation and planning. Specifically, it includes project management and personnel management.

[0094] Project Management: Create projects in the intelligent sample end-to-end management system, establish project files, including Project Management Code (PMC), project name, and project leader; configure a dedicated set of standard operating procedures, equipment access permission pool, and consumable budget for each project; and authorize the operating permissions of personnel within the project (such as the range of samples that can be accessed and the equipment that can be reserved).

[0095] Personnel Management: A multi-dimensional digital profile of personnel is established, storing operator information (ID, name, role, department, etc.). A dynamic skills digital profile is created for each person. The module automatically generates personalized training paths and tasks based on the project to which the person belongs. It records each person's training history, mastery status of standard operating procedures (no training, learned, passed virtual assessment, authorized for practical operation, etc.), assessment scores, and qualification status. The "authorized for practical operation" status is a necessary condition for personnel to obtain the qualification to perform actual on-site operation tasks and use equipment; this status is only assigned by the sample database trainer after the person passes the pre-job intelligent training assessment. The dynamic skills digital profile also includes operational error statistics, used to dynamically adjust the level of detail in guidance and access permissions.

[0096] Module 4, Structured Standard Operating Procedure Management Module This is used to convert text-based standard operating procedures (SOPs) into computer-parseable and executable structured data objects. These SOPs not only contain textual steps but can also be associated with specific equipment models, consumable catalog numbers (CAT numbers), consumable batches, parameter thresholds (such as centrifugal force ranges), quality control points, and AR guidance content. The core data structures include: (1) Standard Operating Procedure Template Table: Stores the basic metadata of standard operating procedures (ID, name, version, applicable items).

[0097] (2) Atomized Step Table: Each standard operating procedure is broken down into the smallest indivisible operating unit (atomic step). Each atomic step contains the following key fields: step description, operation object (associated with equipment or consumables, specifying the equipment number or consumable CAT number of this step), parameter range required for the step stored in JSON format (e.g., centrifugal force: {"min": 1500, "max": 2200, "unit": "xg"}), quality control standards for this step (e.g., "centrifuge tubes must be symmetrically balanced", "pipettes must be aspirated vertically"), warning messages (e.g., "this reagent is corrosive"), expected time for efficiency assessment, and predecessor / successor steps (defining the logical dependencies between steps, forming a directed acyclic graph to ensure that the process does not skip steps).

[0098] Based on the above elements, the system can break down electronic standard operating procedures into a series of indivisible "atomic operation steps" with logical relationships, such as: SOP001 (serum aliquoting, V1.0, applicable to projects with project management number PMC001). a. Scan the barcode on the specimen tube; b. Select cryopreservation tubes (CAT No.: C001); c. Print and paste labels; d. Balance the sample; e. Select centrifuge A001; f. Select centrifugation program A001SOP1 (1500xg, 10min, 4℃); g. Use pipette P001 (adjust the volume range to 200μL); h. Draw 200 μL of serum; i. Add to cryovial; j. Register location.

[0099] Module 5, Process Quality Assessment Module The system continuously analyzes the operational scenarios. If it detects violations such as not wearing protective equipment as required or not operating inside a biosafety cabinet, it immediately displays a red warning icon in the center of the field of view, plays an audio warning, and freezes the standard operating procedure guidance until the compliant behavior is restored.

[0100] Image recognition is used to determine whether the liquid level after dispensing is within the allowable range, whether the pipette range setting is correct, and whether the stratification after centrifugation is clear, ensuring that the operation results meet the requirements of the standard operating procedure.

[0101] When any anomaly is detected, in addition to local alarms, alarm information and on-site images will be pushed in real time to the remote control terminal and the project manager's mobile terminal via the 5G gateway.

[0102] Example 3: An Intelligent Sample Management System An intelligent sample end-to-end management system, integrating the sample dynamic management system described in Embodiment 2, further includes: (a) The multimodal hybrid terminal is configured as the main interactive interface for operator training and assessment and actual operation. The multimodal hybrid terminal enables personnel to collect and process samples. (ii) The Internet of Things sensing and automated execution layer is configured to achieve full-dimensional sensing and automated data collection of samples, environment and equipment during the actual operation phase; and to provide real equipment status simulation data for virtual-real fusion assessment during the training phase. (iii) The network and communication layer is configured to enable communication connections between field devices and cloud / local servers; (iv) The central decision-making platform is configured to integrate a data processing module, a behavior perception and intelligent evaluation module, a practical intelligent analysis module, and a knowledge graph and continuous learning module to realize functions such as personnel training, sample tracking, data processing, behavior evaluation, model learning and updating; and to export intelligently generated full life cycle reports of samples with one click. (v) Remote collaboration and control terminal, which is configured to enable authorized personnel to remotely manage and collaborate on the entire process of training and practical operation on multiple terminals. Specifically, it includes: scenario and project configuration, sample task initiation, training system construction, real-time monitoring, dynamic event injection, proactive intervention and guidance, training assessment and authorization management, as well as review and analysis functions.

[0103] (vi) The digital twin platform layer is configured to build and store digital twin devices, samples and spatial information.

[0104] In this embodiment, the specific configurations for each level are as follows: (a) Multimodal hybrid terminal Serving as the primary interactive interface for both personnel training and practical operation, it supports both general mobile smart devices (smartphones and tablets) and AR glasses. The core functions of this multimodal hybrid terminal cover the entire training and practical operation process. During training, it enables virtual 3D simulation, AR fusion exercises, and assessments; during practical operation, it provides immersive operation guidance and data collection.

[0105] 1. General-purpose mobile smart terminals (smartphones, tablets): serving as the core training platform, supporting two core working modes: (1) Pure Virtual 3D Simulation Mode: Applied to the first stage (new recruits / theoretical learning) and the third stage (regular refresher training / process updates) training. The virtual laboratory 3D scene package delivered by the service layer is rendered and run in full screen on the smartphone or tablet terminal screen, and trainees complete all interactions through touch. This mode supports offline operation.

[0106] (2) Mobile AR Enhanced Mode: Applied to the second stage (pre-job practical simulation assessment), it runs on general smartphones or tablets. Based on the atomic operation steps, it calls the terminal camera to capture the real laboratory environment and accurately overlays virtual guidance information (such as highlighted boxes, arrows, animations, and operation panels) onto real objects in the AR field of view. Trainees interact with virtual devices and consumables through touch screen and virtual-real integrated scenes. The engine verifies in real time whether the trainee's operation sequence, object selection, and parameter settings comply with the SOP and provides immediate error correction feedback.

[0107] 2. AR glasses terminal: (1) Display and projection unit Composed of an optical waveguide display module and a MEMS micro-projection unit, it is used to overlay virtual guidance information (including arrows, text, highlighted boxes, 3D animations, virtual operation panels, etc.) onto the operator's real field of vision in a perspective manner, adapting to immersive guidance for hands-on operation.

