Hospital orthopedic consumable inventory management system and method based on behavior recognition

Abstract

The application relates to the technical field of inventory management, in particular to a hospital orthopedic consumable inventory management system and method based on behavior recognition, which realizes real-time monitoring of consumable taking and replenishment through a behavior recognition technology; when medical staff enter a consumable warehouse, the behavior of the medical staff is automatically captured, the action is analyzed in real time through an embedded behavior recognition model, the operation type is recognized, the specific consumable type and quantity taken are determined through association with an RFID tag on the consumable, a consumable operation information group is generated, an operation identifier and a second identifier auxiliary generation factor are generated, a system calculates an identifier review factor, the operation behavior of the medical staff is reviewed through the relationship between the second identifier auxiliary generation factor and the review factor, and inventory information is automatically updated. The application realizes real-time and automatic inventory updating, does not need manual intervention, greatly reduces operation errors, and improves the efficiency and precision of hospital orthopedic consumable management.

Description

Technical Field

[0001] This invention relates to the field of inventory management technology, specifically to a hospital orthopedic consumables inventory management system and method based on behavior recognition. Background Technology

[0002] With the continuous advancement of modern medical technology, hospital consumable management has become a crucial aspect of ensuring the quality of medical services. This is especially true in orthopedic surgery, where various high-precision, high-cost consumables (such as bone screws, plates, and artificial joints) are used frequently and in great variety. The accuracy of inventory management directly affects the smooth progress of surgery and the control of hospital operating costs. However, traditional consumable management methods have many problems, including reliance on manual records, lack of transparency in consumable usage, and delayed inventory updates. These issues lead to low efficiency in the use of consumables and may even result in shortages, waste, or improper use.

[0003] Chinese invention patent CN111145885A discloses a method for managing orthopedic medical consumables. This method includes acquiring surgical site information and an electronic human model; extracting relevant consumable requirements from the electronic human model based on the surgical site information to form a surgical plan package; generating a configuration requirement table based on the surgical plan package; generating an order set based on the configuration requirement table; filtering orders from the order set whose inventory meets the requirements to obtain shippable orders; generating a shipping notification based on the shippable orders to enable distributors to ship the goods; generating relevant traceable information corresponding to the consumables, which are then inspected and accepted by the hospital. This invention enables automated warehouse management, reduces consumable configuration costs, improves the accuracy of orthopedic consumable procurement, and increases procurement efficiency.

[0004] However, actual inventory updates often lag behind actual operations, making real-time monitoring impossible and leading to untimely inventory information. In complex scenarios, such as multiple people simultaneously taking or returning consumables, it is difficult to effectively distinguish and track each operation, resulting in unclear operation records and consequently affecting the overall efficiency and security of hospital consumable management. Summary of the Invention

[0005] The purpose of this invention is to address the problems existing in the background technology by proposing a hospital orthopedic consumables inventory management system and method based on behavior recognition.

[0006] The technical solution of this invention: a method for managing hospital orthopedic consumables inventory based on behavior recognition, comprising the following specific implementation steps:

[0007] S1. During the system initialization phase, the inventory management module generates inventory list review information, constructs inventory management review information group, counts the inventory of hospital orthopedic consumables, enters information on various consumables used in the hospital's orthopedics department, equips consumables with dedicated RFID tags, and sets minimum thresholds for various consumables.

[0008] S2. Collect actual operation video data of orthopedic surgeons, nurses and other medical staff when picking up and replenishing consumables, and perform fine annotation on these videos to construct a dataset. Then, build a behavior recognition model based on YOLOv5n network, C3Ghost network and SE attention mechanism.

[0009] S3. During the inventory management phase, when any medical staff member registers their identity at the entrance of the consumables inventory area and enters the consumables inventory area, the behavior data collection module begins to capture their behavior through the camera installed in the consumables inventory area, and transmits the captured video data to the behavior recognition module and the inventory management log library in real time.

[0010] S4. The behavior recognition module analyzes the body movements of people in the video through a pre-trained behavior recognition model, determines the type of behavior data of medical staff, and triggers the RFID scanner in the behavior data acquisition module to read the tag information of consumables in the hands of medical staff. It also generates operation identifiers for inventory list review information, constructs operation information groups, and then transmits the constructed operation information groups to the inventory management module.

[0011] S5. The inventory management module reviews the operation information group, determines the type of personnel operation, confirms the quantity of consumables taken or replenished, automatically updates the consumable inventory, and records the confirmed information group {consumable name, replenishment quantity or quantity taken, operator identification, operation time} in the inventory management log library.

