An intelligent blood collection personnel behavior monitoring system

Abstract

The present invention relates to the field of behavioral monitoring technology and is an intelligent phlebotomist behavior monitoring system. The system specifically comprises: a multimodal data processing module, a key quality parameter extraction module, a quality prediction module, and a quality warning module. The system uses multiple sensors to synchronously collect multi-state raw data on phlebotomists' work behaviors, including blood collection, disinfection, and mixing. Through DTW time series alignment and mutual information verification, a hemolysis quality model is constructed, and hemolysis quality parameters are output. A key quality feature set is constructed through spatiotemporal feature matrix and transfer entropy screening. A knowledge-random forest model is constructed to output work product quality prediction results and a mechanical damage index. The system also calculates real-time hemolysis probability values and dynamically adjusts work product quality warning thresholds. The present invention addresses the problem of uneven work product quality in the prior art due to the lack of phlebotomist behavior monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of behavior monitoring, in particular to an intelligent blood sampling personnel behavior monitoring system. Background Art

[0002] In outdoor blood donation vehicles, blood sample quality is directly related to its usability after receipt at blood stations. However, existing technologies have significant shortcomings in multimodal data fusion, real-time quality assessment, and embedded deployment, resulting in high sample rejection rates. Existing systems often rely on single sensors (such as pressure or visual sensors) for local monitoring, lacking collaborative analysis of multi-dimensional parameters such as puncture angle, vessel displacement, and mixing intensity. Furthermore, hardware clock asynchrony leads to misaligned time series data, making it difficult to accurately capture operational dynamics. For example, phase deviations between peak pressure and vessel displacement can mask abnormalities in vascular elasticity, leading to inaccurate hemolysis risk assessments. Furthermore, existing methods often construct feature sets using static statistics, ignoring the spatiotemporal interactions and causal mechanisms between parameters. For example, the statistical correlation between disinfection coverage and vascular visibility can easily mislead the model into learning false associations, failing to identify the true causal chain from "vascular conditions → operating parameters → sample quality." This results in poor model robustness and increased false positive rates in blood donation scenarios, where individual variability is significant. Furthermore, existing hemolysis risk assessment models are often trained on offline laboratory data and employ fixed threshold warning mechanisms, failing to dynamically respond to real-time variables such as ambient temperature and vascular status. For example, vasoconstriction caused by low temperatures can mask abnormalities in operating force, leading to sample rejection due to hemolysis or coagulation abnormalities after delivery to a blood bank. According to statistics, approximately 35% of sample rejections at blood banks are due to such improper operating procedures, and existing technologies can only identify these issues during laboratory retesting, resulting in wasted resources and a reduced donor experience. Furthermore, the high complexity of existing algorithms makes them difficult to run in real time on embedded devices installed in blood donation vehicles. They also lack the ability to provide real-time feedback during the procedure, limiting the feasibility of rapid sample quality testing in outdoor settings. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to address the problem in the prior art that the quality of work results is uneven due to the lack of behavior monitoring of blood drawers, and propose an intelligent blood drawer behavior monitoring system.

[0004] In order to achieve the above objectives, the present invention provides an intelligent blood sampling personnel behavior monitoring system comprising the following modules:

[0005] Multimodal data processing module, key quality parameter extraction module, quality prediction module and quality early warning module;

[0006] The multimodal data processing module collects multi-state raw data of blood collection, disinfection and mixing of blood collection personnel's working behavior through multi-sensor temporal and spatial synchronization, and generates a time series correlation data set containing vascular elasticity characteristics through DTW time series alignment and mutual information verification;

[0007] The key quality parameter extraction module is used to extract multi-dimensional quality parameters from time series correlation data, build a hemolysis quality model, output hemolysis quality parameters, and construct a key quality feature set through spatiotemporal feature matrix and transfer entropy screening;

[0008] The quality prediction module is used to build a knowledge-random forest model, integrate key quality feature sets into model training, and output work product quality prediction results and mechanical damage index;

[0009] The quality warning module is used to calculate the real-time hemolysis probability value based on edge computing, dynamically adjust the work result quality warning threshold, perform graded quality warning of work results through hardware interaction, and generate work result disposal suggestions.

[0010] Specifically, the multimodal data alignment module is configured with a data acquisition architecture construction unit, a multimodal data acquisition unit, a data preprocessing unit, and a redundant data removal unit;

[0011] Preferably, the data acquisition architecture construction unit is used to run the following strategy:

[0012] A11: By adopting a master-slave clock synchronization architecture, a hardware timer (a high-precision timer built into the STM32H7 series microcontroller with a resolution of 1ns) generates a timestamp with a 1ms interval, and the synchronization pulse signal output by the clock subunit starts the multi-state raw data;

[0013] A12: Implement a timestamp embedding mechanism in the sensor driver layer. Specifically, set the data acquisition frequency of the angle sensor and pressure sensor to 10ms, add the current timestamp to the data frame header, and inject timestamps into the camera and infrared imager through the FPGA.