[0108] (2) Sensing and positioning unit ① SLAM localization module for spatial positioning and modeling (including a set of binocular RGB cameras and a time-of-flight (ToF) depth sensor): Constructs a 3D map of the operating environment in real time and achieves accurate spatial registration and tracking of virtual information. The SLAM module adopts an improved algorithm based on ORB-SLAM3, supporting rapid relocalization in dynamic environments and persistent map storage.

[0109] ② High-definition camera: Collects video streams of the operation site for behavior perception, sample quality identification, and operation compliance monitoring.

[0110] ③ Ambient light sensor: Collects ambient light and adjusts display brightness to adapt to different laboratory environments.

[0111] ④ Microphone array: Collects voice commands and supports voice interaction during training and practical operation.

[0112] ⑤ Eye-tracking system: Built-in infrared camera and sensors accurately track the operator's gaze focus, helping to determine whether their attention is focused on the correct object, and providing data for behavioral assessment.

[0113] (3) Computing and Communication Unit It enables high-speed data interaction between local data processing, AR interface rendering, and all system modules, while also supporting remote sharing of training / practical demonstration screens, including: ① Edge computing chipset: integrates CPU, GPU and AI acceleration core (such as NPU) for local real-time processing of sensor data, running lightweight AI models (such as gesture recognition, preliminary object detection, behavior and action analysis) and rendering AR / virtual 3D interfaces to meet the local lightweight data processing needs of training and practical operation.

[0114] ② Multi-mode communication module: Supports 5G / Wi-Fi 6 / Bluetooth, ensuring data exchange with other parts of the system and enabling real-time uploading of training data and operational data, as well as command reception. The AR glasses can use this communication module to share their first-person view and user interface to the control center's large screen in real time, facilitating remote monitoring and collaboration.

[0115] ③ Voice module: integrates a speaker and microphone to receive voice commands and provide voice prompts, alarms and training guidance.

[0116] (II) Internet of Things Sensing and Automated Execution Layer It provides comprehensive perception and automated data collection of samples, environment, and equipment for the actual operation phase, and can also provide simulated data of real equipment status for the virtual-real fusion assessment during the training phase, including the following units: 1. Sample Identification and Tracking Unit This includes cryogenic UHF RFID tags attached to specimen containers (such as vacuum blood collection tubes, urine cups, tissue boxes, etc.), cryopreservation tubes, and cryopreservation boxes; fixed RFID readers deployed at key operational nodes (receiving stations, centrifuges, biosafety cabinets, ultra-low temperature freezers, liquid nitrogen container inlets); and handheld readers integrated into AR glasses or mobile terminals, supporting accurate reading at close range (≤10cm). This enables rapid, batch, and contactless identification and location tracking of samples. During the training phase, virtual RFID tags can be used to track sample models and verify operational procedures.

[0117] 2. Environmental monitoring unit Temperature and humidity sensors, liquid nitrogen level sensors, and door magnetic sensors deployed in laboratories, cold storage facilities, and refrigerators collect real-time data on ambient temperature and humidity, liquid nitrogen level, and equipment operating status within refrigerators and liquid nitrogen tanks. This data is then uploaded to the platform in real-time via an Internet of Things (NB-IoT) gateway. The platform compares the uploaded real-time data with thresholds; any data exceeding these thresholds is marked as "abnormal data," triggering an alarm according to established rules. During training, simulated abnormal environmental scenarios can be used to assess trainees' emergency response capabilities.

[0118] 3. Intelligent experimental equipment This includes equipment such as smart pipettes that support data interfaces, low-temperature high-speed centrifuges that are associated with intelligent sample end-to-end management systems, and fully automated blood component separation systems. These devices can automatically read and upload operating parameters (such as time, temperature, volume, and rotation speed) to achieve automated recording of equipment parameters during the hands-on phase. During the training phase, the operation and parameter settings of the equipment can be simulated through digital twin models of the equipment to achieve an assessment experience consistent with that of the real equipment.

[0119] (III) Network and Communication Layer This integrated device, using a 5G industrial gateway / CPE as a hub, combines a 5G modem, Wi-Fi router, and industrial gateway functions to achieve high-speed, stable connections between field devices and cloud / local servers, while supporting end-to-end data transmission for training and practical operations. Its core functions include: 1. Achieve full-level data interoperability across the field layer (AR glasses, general mobile devices, RFID readers, IoT sensors, smart devices), platform layer (central decision server, local intelligent sample full-process management system server), and cloud resources (cloud fault knowledge base, AI model training platform, teaching management server).

[0120] 2. By leveraging the characteristics of high bandwidth, low latency, and high connection density, the system ensures stable and real-time transmission of video streams, sensor data, training operation data, and hands-on guidance instructions for AR interactive terminals.

[0121] 3. Connect to the cloud-based fault knowledge base and teaching management server to support remote teaching and resource distribution during the training phase, as well as remote diagnosis and emergency guidance during the practical training phase.

[0122] (iv) Central decision-making platform The central decision-making platform server is deeply integrated with the intelligent sample full-process management system. It is the core computing and decision-making unit of the system, supporting intelligent decision-making in both training and practical stages. It includes the following core intelligent modules: 1. Data Processing Module The system performs preprocessing such as decoding, cleaning, and feature extraction on multi-channel audio, video, sensor timing data, touch screen operation data, training operation data, and practical operation data from AR terminals / general mobile devices to provide standardized data for training evaluation and practical analysis.

[0123] 2. Pre-employment Intelligent Training Engine: Responsible for the logical processing of pre-employment training operations, including: (1) Adaptive Virtual Scene Generation Engine As the core intelligent module for the training phase, based on the pre-set spatial layout units, digital twin resource units, and structured standard operating procedures of the intelligent sample end-to-end management system, it automatically generates the optimal virtual laboratory 3D scene matching the project. Its core comprises two major algorithms: ① Template matching algorithm based on SOP semantics: Extract keywords from SOP (such as "biosafety cabinet", "centrifugation", "dispensing"), calculate the cosine similarity between keywords and the description vector of each scenario template, and select the template with the highest similarity as the basis.

[0124] ② Automatic layout algorithm based on bounding box collision detection: Equipment resources are allocated to the predefined functional area of ​​the virtual laboratory. Core equipment is placed at the template "anchor point" position. Subsequent equipment finds the optimal collision-free placement position through gradient descent. Consumables are placed in easily accessible positions according to the order and frequency of use, thus achieving a rational layout of the virtual scene.

[0125] (2) Multimodal training scheduling and allocation engine Based on the trainees' training stage, learning progress, and current environment, the system intelligently allocates and schedules the most suitable training mode and content, achieving adaptive adjustment of training. The core scheduling rules are as follows: ① First stage (new recruits / theoretical learning): Assigned to "pure virtual 3D mode", the system generates a complete virtual laboratory panoramic scene, which trainees can freely explore and learn the process logic through touch screen interaction on general mobile devices. No real laboratory environment is required and offline operation is supported.