[0012] S6. The inventory management module monitors consumable inventory data in real time. If the inventory of a certain type of consumable approaches the set minimum threshold, the inventory management module will automatically remind the management personnel, generate replenishment suggestions, and prompt that replenishment is needed.

[0013] Preferably, the specific implementation steps for constructing the inventory management audit information group in the inventory management module are as follows:

[0014] S2-1. The inventory management module selects a prime number p=521, which makes n=2. p -1 is a Mersenne prime, and the inventory management module then performs the following operations:

[0015] S2-2, Calculate the p-th degree irreducible primitive polynomial q=x 521 +x 32 +1;

[0016] Where x is a polynomial indefinite variable;

[0017] S2-3, Randomly select an element a∈ , satisfying a n mod q = 1;

[0018] in, For a finite field The multiplication group;

[0019] S2-4. Select the hash function H();

[0020] S2-5. Define the operation type: the fetch operation is defined as 0x00, and the supplement operation is defined as 0x01.

[0021] S2-6. Generate inventory list and review information m={0,1} * And publicly disclose the inventory management audit information group {p, q, H, a, m, type}.

[0022] Preferably, the operation information group is divided into a consumable dispensing information group and a consumable replenishment information group.

[0023] Preferably, the process for constructing the operation information group is as follows:

[0024] S4-1, The behavior recognition module randomly selects a number k. i ∈Z n * ={1, 2, ..., n-1}, calculate the first identifier auxiliary generation factor AcⅠfactor i The calculation method is as follows:

[0025] ;

[0026] S4-2, Calculate the second identifier auxiliary generation factor AcⅡfactor i =H(AcⅠfactor i ||m);

[0027] Where || represents string concatenation;

[0028] S4-3, Select a random number x i ∈Z n * Calculate the identifier recognition factor y i :

[0029] ;

[0030] S4-4, Calculation Operation Identifier AcID i =(k i +Aufactor i ×x i ) mod n;

[0031] S4-5, Generate operation information group {operation type} i Operation ID AcID i The second identifier auxiliary generation factor AcⅡfactor iIdentification factor y i Consumable name, quantity used or replenished (num), operator identification ID i Operation time T i}

[0032] Preferably, the implementation process for the inventory management module's audit operation information group is as follows:

[0033] S5-1, From the received operation information group {operation type} i Operation ID AcID i The second identifier auxiliary generation factor AcⅡfactor i Identification factor y i Consumable name, quantity used or replenished (num), operator identification ID i Operation time T i Extract the parameters from the list:

[0034] operation type i Operation Identifier AcID i Second identifier auxiliary generation factor AcⅡfactor i Identification factor y i Consumable name, quantity used or replenished (num), operator identification ID i Operation time T i ;

[0035] S5-2, Calculate the AcAfactor (AcAfactor) – Auxiliary Audit Factor i :

[0036] ;

[0037] S5-3, Calculate the Aufactor (Identification Factor) i =H(AcAfactor i ||m);

[0038] S5-4, If the equation Aufactor is true i =AcⅡfactor i Once established, the inventory management module confirms the identity of the medical staff member. i The operational behavior.

[0039] Preferably, the behavior detection model replaces the C3 module in the YOLOv5n backbone network with the C3Ghost module, and adds an SE attention module before the first C3 module in the feature fusion network part.

[0040] The preferred early warning and automatic replenishment process for high-frequency orthopedic consumables is as follows:

[0041] S7-1. Construct a behavior-driven consumable consumption prediction model. By performing time-series statistics on the usage behavior of medical staff, construct a consumable usage time series, denoted as: ;

[0042] Among them, U t B represents the predicted consumption of consumables at time t; t Represents behavioral event traffic; H t This represents the actual consumption data of consumables in adjacent historical time periods; P t The intensity of the surgical plan is represented by f(); f() represents the weighted prediction function, which is the LSTM model.

[0043] S7-2, Automatically calculate dynamic safety stock level S s : ;

[0044] Among them, S s Z represents the lower limit of safety stock; Z represents the set service level factor. U represents the standard deviation of consumable demand. t Variance estimation of the sequence; L represents the replenishment lead time;

[0045] S7-3. Update the real-time inventory level I after each use of consumables. t And compare it with the dynamic threshold S. s If the real-time inventory level I t ≤Dynamic threshold S s This will trigger a consumables warning;

[0046] S7-4. When the warning is triggered, automatically calculate the recommended replenishment quantity Q: ;

[0047] in, This represents the average usage over the past N periods; Indicates the regulating factor; This represents the standard deviation of the demand per unit time in the consumable consumption prediction model.