[0126] ② Second stage (pre-job practical assessment): Trainees are assigned to the "AR virtual-real fusion mode". In a real laboratory, trainees use AR terminals to overlay virtual operation instructions and verification information onto real equipment and consumables to conduct high-fidelity practical assessments.

[0127] ③ Third stage (regular refresher training / process update): It is recommended to use the "pure virtual 3D mode" for quick review and practice, and AR mode can be selected for targeted reinforcement.

[0128] (3) Multimodal behavior perception and intelligent evaluation engine The core assessment module in the training phase collects operational data from trainees under different training modes (touchscreen event sequences, AR interaction actions, operation time, and error records). Using a pre-set rule engine and machine learning model, it quantitatively scores the fidelity, accuracy, standardization, and efficiency of the operational process from multiple dimensions, generating a detailed training assessment report, which is then stored in the personnel's skill profile within the intelligent sample-based full-process management system. The assessment results serve as the core basis for authorizing personnel for practical operations. During the practical operation phase, the proficiency of operators can be continuously evaluated, and the level of guidance detail can be dynamically adjusted.

[0129] 3. Practical Intelligent Analysis Module The core analysis module in the practical phase integrates and analyzes multi-dimensional data from AR terminals to achieve sample quality identification, operational compliance monitoring, and operational status judgment. It comprises three sub-modules: (1) Multimodal operation tracking submodule: integrates and analyzes the operator's gaze focus, gestures, voice command sequence and decision response time to construct the operator's "attention-action" time sequence chain and accurately judge the current operation steps and status.

[0130] (2) Evaluation and identification submodule: Run a deep learning model (such as convolutional neural network CNN) to analyze the sample images captured by the AR camera in real time, and realize sample quality identification (hemolysis, lipemia, clots, label blurring), operation compliance identification (wearing gloves, masks, and lab coats; operation in biosafety cabinet, standardized sample addition action) and result verification (whether the stratification after centrifugation is clear and whether the liquid level height meets the standard).

[0131] (3) Analysis instruction submodule: Make decisions based on SOP, personnel operation tracking results and evaluation and identification results. If normal, generate the next AR guidance instruction; if an abnormality is identified, generate the corresponding alarm and intervention instructions.

[0132] 4. Knowledge Graph and Continuous Learning Module Based on historical training data, abnormal assessment events, practical operation data, and sample quality anomaly data, a full-process knowledge graph is constructed. The system can learn from newly generated data, continuously optimize the training evaluation model, training program, guidance strategy, and sample quality identification model, and achieve self-iterative upgrades.

[0133] (v) Remote collaboration and control terminal This is a web-based, cross-platform application that authorized professionals (such as administrators, project managers, and trainers) can access and run on computers, tablets, mobile phones, and control center screens. Its functions cover two main scenarios: training management and practical monitoring, enabling remote management and collaboration throughout the entire process. Core functions include: 1. Scenario and Project Configuration Configure SOP processes, associate resources, and set quality and alarm thresholds for different projects; at the same time, create exclusive training programs for different projects, associate electronic SOPs with training resources, and set training assessment standards and quantitative indicators.

[0134] 2. Real-time monitoring View the training / practical operation shared perspective and system status of any AR terminal / general mobile device in real time, and keep track of the training progress and the current operation status of the personnel.

[0135] 3. Dynamic event injection During training or drills, scenarios such as equipment failure, sample anomalies, and environmental anomalies are remotely simulated to test the operators' responsiveness and emergency response capabilities.

[0136] 4. Proactive intervention and guidance During the practical operation phase, when the system alarms (such as excessive temperature or operational violations), the air conditioning, ventilation, lighting and other facilities can be remotely adjusted directly through this interface. During the training and practical operation phase, professionals can remotely annotate or initiate two-way voice and video calls through the AR interface to provide real-time guidance.

[0137] 5. Training, assessment, and authorization management The system allows remote viewing of trainees' training records and assessment results, and online authorization for trainees who pass the assessment. It also allows adjustment of personalized training paths based on trainees' assessment performance.

[0138] 6. Review and Analysis Playback of historical training operation videos, practical operation videos, and problem debriefings can be used for training, assessment, accident analysis, and process optimization. At the same time, typical cases can be included in the training resource library to improve the relevance of training.

[0139] (vi) Digital Twin Platform Layer Before project configuration, the laboratory space layout units and digital twin resource units are set up and saved on the digital twin platform. During sample management, the digital twin platform resources are invoked in real time to achieve refined, end-to-end sample management.

[0140] 1. Spatial Layout Unit Primarily used for virtual laboratory layout, including basic laboratory function templates and scene template library.

[0141] Laboratory Basic Function Templates: The system pre-designs basic laboratory layout templates based on the characteristics of experimental functions and the most commonly used ones. These include basic function templates such as "Sample Receiving Area", "Centrifugation Area", "Biosafety Cabinet Operation Area", "Automatic Cryopreservation Tube Cap Opening Area", "Cryopreservation Tube Bottom QR Code Scanning Area", "2℃-8℃ Low Temperature Temporary Storage Area", "-80℃ Ultra-Low Temperature Storage Area", and "-196℃ Deep Low Temperature Storage Area".

[0142] Scene Template Library: Generated based on basic laboratory function templates, each scene template defines the division and relative positional relationships of core functional areas. For example, the core area of ​​scene template A (biosafety cabinet operation scene) is a large biosafety cabinet occupying the center of the screen, with auxiliary areas on the left and right sides for placing samples and consumables; the core area of ​​scene template B (centrifuge area centrifugation scene) is a floor-standing low-temperature high-speed centrifuge occupying the center of the screen, with auxiliary areas next to the centrifuge for placing specimen tubes such as blood collection tubes and a workbench for placing a balance.

[0143] 2. Digital Twin Resource Unit It includes digital twin models of equipment and consumables. The models support Level of Detail (LOD), are structured, searchable, and renderable, and can be dynamically linked to the physical asset status (such as equipment operation data and consumable inventory) in an intelligent sample end-to-end management system. When applied to the training module, it only contains static modeling data and interaction logic definitions; it does not connect to real equipment, collect real-time status, or perform physical control. It is used to support AR training systems in dynamically generating virtual laboratory scenes according to standard operating procedure types.

[0144] 3. Digital Twin Archive Unit A unique digital twin record is created for each entity sample, using the sample's unique identifier as the primary key, and dynamically aggregating four main categories of data generated throughout its entire lifecycle, including: (1) Original information: donor informed consent form, demographic characteristics, clinical diagnosis, collection time and location, etc.