[0048] The technical solution of this invention: A hospital orthopedic consumables inventory management system based on behavior recognition, which is used to execute the aforementioned hospital orthopedic consumables inventory management method based on behavior recognition, comprising:

[0049] The behavioral data acquisition module is equipped with a high-definition camera and an RFID scanner to collect personnel operation data.

[0050] The behavior recognition module uses computer vision algorithms to identify human operational behaviors;

[0051] The inventory management module is used to monitor and update the inventory information of consumables in real time.

[0052] The inventory management log repository is used to store data generated during the management of orthopedic consumables in the hospital.

[0053] The visualization platform is used to display the real-time status of the hospital's orthopedic consumables inventory area, and to graphically show the current inventory change trend and display key inventory warning information.

[0054] Preferably, the inventory management module is linked to the electronic health record system and the supply chain management system, thereby completing the matching of surgical scheduling with consumable demand, dynamic adjustment of inventory strategies, association of consumable usage records with patient files, intelligent procurement, and automatic replenishment.

[0055] Compared with the prior art, the above-mentioned technical solution of the present invention has the following beneficial technical effects:

[0056] This invention designs a hospital orthopedic consumables inventory management system and method based on behavior recognition. The method uses behavior recognition technology to achieve real-time monitoring of consumable retrieval and replenishment: when medical staff enter the consumables warehouse to perform operations, cameras automatically capture their behavior. The system analyzes the actions in real time using a built-in behavior recognition model, identifying the operation type based on YOLOv5n networks, C3Ghost networks, and SE attention mechanisms. It then associates the operation with RFID tags on the consumables to determine the specific type and quantity of consumables retrieved. Subsequently, it generates a consumable operation information group, generates an operation identifier and a second identifier auxiliary generation factor. The system calculates an identifier verification factor based on the operation identifier and verifies the medical staff's operation behavior through the relationship between the second identifier auxiliary generation factor and the verification factor. Furthermore, it integrates behavior recognition data flow with multi-dimensional inventory dynamic modeling, introducing behavior flow Bt as a direct driving factor. The parameters are dynamically updated based on behavior recognition feedback-driven prediction. This significantly enhances the "sensitivity and adaptability" of safety stock by introducing a volatility adjustment term. It also incorporates an over-limit protection mechanism in case of supply instability, making the replenishment algorithm "scenario-adaptive," enabling real-time and automatic inventory updates without manual intervention, greatly reducing operational errors, and improving the efficiency and accuracy of hospital orthopedic consumable management. Attached Figure Description

[0057] Figure 1 This is an architecture diagram of a hospital orthopedic consumables inventory management system based on behavior recognition proposed in this invention;

[0058] Figure 2 This is a network structure diagram of the behavior recognition model proposed in this invention;

[0059] Figure 3 Here is the network structure diagram of the SE module;

[0060] Figure 4 This is a flowchart of the compression and excitation operations in the SE module. Detailed Implementation

[0061] Example 1, as Figure 1 As shown, the present invention proposes a hospital orthopedic consumables inventory management system based on behavior recognition, comprising: a behavior data acquisition module, a behavior recognition module, an inventory management module, an inventory management log library, and a visualization platform.

[0062] The behavioral data acquisition module is equipped with a high-definition camera and an RFID scanner for collecting personnel operation data.

[0063] High-definition cameras are used to capture people's actions of storing and retrieving goods;

[0064] RFID scanners are used to scan consumable information;

[0065] The behavior recognition module uses computer vision algorithms to identify human operational behaviors, including but not limited to: picking up consumables and registering them in the warehouse;

[0066] The inventory management module updates the inventory information of consumables in real time, and automatically updates the inventory quantity whenever it detects that medical staff have taken or replenished the supplies.

[0067] The inventory management log repository is used to store data generated during the management of orthopedic consumables in the hospital.

[0068] The visualization platform is used to display the real-time status of the hospital's orthopedic consumables inventory area, and to graphically show the current inventory change trend and display key inventory warning information.

[0069] Example 2: The present invention proposes a hospital orthopedic consumables inventory management method based on behavior recognition, which is applicable to the hospital orthopedic consumables inventory management system based on behavior recognition proposed in Example 1. Specifically, it includes the following implementation steps:

[0070] System initialization phase:

[0071] S1. Conduct environmental surveys of key areas in the consumables inventory area to determine the installation locations of cameras and RFID scanners, ensuring the accuracy of behavior recognition.