[0145] (2) Process data: standard operating procedure version, operation step log, equipment operating parameters (speed / temperature / time), reagent batch, operator digital signature, etc.

[0146] (3) Quality data: AI initial screening images, quality control level, anomaly handling records, retest results, etc.

[0147] (4) Location trajectory: The complete spatiotemporal trajectory from the collection point to the final storage, including the timestamp of each transfer and the person in charge.

[0148] Digital twin records use blockchain hash chain technology to ensure that the data is immutable. Each record contains the hash value of the previous record, forming a complete audit trail.

[0149] Through deep collaboration among various subsystems, a full-process core capability is formed, consisting of "intelligent training and competency certification - intelligent guidance and SOP execution - real-time quality monitoring and proactive intervention - automated metadata capture and inventory management - quality-driven outbound delivery and full traceability". Among them, intelligent training and competency certification is the core function in the early stage, while the other functions are the core capabilities in the practical stage. All functions have proactive intervention and adaptive characteristics.

[0150] Core Function 1: Intelligent Training and Competency Certification As a prerequisite for actual operation, standardized, quantifiable, and immersive training and competency certification are implemented for personnel. Only after passing the assessment can practical authorization be granted. The core process is as follows: 1. Professionals create training programs for specific projects in the intelligent sample full-process management system, linking the project's exclusive atomic SOP, digital twin resource units and spatial layout units, and setting quantitative assessment standards.

[0151] 2. The system automatically generates a personalized training path based on the trainer's role and project permissions. Trainers can log in to the system through a common mobile device / AR terminal and enter the corresponding training project.

[0152] 3. The multimodal training scheduling and allocation engine of the central decision-making platform intelligently allocates pure virtual 3D mode / AR virtual-real fusion mode according to the trainees' learning progress, and the adaptive virtual scene generation engine automatically generates matching training scenarios.

[0153] 4. Trainees practice their skills in the training scenario. The system collects operation data in real time through multimodal behavior perception and intelligent evaluation engine, and performs process verification and immediate error correction feedback.

[0154] 5. After completing the training, trainees will participate in a standardized assessment. The system will conduct multi-dimensional quantitative scoring and generate an assessment report. After passing the assessment, the sample library trainers will grant them practical operation permissions for the corresponding projects in the intelligent sample full-process management system, and the system will update their skill digital files at the same time.

[0155] 6. The system supports regular refresher training and process update training for trainees. For operational errors that occur during practice, it pushes targeted reinforcement training content in reverse to achieve a closed loop of "training-practice-refresher training".

[0156] Core Function Two: Intelligent Guidance and SOP Execution For authorized personnel, an immersive, step-by-step AR-guided sample processing procedure is implemented to ensure strict adherence to Standard Operating Procedures (SOPs). Core process: 1. After logging into the AR terminal and verifying your identity, the system will automatically load the project-specific SOP from the intelligent sample full-process management system to the central decision-making platform server according to your authorized project.

[0157] 2. The AR terminal uses SLAM to locate the operating table and guides the operation step by step in the real field of vision using highlighted arrows, virtual outlines, 3D animations, etc. The guidance content is consistent with the training stage.

[0158] 3. At critical steps, the system enforces verification: for example, it confirms that the centrifuge lid is closed through image recognition and automatically records the centrifugation time, or it requires the operator to enter parameters through voice or virtual keyboard. The next step is only unlocked after the parameters are verified to be correct, to prevent problems such as skipping steps or incorrect parameters.

[0159] Core Function Three: Real-time Quality Monitoring and Proactive Intervention Throughout the entire sample processing process, real-time monitoring of sample quality, operational compliance, and process accuracy is implemented. Upon detecting anomalies, local and remote alarms are immediately triggered, and the process is paused until the problem is resolved. Core content: 1. Initial screening of sample quality: Before the specimen is aliquoted, the AR terminal camera automatically takes a picture. After the image is uploaded, the evaluation and recognition module returns a quality rating within a few seconds and displays it on the lens. If a serious abnormality is found (such as severe coagulation), the process is paused, the sample is prompted to be replaced, and the reason for the non-compliance is recorded.

[0160] 2. Operational compliance monitoring: The system continuously analyzes operational scenarios. If it detects violations such as not wearing protective equipment as required or not operating inside a biosafety cabinet, it will immediately display a red warning icon in the center of the field of view, play an audio warning, and freeze the SOP guidance until the compliant behavior is restored.

[0161] 3. Process accuracy verification: Image recognition is used to determine whether the liquid level is within the allowable range after dispensing, whether the pipette range setting is correct, and whether the stratification is clear after centrifugation, to ensure that the operation results meet the SOP requirements.

[0162] 4. Remote Anomaly Push: When any anomaly is detected, in addition to local alarms, alarm information and on-site images will be pushed in real time to the remote control terminal and the project manager's mobile terminal via the 5G gateway, supporting remote consultation and decision-making.

[0163] Core Function 4: Automated Metadata Capture and Inventory Management The core content includes: automating the collection of all data during sample processing, creating digital twin profiles for samples, and dynamically updating inventory. 1. Seamless data entry: When the operator scans the sample tube, the relevant sample information is automatically retrieved and displayed from the intelligent sample full-process management system; equipment operating parameters such as centrifugation parameters and pipetting volume are automatically read through the equipment interface; all data such as operator ID, timestamp, sample quality data, and operation process data are written into the digital twin file of the corresponding sample in real time, forming an unalterable holographic record.

[0164] 2. Dynamic inventory update: When samples are received, operators place the cryopreservation tubes into the designated boxes under AR guidance. The AR terminal confirms the matching of the tubes and positions through image recognition, and the RFID reader tracks the sample position. When a sample is placed into or removed from the storage device, the physical location information and inventory status of the sample in the intelligent sample full-process management system are automatically and seamlessly updated without the need for manual scanning.

[0165] Core Function 5: Quality-Driven Outbound and End-to-End Traceability To achieve accurate retrieval and warehousing based on sample quality, and to enable full lifecycle traceability of samples from receipt to warehousing, meeting regulatory review requirements, the core process is as follows: 1. Searchers can directly perform combined searches based on sample quality attributes (such as tumor cell content, RIN value, and hemolysis grade) in the intelligent sample end-to-end management system to accurately screen high-quality samples that meet research needs.

[0166] 2. When the sample is being taken out of the warehouse, the AR terminal guides the operator to the precise physical location of the target sample and confirms the barcode of the sample tube being taken out matches the application form through image recognition to prevent mis-taking; the RFID reader tracks the location of the sample and automatically updates the status of the sample to "out of warehouse" in the intelligent sample full-process management system.

[0167] 3. The system can generate a complete "sample lifecycle report" from receipt to shipment with one click, which includes all operations, quality data, audit trails, and training, assessment and qualification information of operators, which helps to meet strict audit requirements.

[0168] Example 4: An Intelligent Sample End-to-End Management Method An intelligent sample end-to-end management method includes the following steps: Step 1: Establish the project configuration and training system; Step 2: Conduct personnel training and certification authorization; Step 3: Start the sample task; Step 4: Sample collection and initial quality screening; Step 5: Process the samples; Step 6: Process the sample data; Step 7: Perform full lifecycle tracing of the sample and generate a full lifecycle report of the sample.

[0169] This embodiment achieves digital and intelligent management through a complete intelligent processing loop from "project configuration - training and assessment - personnel authorization - task initiation - sample processing - warehousing and archiving - outbound traceability," ensuring the operational quality of relevant personnel and the quality control of samples throughout their entire lifecycle. It is based on the system implementation provided in Embodiment 3.

[0170] Step 1: Configure the project and establish the training system During the preparation phase, professionals complete the integrated configuration of project and training: 1. Project Creation and Standard Operating Procedure Configuration: Professionals log in to the administrator terminal of the intelligent sample end-to-end management system to create a biological sample processing project to be executed, entering basic information such as project name, research direction, sample type, processing batch, and quality control standards. Simultaneously, the project plan is converted into a structured, parsable electronic standard operating procedure. Following the operational logic and sequence of sample processing, the complete standard operating procedure is broken down into a series of indivisible "atomic operation steps." Each step is associated with the corresponding equipment model, consumable batch, parameter threshold, quality control point, and operational requirements. For example, the serum separation standard operating procedure can be broken down into… The process involves several steps: scanning the specimen tube barcode, selecting a 1.0mL cryopreservation tube, printing and affixing a label, balancing the sample, setting centrifuge parameters, centrifuging, aliquoting serum, storing the sample, and registering its location. After disassembly, the electronic standard operating procedure is linked to the project, along with the necessary reagents, consumables, storage locations, and intelligent experimental equipment. The intelligent sample management system's dynamic resource allocation module automatically locks the associated resources, generates resource replenishment reminders based on consumable expiration dates and inventory levels, and automatically schedules usage periods for relevant equipment based on the operational steps, ensuring that resources and equipment are available during actual sample processing.

[0171] 2. Training Program Creation: In the "Personnel Training" module of the intelligent sample end-to-end management system, the sample bank administrator can create a matching "training program" for the configured sample processing projects. This program is directly linked to the electronic atomized standard operating procedure of the project. The system automatically retrieves the corresponding 3D lightweight models of equipment and consumables from the digital twin resource unit, and loads the required operating scenarios and guidance logic for AR training simulation, forming standardized training content that is completely consistent with actual operation. Administrators can add abnormal scenario drills to the training program according to training needs, such as sample hemolysis, centrifuge parameter errors, and consumable usage errors, to test the emergency response capabilities of operators.

[0172] 3. Training assessment standard setting: Professionals set quantitative assessment standards for the training project in the system, including scoring indicators for dimensions such as the fidelity of the operation process, the accuracy of the steps, the standardization of operation, and efficiency, as well as the threshold for passing the assessment, and set assessment requirements for abnormal scenario drills.

[0173] Step Two: Conduct personnel training and certification authorization This stage is the competency certification stage for relevant personnel, achieving standardized, immersive training and quantitative assessment. Practical authorization is granted only after passing the assessment; this is the core prerequisite stage, with the following core steps: 1. Training plan development and personnel allocation: Trainers develop training plans in the intelligent sample full-process management system and assign training projects to relevant trainees. The system automatically generates personalized training paths based on the basic information of each trainee.

[0174] 2. Three-stage progressive training: Trainees log into the system via mobile smart terminals and access training programs. The central decision-making platform's multimodal training scheduling and allocation engine intelligently allocates training modes based on the trainees' learning progress, achieving adaptive learning. (1) Pure Virtual 3D Mode Learning (Newcomers / Theoretical Learning): The system generates a complete virtual laboratory panoramic scene. Trainees can freely explore on a general mobile device through touch screen interaction, learn the standard operating procedures and logic, and the basic operation of equipment and consumables. The system provides real-time operation guidance and instant error correction, and supports offline operation.

[0175] (2) AR virtual-real fusion mode practice (practical simulation): In a real laboratory, trainees use mobile smart terminals to overlay virtual operation instructions and verification information onto real equipment and consumables to conduct high-fidelity practical simulation practice. The system collects operation data in real time through multimodal behavior perception technology and verifies the operation behavior throughout the process.

[0176] (3) Pure Virtual 3D Mode Retraining (Reinforcement / Update Learning): Reinforcement training is conducted for the weak steps of trainees. If the standard operating procedures of the project are updated, the system will automatically push the updated content, and trainees can quickly review and learn through pure virtual 3D mode.

[0177] 3. Standardized and Quantitative Assessment: After completing the training, trainees submit an assessment application to the system. Professional personnel initiate online assessments via remote collaboration and control. The system randomly injects assessment tasks for both routine operations and abnormal scenarios, comprehensively evaluating performance across multiple dimensions, including compliance of operational procedures, accuracy of parameter settings, standardized use of consumables and equipment, and ability to handle abnormal scenarios, generating a detailed assessment report. All operational data during the assessment process is uploaded in real-time to the intelligent sample full-process management system for storage.

[0178] 4. Personnel Authorization: Trainees whose assessment scores reach the preset threshold will be granted practical operation permissions for the project by the trainer in the intelligent sample full-process management system. The system will simultaneously update the trainee's dynamic skill digital profile and mark their mastery of standard operating procedures as "authorized for practical operation". For trainees who fail the assessment, the system will automatically push targeted reinforcement training content and reassess until they pass.

[0179] Step 3: Start the sample task This step is the initiation phase of actual sample processing, completing system initialization and boot preparation before operation. Core steps: 1. Authorized operators wear AR glasses and log in to the system via facial recognition or employee ID password. They then select the authorized project task, and the system completes dual verification of identity and permissions.

[0180] 2. The AR terminal activates the SLAM positioning module to quickly map and register the current experimental environment, achieving accurate matching between virtual information and the real environment.

[0181] 3. The system retrieves the atomized standard operating procedure for the project from the intelligent sample full-process management system and converts it into a series of spatially anchored AR guidance instructions (such as "highlighting the arrow to the biosafety cabinet" and "overlaying virtual buttons and parameter input boxes on the centrifuge panel"), loads them into the AR terminal, and completes the operation guidance initialization.

[0182] 4. The ambient light sensor of the AR glasses collects the light intensity of the current operating environment in real time, and automatically adjusts the brightness and contrast of the display and projection units to ensure that the virtual guidance information is clearly visible and does not affect the operator's real field of vision, completes all the initialization work of AR guidance, and waits for the operator to start the sample processing operation.