[0072] S2. Install the camera in a location that can fully cover the consumables storage area, such as directly above the consumables storage rack. The RFID scanner can be installed near the consumables storage area. When wiring, ensure a stable connection between the power supply and data cable and the system host.

[0073] S3. Connect hardware devices such as cameras and RFID scanners to the hospital's internal network to ensure that the devices can transmit data to the behavior recognition module in real time. Perform preliminary equipment debugging to ensure that the clarity of the video stream and the sensitivity of the RFID scanner meet the requirements.

[0074] S4. The inventory management module selects a prime number p=521, which makes n=2. p -1 is a Mersenne prime, and the inventory management module then performs the following operations:

[0075] (1) Calculate the irreducible primitive polynomial q=x of degree p. 521 +x 32 +1;

[0076] Where x is a polynomial indefinite variable;

[0077] (2) Randomly select an element a∈ , satisfying a n mod q = 1;

[0078] in, For a finite field The multiplication group;

[0079] (3) Select the hash function H();

[0080] (4) Define the operation type: the fetch operation is defined as 0x00, and the supplement operation is defined as 0x01;

[0081] (5) Generate inventory list review information m={0,1} * And publicly disclose the inventory management audit information group {p, q, H, a, m, type}.

[0082] S5. The inventory management module takes stock of the hospital's orthopedic consumables inventory and performs the following operations:

[0083] (1) Enter information on various consumables used in the hospital's orthopedics department, including but not limited to the type, specifications, inventory, production batch and expiration date of the consumables;

[0084] (2) Equip consumables with dedicated RFID tags. The tag content must contain the unique identification information of the consumables to ensure that the system can identify the consumable information by recognizing the tag each time it is used or replenished;

[0085] (3) Verify that the RFID scanner can accurately read the dedicated RFID tag of the consumable and match it with the consumable information in the system to ensure that the consumable tag is consistent with the system record;

[0086] (4) Set the minimum threshold for various consumables.

[0087] Behavior recognition model training and deployment phase:

[0088] S1. Behavioral Data Collection and Annotation: Collect video data of actual operations performed by orthopedic surgeons, nurses, and other medical personnel when retrieving and replenishing consumables, and annotate these videos in detail. The annotations include the specific types of operations, including but not limited to "retrieving consumables", "replenishing consumables", and "moving and replenishing supplies".

[0089] S2. Training the behavior recognition model: Based on the labeled dataset, use deep learning algorithms to train the behavior recognition model. The model needs to be able to accurately recognize actions in different scenarios and also have the ability to adapt to different operating styles.

[0090] S3. Model Testing and Optimization: Test the trained behavior detection model in real-world scenarios to identify actions such as picking up and replenishing goods. If the model's recognition accuracy is not high, it needs to be optimized by increasing training data or adjusting model parameters.

[0091] It should be noted that, as Figure 2 As shown, the behavior detection model introduces an SE (Squeeze-and-Excitation) module before the first C3 module in the feature fusion network part based on the YOLOv5n model, and replaces the C3 module in the YOLOv5n backbone network with the C3Ghost module.

[0092] In the diagram, Conv represents the convolution operation, C3Ghost is a lightweight convolutional neural network structure composed of a C3 structure and a GhostConv module. C3 convolution is a convolutional module that combines depthwise separable convolution and dilated convolution. Concat represents the concatenation operation, Upsample represents the upsampling operation, SE represents SE attention, and the Detect module completes the classification and detection of human behavior in the operation video data through multi-scale object detection, anchor box and prediction box generation, class prediction and confidence evaluation.

[0093] It should be noted that, as Figure 3As shown, the SE module consists of two main parts: Squeeze and Excitation. The first step is compression, which involves taking a W×H×C feature map as input, where W is the width, H is the height, and C is the number of channels. An average pooling operation compresses the feature map into a 1×1×C vector. The second step is activation, which consists of two fully connected layers and two activation functions. The first fully connected layer has C×SERatio neurons. The input 1×1×C becomes 1×1×C×SERatio after passing through the first fully connected layer, where SERatio is the scaling parameter. This operation reduces the number of channels, thus reducing computation. The second fully connected layer has 1×1×C parameters, transforming the input back into 1×1×C. The final step is scaling, which multiplies the weights. The weights calculated by the SE module are multiplied by the corresponding channels of the original feature map to obtain the final output. The compression and activation operations are as follows: Figure 4 As shown.