[0183] Step 4: Sample collection and automated initial quality screening This step enables rapid sample reception and automated initial quality screening, laying the foundation for subsequent processing. It includes the following core steps: 1. Guided by the virtual arrows and voice prompts on the AR glasses, the operator places the sample box on the sample receiving station with the integrated RFID reader and activates the RFID batch identification function. The reader performs rapid, non-contact batch identification of the low-temperature resistant UHF RFID tags on all specimen containers (vacuum blood collection tubes, tissue boxes, etc.) in the sample box. The system automatically reads the unique identification information of all samples and retrieves the matching sample list from the intelligent sample full-process management system. The number of sample tubes, sample type, donor basic information, etc. are displayed in the AR field of view, realizing rapid verification of sample information.

[0184] 2. If the sample information is correct, the operator, guided by the AR voice prompts, places the sample tubes one by one into the fixed field of view imaging area pre-marked by the SLAM positioning module. This area is an optimized image acquisition area of ​​the system, which can ensure clear imaging of the sample tubes and avoid shooting deviations affecting quality recognition. If the sample information is inconsistent, the system will pop up a red warning box in the AR field of view, indicating "sample information mismatch" and marking the abnormal sample. The operator must manually check and process the information before proceeding to the next step.

[0185] 3. The evaluation and identification submodule of the central decision-making platform runs the sample quality identification model, which completes the analysis and rating of hemolysis, clots, lipemia, contamination and label integrity of a single tube sample within 0.5-2 seconds. The quality rating results are displayed in the AR field of view of the corresponding sample tube in the form of "color text box + voice prompt". The rating is divided into multiple levels such as "qualified", "mild hemolysis - suggested remarks", "moderate hemolysis - use with caution", "severe hemolysis / clots / lipemia - unqualified" and "label blurred - information supplementation". All rating results are automatically synchronized to the intelligent sample full-process management system and linked to the digital twin file of the corresponding sample.

[0186] 4. If a sample is detected as having "severe hemolysis, severe clots, or other quality defects," the system will immediately pause the operation. A flashing red warning icon will appear in the center of the AR view, indicating "Sample quality is abnormal; sample replacement is recommended." The system will also automatically record the reason for the sample's failure. Operators can report the abnormal sample via the microphone array on the AR glasses. Administrators can view the sample image and anomaly information in real time through remote collaboration and control, providing remote guidance to operators. After the abnormal sample is processed, the operator manually confirms, and the system resumes the process. If the quality defect is only minor or the label is blurry, the operator can continue with subsequent sample processing after completing the notes or information additions as prompted by the system. All relevant records are stored in real time in the intelligent sample end-to-end management system, enabling source traceability of sample quality.

[0187] Step 5: Process the sample This step is the core stage of sample processing, enabling step-by-step immersive guidance of standard operating procedures and real-time monitoring and compliance verification throughout the entire process. Core steps: 1. Operation Guidance: Based on the atomized standard operating procedures, the system uses virtual highlighted outlines to mark the equipment / consumables or operating areas to be operated on next, such as centrifuges, biosafety cabinets, and cryopreservation tubes of specified sizes. The highlighted outlines adjust their view in real time as the operator moves to ensure they are always clearly visible. Animated arrows guide hand movement, and floating text / voice explanations illustrate key operating points. The guidance content is consistent with the training phase, allowing operators to quickly adapt.

[0188] 2. Parameter Verification: The system verifies that the actual parameters on the device panel match the requirements of the standard operating procedure through image recognition, or directly reads the parameters through the smart device interface. Verification is a necessary condition for proceeding to the next step, reducing or avoiding parameter setting errors. If the parameter settings match the standard operating procedure requirements, the system will display a green "Parameter Verification Passed" message in the AR field of view, automatically unlocking the next operation. If the parameter settings are incorrect or exceed the threshold, the system will display a red warning box, indicating the incorrect parameters and showing the correct parameter range. Simultaneously, a voice prompt will say, "Parameter settings are incorrect, please readjust," guiding the operator to correct the parameters until verification is successful. If parameter verification fails, the system will lock subsequent operations to prevent erroneous operations.

[0189] 3. Compliance Monitoring: Throughout the sample processing process, the multimodal operation tracking and evaluation / recognition submodules of the central decision-making platform continuously collaborate, integrating and analyzing multi-dimensional data from the AR glasses, including "operator's gaze focus, hand gestures, on-site video stream, and voice command sequences," to construct the operator's "attention-action" time-series chain and assess the compliance of operational behaviors in real time. The system focuses on monitoring core compliance requirements, including: whether operators wear protective equipment (masks, gloves, lab coats) as required, whether sample tubes are handled within the compliant working area of ​​the biosafety cabinet, whether sample addition / dispensing operations are standardized, and whether equipment operation meets safety requirements. If violations are detected (such as ungloved hands reaching into the sample area, sample tubes being removed from the biosafety cabinet, or starting equipment without closing the centrifuge lid), a flashing red warning icon immediately pops up in the center of the AR field of view, and a voice warning is played. Simultaneously, all subsequent guidance is suspended until the operator resumes compliant behavior. The time of occurrence, type of violation, and rectification status of all violations are uploaded in real time to the intelligent sample full-process management system and recorded in the operator's operation file, serving as the basis for subsequent personnel competency assessment and retraining.

[0190] 4. Result Confirmation: After a step is completed, the system automatically confirms the operation result through image recognition. For example, after the "serum aliquoting" step, the system identifies whether the liquid level in the new cryopreservation tube is within the preset range and prompts "Liquid level is qualified" or "Liquid level is insufficient, please add more" in the AR field of view. Only after the result is confirmed can the next step be performed to ensure that the result of each step meets the requirements.

[0191] 5. Remote Anomaly Push: If the system detects any anomaly (quality anomaly, operational violation, equipment parameter error), in addition to local alarms, it will also push alarm information and on-site images in real time to the remote control terminal and the project manager's mobile terminal via the 5G gateway, supporting remote consultation and decision-making by professionals. For serious anomalies (such as excessive temperature), the system will automatically suspend all tasks of the operator until administrator confirmation.

[0192] Step Six: Processing Sample Data This step enables seamless and automated data collection throughout the entire sample processing process, building a complete digital twin profile for each sample. Specifically, it includes the following core steps: 1. Automatic metadata association: When operators scan the sample tube using an RFID reader or barcode scanner, metadata such as the sample donor ID, collection time, and sample type are automatically retrieved from the intelligent sample end-to-end management system and associated with the sample digital twin file.

[0193] 2. Automatic upload of process data: Process data such as the dispensing volume of the intelligent pipette, the running time and speed of the centrifuge, the timestamps of each step of sample processing, and the operator ID are automatically uploaded to the system through the device interface and AR terminal sensors.

[0194] 3. Automatic capture of quality data: Quality data such as initial screening images and results of sample quality, confirmation images of key steps (such as the stratification state after centrifugation), and operation compliance verification results are automatically captured and uploaded by the AR terminal.

[0195] 4. Digital Twin Archive Update: All collected metadata, process data, and quality data are written in real time through a 5G gateway into the "digital twin" record corresponding to the sample in the intelligent sample full-process management system, forming a structured and tamper-proof electronic record, realizing the full retention of sample processing data.

[0196] To achieve seamless and intelligent sample entry and dynamic automated inventory updates, ensuring accurate recording of sample locations, the following core steps are involved: 1. When the sample is ready to be stored after processing, the AR terminal guides the operator to the designated storage unit (such as a -80℃ refrigerator or liquid nitrogen tank) and highlights the location of the storage unit in the AR field of view.

[0197] 2. When the operator opens the refrigerator / liquid nitrogen tank drawer, the AR view will further highlight the specific cryopreservation box to be used and the target hole position inside the box, guiding the operator to place the sample tube.

[0198] 3. After the operator places the sample tube into the target hole, the system uses the image recognition function of the AR camera and combines it with SLAM positioning information to automatically calculate the precise three-dimensional physical coordinates of the sample tube in the storage device, such as F01 A1-1-1 A1 (indicating that the sample is placed at position A1 in the first box of the first shelf of the first shelf of the first shelf of the A layer of refrigerator F01).

[0199] 4. A fixed RFID reader deployed at the entrance of the storage unit performs a "secondary non-contact identification" of the placed sample tubes, verifying the match between the sample identifier and the target well position. If the match is correct, the system updates the sample's three-dimensional physical coordinate information to the inventory management module of the intelligent sample end-to-end management system in real time, automatically changing the sample's inventory status to "in stock," and simultaneously updating the sample's digital twin file, recording the stock entry time and storage location. If the match is inconsistent, the system immediately pops up an alert, prompting "Sample placed in the wrong location, please reposition." When samples are subsequently removed, transferred, or their storage location changes, the RFID reader will capture the sample position changes in real time, and the inventory information of the intelligent sample end-to-end management system will be automatically updated synchronously, realizing dynamic and intelligent management of sample inventory without the need for manual entry or modification.

[0200] Step 7: Perform full lifecycle tracing of the sample and generate a full lifecycle report of the sample. To achieve accurate sample release based on sample quality and full lifecycle traceability of samples from receipt to release, and to meet the requirements of scientific research and regulatory review, the following core steps are required: 1. High-quality sample precise retrieval: Researchers can log in to the web or client version of the intelligent sample end-to-end management system and enter the sample retrieval module. They can perform combined searches based on multiple dimensions such as "sample quality parameters, basic information, and processing procedures," such as "retrieving RNA samples extracted from fresh frozen tumor tissue with a tumor cell content greater than 70% and a RIN value greater than 8.0." The system quickly filters out samples that meet the criteria from the sample digital twin archive and generates a sample list. The list displays key information such as sample quality data, storage location, and processing batch. Researchers can select samples according to their needs and submit a request for release.

[0201] 2. Sample Outbound Application Review: After receiving a sample outbound application in the intelligent sample end-to-end management system, the administrator reviews the intended use, quantity, and quality requirements of the sample. If the review is approved, the system assigns a unique outbound order number to the application and links it to the digital twin file of the corresponding sample, recording the outbound application information. If the review is not approved, the administrator marks the reason for rejection and returns the application to the researchers.

[0202] 3. Precise Outbound Navigation and Sample Verification: Authorized operators wearing AR glasses log into the system, select the outbound task to be processed, and the system guides the operator to the storage unit of the target sample through AR real-scene navigation. Upon arrival, the 3D physical coordinates of the target cryopreservation box and sample tubes are highlighted in the AR view. The operator then retrieves the sample tubes according to the guidance. After the sample tubes are retrieved, the system immediately activates a "dual verification mechanism." The high-definition camera of the AR glasses performs image recognition on the sample tube label, and simultaneously, a handheld RFID reader scans the RFID tag on the sample tube. The recognition results are compared with the sample information in the outbound application form to confirm that the sample identification, type, quality parameters, and other information are completely matched. If the verification matches, the system pops up a green "Sample Verification Passed" message in the AR view; if the verification does not match, the system pops up a red warning message, "Sample does not match the application form, do not retrieve," and guides the operator to put the sample tubes back in their original positions. At the same time, the administrator will receive an anomaly alert and promptly conduct manual verification.

[0203] 4. Automatic inventory status update: After the sample is verified and retrieved, the RFID reader deployed at the entrance of the storage unit captures the changes in the sample's location in real time. The intelligent sample full-process management system automatically updates the sample's inventory status to "outbound". At the same time, it records information such as outbound time, outbound operator, outbound order number, and receiving unit in the sample's digital twin file, realizing real-time traceability of outbound information.

[0204] 5. Sample Lifecycle Traceability Report Generation: Researchers, administrators, or relevant reviewers can generate a complete "Sample Lifecycle Report" for any sample with a single click within the intelligent sample lifecycle management system, covering the entire process from receipt, processing, storage, to release. The report includes basic sample metadata, atomic operation step records for the entire processing, equipment operating parameters at each stage, quality testing data and images, operational compliance records, operator information, environmental monitoring data, inventory location change records, audit trail records, and more. The report supports export in multiple formats such as PDF, Excel, and XML, meeting the review requirements of domestic and international biobank quality management systems and regulations, including ISO 20387 and GLP.

[0205] 6. Sample follow-up traceability: After the sample is released from the warehouse, the digital twin file of the sample can be supplemented and entered through the intelligent sample full-process management system, realizing full-chain traceability of the sample from collection to use, and providing complete data support for the reliability of scientific research results.

[0206] 7. Reverse optimization training based on operational behavior: The system continuously analyzes the operational behavior data of operators. If it finds that an operator has high-frequency operational errors, the system will push corresponding targeted reinforcement training content and require the operator to complete the retraining, so as to realize the closed-loop optimization of "operation-training" and continuously improve the operator's operational ability.