[0094] Inventory management phase:

[0095] S1, when any medical personnel person i Register your personal identification ID at the entrance to the consumables inventory area. i When the device enters the consumables inventory area, the behavior data acquisition module begins to capture its behavior through cameras installed in the consumables inventory area, and transmits the captured video data to the behavior recognition module and inventory management log library in real time.

[0096] S2. The behavior recognition module analyzes the body movements of people in the video using a pre-trained behavior recognition model to determine whether medical personnel are performing operations such as "retrieving consumables" or "replenishing consumables." The specific detection process is as follows:

[0097] S2-1. When a medical staff member reaches out to the consumables shelf and picks up a certain type of consumable, the behavior recognition model will identify the staff member's posture, track the staff member's movement, and identify the "taking" behavior based on the staff member's arm extension and grasping movements.

[0098] For example, if a medical staff member takes a box from the consumables shelf, the system can track the entire process of taking the box, ensuring that all details are recorded;

[0099] S2-2. When the behavior recognition model recognizes the "take" action, the behavior recognition module triggers the RFID scanner in the behavior data acquisition module to read the tag information of the consumables in the hands of medical staff, including but not limited to the name and quantity of the consumables.

[0100] S2-3, The behavior recognition module generates an operation identifier AcID for the inventory list review information m. iGenerate a consumables dispensing information group;

[0101] The process by which the behavior recognition module generates the consumable dispensing information group is as follows:

[0102] (1) Randomly select a number k i ∈Z n * ={1, 2, ..., n-1}, calculate the first identifier auxiliary generation factor AcⅠfactor i The calculation method is as follows:

[0103] ;

[0104] (2) Calculate the second identifier auxiliary generation factor AcⅡfactor i =H(AcⅠfactor i ||m);

[0105] Where || represents string concatenation;

[0106] (3) Select a random number x i ∈Z n * Calculate the identifier recognition factor y i :

[0107] ;

[0108] (4) Calculate the operation identifier AcID i =(k i +Aufactor i ×x i ) mod n;

[0109] (5) Generate consumables dispensing information group {operation type} i Operation ID AcID i The second identifier auxiliary generation factor AcⅡfactor i Identification factor y i Consumable name, quantity (-num), operator ID i Operation time T i};

[0110] S2-4. When the behavior recognition module detects the medical staff's retrieval behavior and confirms the specific consumable information, the behavior recognition module groups the consumable retrieval information into {Operation Identifier AcID}. i The second identifier auxiliary generation factor AcⅡfactor i Identification factor y iConsumable name, quantity (-num), operator ID i Operation time T i The data is then transferred to the inventory management module.

[0111] It should be noted that the replenishment action recognition is similar to the retrieval behavior. When medical staff replenish consumables to the warehouse or storage area, the behavior recognition model automatically identifies the "replenishment" action characteristics by capturing the personnel's behavior action data through the camera. After the "replenishment" action is identified, the behavior recognition module reads the replenished consumable tag through the RFID scanner, confirms the name and quantity of the replenished consumables, generates a consumable replenishment information group, and transmits the consumable replenishment information group to the inventory management module.

[0112] The generation method for the above-mentioned consumable replenishment information group is similar to that for the generation method of consumable dispensing information group, and the specific implementation process is as follows:

[0113] (1) Randomly select a number k i ∈Z n * ={1, 2, ..., n-1}, calculate the first identifier auxiliary generation factor AcⅠfactor i The calculation method is as follows:

[0114] ;

[0115] (2) Calculate the second identifier auxiliary generation factor AcⅡfactor i =H(AcⅠfactor i ||m);

[0116] (3) Select a random number x i ∈Z n * Calculate the identifier recognition factor y i :

[0117] ;

[0118] (4) Calculate the operation identifier AcID i =(k i +Aufactor i ×x i ) mod n;

[0119] (5) Generate consumables replenishment information group {operation type} i Operation ID AcID i The second identifier auxiliary generation factor AcⅡfactor i Identification factor y iConsumable name, replenishment quantity (+num), operator identification ID i Operation time T i};

[0120] It should be noted that in some complex scenarios, multiple medical personnel may enter the consumables warehouse at the same time and take different consumables. The behavior recognition module uses multi-target tracking technology to identify multiple operators at the same time and create independent records for each person's behavior. Videos captured by the camera from different angles are merged into the behavior recognition model, thereby accurately distinguishing the actions of different personnel.

[0121] S3. Upon receiving the consumables usage information group and the consumables replenishment information group, the inventory management module performs the following operations:

[0122] S3-1, Retrieve consumable information group {operation type} i Operation ID AcID i The second identifier auxiliary generation factor AcⅡfactor i Identification factor y i Consumable name, quantity (-num), operator ID i Operation time T i} or Consumables Supplement Information Group {Operation Type} i Operation ID AcID i The second identifier auxiliary generation factor AcⅡfactor i Identification factor y i Consumable name, replenishment quantity (+num), operator identification ID i Operation time T i Extract the parameters from the list:

[0123] Operation ID AcID i Second identifier auxiliary generation factor AcⅡfactor i Identification factor y i Consumable name, quantity used (-num) or replenishment quantity (+num), operator identification ID i Operation time T i ;

[0124] S3-2, Calculate the AcAfactor (AcA Factor) – Auxiliary Audit Factor i :

[0125] ;

[0126] S3-3, Calculate the Aufactor (Identification Factor) i=H(AcAfactor i ||m);

[0127] S3-4, If the equation Aufactor is true i =AcⅡfactor i Once established, the inventory management module confirms the identity of the medical staff member. i Retrieve or replenish consumables and identify the type of operation:

[0128] (1) If the operation type is type i =0x00, then the operation type is "retrieval" and the retrieved quantity is num. The inventory management module automatically deducts the corresponding num quantity of consumables based on the consumable name and retrieved quantity, updates the inventory data in real time, and groups the consumable retrieval confirmation information into {operation type}. i Consumable name, quantity (-num), operator ID i Operation time T i Record this in the inventory management log database;

[0129] (2) If the operation type is type i =0x01, then the operation type is replenishment and the replenishment quantity is num. The inventory management module updates the inventory record according to the type and quantity of replenished consumables, adds num to the quantity of that type of consumable in the inventory, and adds the replenishment confirmation information group {operation type}. i Consumable name, replenishment quantity (+num), operator identification ID i Operation time T i Record it in the inventory management log database.

[0130] S4, the inventory management module, is based on an intelligent early warning and automatic replenishment mechanism. It integrates behavioral recognition data streams and multi-dimensional dynamic inventory modeling. Through a time-series prediction model driven by consumable usage behavior, combined with a multi-factor weighted safety stock early warning algorithm, it achieves accurate early warning and automated replenishment control for high-frequency orthopedic consumables. Specifically:

[0131] S4-1. Construct a behavior-driven consumable consumption prediction model. By performing time-series statistics on the consumption behavior of medical staff ("consumption behavior" identified by the behavior recognition module), construct a consumable usage time series, denoted as: ;

[0132] Among them, U t B represents the predicted consumption of consumables at time t; t H represents the flow of behavioral events (the number of times consumables are retrieved per unit of time); tThis represents the actual consumption data of consumables in adjacent historical time periods (smoothed statistics within the time window); P t represents the intensity of the surgical plan; f() represents the weighted prediction function, which is an LSTM model in this embodiment;

[0133] S4-2, Automatically calculate dynamic safety stock level S s : ;

[0134] Among them, S s Z represents the minimum safety stock level; Z represents the service level factor (selected based on the set stockout risk tolerance, e.g., Z=1.65 corresponds to a 95% service level). This represents the standard deviation of consumable demand (source: U). t (Variance estimation of the sequence); L represents the replenishment lead time (determined by the average response time of the consumable supplier);

[0135] S4-3. Update the real-time inventory level I after each use of consumables. t And compare it with the dynamic threshold S. s If the real-time inventory level I t ≤Dynamic threshold S s This will trigger a consumables warning;

[0136] S4-4. When the warning is triggered, automatically calculate the recommended replenishment quantity Q: ;

[0137] in, This represents the average usage over the past N periods; This represents the adjustment factor (dynamically adjusted based on historical stockout rates, initially set to 1). This represents the standard deviation of the demand per unit time in the consumable consumption prediction model.

[0138] S5. The visualization platform retrieves the consumables usage confirmation information group or consumables replenishment confirmation information group from the inventory management log library and displays the medical staff's personal information. i This operation data can be used to display personnel behavior data.