Claims

1. An intelligent sample dynamic management system, characterized in that, This is applied to an intelligent sample end-to-end management system, including a dynamic resource configuration module; the dynamic resource configuration module includes the following sub-modules: (1) The resource requirement parsing module is configured to perform resource requirement parsing, including the following steps: Step A: When the administrator configures the standard operating procedure for the project in the intelligent sample full-process management system, the resource requirements associated with each atomic step of the standard operating procedure are parsed, and the required reagent or consumable list is extracted based on the resource requirements. Step B: Combining the database of the intelligent sample end-to-end management system, the recommended batches of reagents or consumables are obtained through the algorithm configuration module; Step C: Based on the estimated single-step time and total process duration of the standard operating procedure, and combined with the corresponding equipment usage time slot, the device status is synchronized in real time through the IoT sensing and automated execution layer of the intelligent sample full-process management system to dynamically adjust the availability of the reservation. After steps D, B, and C are completed, the system locks the consumable inventory and schedules the corresponding time slot in the equipment calendar; the locking information is synchronized to the intelligent sample end-to-end management system in real time. (2) The algorithm configuration module is configured as follows: A priority-based preemptive resource allocation algorithm is used to calculate the overall priority score, as shown in the following formula: P=w1·T urgency +w2·V sample +w3·S level Where T urgency V is the sample timeliness factor. sample S is the sample rarity factor. level The project strategic level factor is P, which is the comprehensive priority score, with a normalized value range of 0 to 1. w1, w2, and w3 are the weights, satisfying ∑w i =1; The priority of obtaining reagents, consumables or equipment for the project is determined by the ranking of the comprehensive priority score, and the recommended batch of reagents or consumables is obtained. (3) The dynamic inventory early warning and replenishment module is configured as follows: The system dynamically monitors the inventory status of consumables in the inventory database of the intelligent sample full-process management system, updates the consumables status to the administrator in real time, and makes predictions, alarms, and notifications to the administrator based on the consumables status.

2. The intelligent sample dynamic management system according to claim 1, characterized in that, The list of reagents or consumables includes: CAT number, specifications, single dosage, and compatible equipment model.

3. The intelligent sample dynamic management system according to claim 1, characterized in that, In the algorithm configuration module: It reflects the urgency of the maximum allowable time window from sample collection to processing completion; The calculation formula is: in, For the current time, Sampling time, This is the maximum processing time specified in the standard operating procedures; V sample The samples are graded and quantified based on their scarcity, non-renewability, and scientific and clinical value. S level Preset by the administrator in the intelligent sample end-to-end management system; When a high-priority task is submitted, the system will check whether the resources it requires are occupied by low-priority tasks; if so, the system can automatically negotiate to release them or trigger an alert to notify the administrator to make a manual decision.

4. The intelligent sample dynamic management system according to claim 1, characterized in that, The intelligent sample dynamic management system includes the following modules: Dynamic resource configuration module: Dynamically configures the required resources, coupling the execution of standard operating procedures with resource requirements; Sample quality assessment module: Monitors samples and sample containers in real time and collects background data, and conducts assessments or issues warnings based on changes in sample quality; Intelligent Project and Personnel Management Module: Establishes digital profiles for projects and personnel; Structured Standard Operating Procedures Management Module: Used to convert text-based standard operating procedures into structured data objects that the system can parse and execute; Process quality assessment module: Real-time monitoring of personnel operations and sample processing, collection of background data, and assessment or early warning based on comparison with atomized standard operating procedures.

5. The intelligent sample dynamic management system according to claim 4, characterized in that, The sample quality assessment module includes the following sub-modules: (1) The sample tracking module is configured to track images captured in real time by the AR terminal through an integrated AI recognition engine. The image tracking specifically includes: A. The sample receiving station with integrated RFID reader can track the tags on the sample containers inside the sample box; B. Place the sample container in a pre-defined fixed shooting area and perform image tracking on the sample collected inside the sample container; (2) The quality assessment module is configured as follows: a. Retrieve the matching sample list from the intelligent sample full-process management system and verify the retrieved sample list information against the sample container label information; b. Compare and verify the collected sample images with the sample container labels; c. Evaluate the quality of the collected sample images.

6. An intelligent sample end-to-end management system, characterized in that, The system integrating the intelligent sample dynamic management system according to claim 1 further includes: The multimodal hybrid terminal is configured as the main interactive interface for personnel training and assessment and actual operation. Through the multimodal hybrid terminal, personnel can complete the collection and processing of samples. The IoT sensing and automated execution layer is configured to achieve full-dimensional sensing and automated data collection of samples, environment, and devices during the actual operation phase; and to provide real device status simulation data for virtual-real fusion assessment during the training phase. The network and communication layer is configured to enable communication connections between field devices and cloud / local servers; The central decision-making platform is configured to integrate a data processing module, a behavior perception and intelligent evaluation module, a practical intelligent analysis module, and a knowledge graph and continuous learning module to realize functions such as personnel training, sample tracking, data processing, behavior evaluation, model learning and updating. The remote collaboration and control terminal is configured to enable authorized personnel to remotely manage and collaborate on the entire process of training and practical operation on a multimodal hybrid terminal. Specifically, it includes scenario and project configuration, sample task initiation, training system construction, real-time monitoring, dynamic event injection, proactive intervention and guidance, training assessment and authorization management, as well as review and analysis functions. The digital twin platform layer is configured to build and store digital twin device, sample, and spatial information.

7. An intelligent dynamic sample management method, characterized in that, The intelligent sample dynamic management system according to any one of claims 1-5 realizes dynamic resource allocation, including: Step 1: When the administrator configures the standard operating procedure for a project in the intelligent sample full-process management system, the resource requirements associated with each atomic step of the standard operating procedure are parsed, and the required reagent or consumable list is extracted based on the resource requirements. Step 2: Combine the database of the intelligent sample end-to-end management system to configure and obtain recommended batches of reagents or consumables through algorithms; Step 3: Based on the estimated single-step time and total process duration of the standard operating procedure, and combined with the corresponding equipment usage time slot, the availability of the reservation is dynamically adjusted by synchronizing the equipment status in real time through the IoT sensing and automated execution layer of the intelligent sample full-process management system. After steps 4, 2, and 3 are completed, the system locks the consumable inventory and schedules the corresponding time slot in the equipment calendar; the locking information is synchronized to the intelligent sample end-to-end management system in real time. A priority-based preemptive resource allocation algorithm is used to calculate the overall priority score, as shown in the following formula: P=w1·T urgency +w2·V sample +w3·S level Where T urgency V is the sample timeliness factor. sample S is the sample rarity factor. level The project strategic level factor is P, which is the comprehensive priority score, with a normalized value range of 0 to 1. w1, w2, and w3 are the weights, satisfying ∑w i =1; The priority of obtaining reagents, consumables, or equipment for the project is determined by the ranking of the comprehensive priority score, and the recommended batch of reagents or consumables is obtained.

8. An intelligent sample end-to-end management method, said method being implemented based on the intelligent sample end-to-end management system of claim 5, characterized in that, Includes the following steps: Step 1: Establish the project configuration and training system; Step 2: Conduct personnel training and certification authorization; Step 3: Start the sample task; Step 4: Sample collection and initial quality screening; Step 5: Process the samples; Step 6: Process the sample data; Step 7: Perform full lifecycle tracing of the sample and generate a full lifecycle report of the sample.

9. A computer-readable storage medium, characterized in that: It stores a computer program for implementing the intelligent sample full-process management method as described in claim 8.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the intelligent sample full-process management method as described in claim 8.

Images