[0139] Example 3: A hospital orthopedic consumables inventory management system based on behavior recognition. Unlike Example 2, this system integrates the inventory management module with the Electronic Health Record (EHR) system and the supply chain management system. The integrated system functions are as follows:

[0140] (1) Matching surgical schedules with consumable needs:

[0141] The EHR system records the schedule of each surgery and the specific patient's condition information. The inventory management module automatically generates a list of consumables required for the surgery based on the above data. If the consumables actually used during the surgery do not match the list, the hospital orthopedic consumables inventory management system based on behavior recognition will automatically adjust the inventory consumption and update the prediction model.

[0142] For example, when a patient is about to undergo fracture repair surgery, the hospital orthopedic consumables inventory management system based on behavior recognition automatically matches the required orthopedic consumables (including but not limited to plates, screws, and scalpel blades) according to the surgical plan and standard operating procedures recorded in the EHR, and locks the inventory in advance.

[0143] (2) Dynamically adjust inventory strategy:

[0144] By obtaining the surgical schedule from the EHR system in real time, the inventory management module dynamically adjusts the inventory strategy of consumables based on the surgical arrangements for the next few days. For some emergency surgical arrangements, the inventory management module can adjust the priority order of consumables in a timely manner to ensure that the consumables required for emergency surgeries are not consumed.

[0145] For example, the inventory management module determines that a large number of orthopedic screws or surgical instruments will be needed in the next week based on the schedule, and will prioritize ensuring sufficient inventory of these consumables to reduce the risk of consumable shortages.

[0146] (3) Consumable usage records are linked to patient health records:

[0147] The inventory management module automatically links the usage of consumables for each surgery with the patient's health record;

[0148] For example, information such as the specific type of steel plate and screws used during surgery can be automatically recorded in the patient's health record for future follow-up. If the patient needs subsequent surgery or rehabilitation examination, the doctor can quickly learn about the consumables used through the EHR system to ensure the continuity of the medical process.

[0149] (4) Intelligent procurement and automatic replenishment:

[0150] The supply chain management system is responsible for the hospital's procurement, supplier management, and order processing. It integrates the inventory management module with the supply chain system to automate the entire process of consumables from demand forecasting to procurement and warehousing, thereby improving efficiency and reducing errors caused by manual operation.

[0151] When the inventory management module detects that a certain type of consumable is about to run out, it can automatically generate a replenishment plan and send a purchase order to the supplier through the supply chain management system.

[0152] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A method for managing hospital orthopedic consumable inventory based on behavior recognition, characterized in that, The specific implementation steps include the following: S1. During the system initialization phase, the inventory management module generates inventory list review information, constructs inventory management review information group, counts the inventory of hospital orthopedic consumables, enters information on various consumables used in the hospital's orthopedics department, equips consumables with dedicated RFID tags, and sets minimum thresholds for various consumables. The specific implementation steps for constructing the inventory management audit information group in the inventory management module are as follows: A1. The inventory management module selects a prime number p = 521, which makes n = 2 p -1 is a Mersenne prime number, after which the inventory management module implements the following operations: A2, compute the irreducible primitive polynomial of degree p q = x 521 + x 32 + 1; Where x is a polynomial indefinite variable; A3. Randomly select an element a∈ , satisfying a n mod q = 1; in, For a finite field The multiplication group; A4. Select the hash function H(); A5. Define the operation type: the fetch operation is defined as 0x00, and the supplement operation is defined as 0x01. A6, generating inventory audit information m = {0, 1} * and discloses the inventory management audit information group {p, q, H, a, m, type}; S2. Collect actual operation video data of orthopedic surgeons and nurses when picking up and replenishing consumables, and perform fine annotation on these videos to construct a dataset. Then, build a behavior recognition model based on YOLOv5n network, C3Ghost network and SE attention mechanism. S3. During the inventory management phase, when any medical staff member registers their identity at the entrance of the consumables inventory area and enters the consumables inventory area, the behavior data collection module begins to capture their behavior through the camera installed in the consumables inventory area, and transmits the captured video data to the behavior recognition module and the inventory management log library in real time. S4. The behavior recognition module analyzes the body movements of people in the video through a pre-trained behavior recognition model, determines the type of behavior data of medical staff, triggers the RFID scanner to read the tag information of consumables in the hands of medical staff, generates operation identifiers for inventory list review information, constructs operation information groups, and then transmits the constructed operation information groups to the inventory management module. The operation information group is divided into the consumable dispensing information group and the consumable replenishment information group; The process of constructing the operation information group is as follows: B1. The behavior recognition module randomly selects a number k. i ∈Z n * ={1, 2, ..., n-1}, calculate the first identifier auxiliary generation factor AcⅠfactor i The calculation method is as follows: ； B2, compute a second identity-aided generation factor AcII factor i = H(AcII factor i ||m); Where || represents string concatenation; B3. Select a random number x i ∈Z n * Calculate the identifier recognition factor y i : ； B4. Calculation Operation Identifier AcID i =(k i +Aufactor i ×x i ) mod n; B5. Generate operation information group {operation type} i Operation ID AcID i The second identifier auxiliary generation factor AcⅡfactor i Identification factor y i Consumable name, quantity used or replenished (num), operator identification ID i Operation time T i }; S5. The inventory management module reviews the operation information group, determines the type of personnel operation, confirms the quantity of consumables taken or replenished, automatically updates the consumable inventory, and records the confirmed information group {consumable name, replenishment quantity or quantity taken, operator identification, operation time} in the inventory management log library. The implementation process for the inventory management module's audit operation information group is as follows: C1. From the received operation information group {operation type} i Operation ID AcID i The second identifier auxiliary generation factor AcⅡfactor i Identification factor y i Consumable name, quantity used or replenished (num), operator identification ID i Operation time T i Extract the parameters from the list: operation type i Operation Identifier AcID i Second identifier auxiliary generation factor AcⅡfactor i Identification factor y i Consumable name, quantity used or replenished (num), operator identification ID i Operation time T i ; C2. Calculate the AcAfactor (AcAfactor) – an auxiliary audit factor. i : ； C3. Calculate the Aufactor (Identification Factor) i =H(AcAfactor i ||m); C4, if the equation Aufactor i =AcⅡfactor i Once established, the inventory management module confirms the identity of the medical staff member. i Operational behavior; S6, the inventory management module integrates behavioral recognition data flow with multi-dimensional dynamic inventory modeling. Through a time-series prediction model driven by consumable usage behavior, combined with a multi-factor weighted safety stock early warning algorithm, it provides early warning and automatic replenishment for high-frequency orthopedic consumables. The early warning and automatic replenishment process for high-frequency orthopedic consumables is as follows: D1. Construct a behavior-driven consumable consumption prediction model. By performing time-series statistics on the usage behavior of medical staff, construct a consumable usage time series, denoted as: ; Among them, U t B represents the predicted consumption of consumables at time t; t Represents behavioral event traffic; H t This represents the actual consumption data of consumables in adjacent historical time periods; P t The intensity of the surgical plan is represented by f(); f() represents the weighted prediction function, which is the LSTM model. D2. Automatically calculate dynamic safety stock level S s : ; Among them, S s Z represents the lower limit of safety stock; Z represents the set service level factor. U represents the standard deviation of consumable demand. t Variance estimation of the sequence; L represents the replenishment lead time; D3. Update the real-time inventory level I after each use of consumables. t And compare it with the dynamic threshold S s If the real-time inventory level I t ≤Dynamic threshold S s This will trigger a consumables warning; D4. When the warning is triggered, the recommended replenishment quantity Q will be automatically calculated: ; in, This represents the average usage over the past N periods; Indicates the regulating factor; This represents the standard deviation of the demand per unit time in the consumable consumption prediction model.

2. The method for managing hospital orthopedic consumables inventory based on behavior recognition according to claim 1, characterized in that, The behavior detection model replaces the C3 module in the YOLOv5n backbone network with the C3Ghost module, and adds an SE attention module before the first C3 module in the feature fusion network part.

3. A hospital orthopedic consumables inventory management system based on behavior recognition, used to execute the hospital orthopedic consumables inventory management method based on behavior recognition as described in any one of claims 1 to 2, characterized in that, include: The behavioral data acquisition module is equipped with a high-definition camera and an RFID scanner to collect personnel operation data. The behavior recognition module uses computer vision algorithms to identify human operational behaviors; The inventory management module is used to monitor and update the inventory information of consumables in real time. The inventory management log repository is used to store data generated during the management of orthopedic consumables in the hospital. The visualization platform is used to display the real-time status of the hospital's orthopedic consumables inventory area, and to graphically show the current inventory change trend and display key inventory warning information.

4. A hospital orthopedic consumables inventory management system based on behavior recognition according to claim 3, characterized in that, The inventory management module is linked to the electronic health record system and the supply chain management system, thereby enabling the matching of surgical scheduling with consumable demand, dynamic adjustment of inventory strategies, association of consumable usage records with patient files, intelligent procurement, and automatic replenishment.

Images