[0014] A sliding window algorithm is used to calibrate the time stamp deviation received by the data processing center in real time;

[0015] A13: The work behavior of the blood collection personnel is divided into four working states: standby, action, feedback, and end. The switching trigger conditions between the states are set, including: switching to the first working state, switching to the second working state, and switching to the third working state;

[0016] The triggering condition for switching the first working state includes: a falling edge of the pressure sensor data and a change in the blood color detected by the camera;

[0017] The triggering conditions for switching to the second working state include: the feedback state lasts for more than 3 seconds or the pressure sensor data is lower than the threshold of 0.5N;

[0018] The triggering condition for switching to the third working state includes: the blood sampling personnel's behavior enters the end state and all sensor data returns to the standby state;

[0019] A14: In each working state, establish a parameter collection queue, including:

[0020] Behavior onset angle, behavior onset pressure, and behavior pressure fluctuation rate in the action state;

[0021] Blood feedback rate and behavioral feedback pressure in feedback state;

[0022] the rate of change of the behavior end angle and the behavior end pressure in the end state;

[0023] Preferably, the multimodal data acquisition unit collects multi-state raw data of the working behavior of the blood sampling personnel through the intelligent hardware device, including: action (puncture) state data, standby (disinfection) state data, mixing state data and characteristic data of the blood sampling personnel;

[0024] The puncture status data includes: action angle, action pressure (puncture force), action pressure change rate, infrared blood vessel displacement data and action feedback time (blood return time);

[0025] The disinfection status data includes: calculation of disinfection coverage through a visual recognition algorithm;

[0026] The mixing status data includes: mixing times and shaking intensity;

[0027] The characteristic data of the person whose blood is being drawn include: blood vessel visibility;

[0028] The data pre-processing unit is used to run the following strategy:

[0029] A21: Jointly align the pressure curve with the angle sensor data and infrared vascular displacement data, normalize the behavior start angle, behavior pressure, and infrared vascular displacement data, and construct a three-dimensional feature vector. This 3D feature vector is then imported into the difference distance calculation strategy to obtain the difference distance of the 3D feature vector.

[0030] Preferably, the difference distance calculation strategy is specifically as follows:

[0031] ;

[0032] in, Represents the value of the k-th dimension feature at time i and j;

[0033] is the weight of the k-th dimension feature;

[0034] is the standard deviation of historical data of the k-th dimension feature;

[0035] A22: Extract the difference distance of the three-dimensional feature vectors and construct a difference distance matrix. Use the dynamic time warping path search algorithm to find the optimal path from the starting point to the end point.

[0036] It should be noted that the optimal path is the mapping relationship between two time series, which indicates how to minimize the morphological difference between the two series through alignment (stretching / compression);

[0037] A23: Using the optimal path, a time mapping table is established between the behavior start pressure curve and the behavior feedback state. Using a gradient detection algorithm, the local minimum point of the difference distance is found in the pressure sequence after path mapping and marked as the peak moment of the behavior start pressure curve.

[0038] Calculate the vascular elasticity parameters of the person whose blood is being drawn through the time mapping table ;

[0039] ;

[0040] in, is the starting moment of the blood sampling personnel’s behavior feedback state, This is the peak moment of the pressure curve for the phlebotomist's behavior.

[0041] Specifically, the redundant data removal unit is used to execute the following strategy:

[0042] A31: Quantify the correlation between disinfection coverage and vascular visibility using a probabilistic algorithm, and calculate the mutual information value between the two;

[0043] The calculation strategy of the mutual information value is:

[0044] ;

[0045] in, They are disinfection coverage and blood vessel visibility;

[0046] Indicates the time window and The joint probability density of

[0047] Respectively represent the time window , The marginal probability density of ;

[0048] A32: Extract the mutual information value and compare it with the preset correlation threshold. When the mutual information value is greater than the preset correlation threshold, it is determined that the disinfection coverage is highly correlated with the vascular visibility. Before extracting the key quality parameters, redundant parameters are eliminated.

[0049] Specifically, the key quality parameter extraction module includes: a hemolysis quality estimation unit and a key quality feature set screening unit;

[0050] The hemolysis quality estimation unit is configured with the following strategy:

[0051] B11: Construct a red blood cell shear damage prediction model, calculate the shear stress on red blood cells during the mixing operation by the blood collector through the shear stress calculation strategy, input the shear stress into the red blood cell shear damage prediction model to output the cumulative shear damage ;

[0052] Preferably, the shear stress calculation strategy is:

[0053] ;

[0054] in, The shear stress on red blood cells caused by blood collection personnel during mixing operations;

[0055] K is the blood viscosity of the person whose blood is being drawn (characterizing the viscosity characteristics of the fluid); It is the intensity of the shaking during the mixing operation of the blood collector; is the apparent viscosity of blood (related to G); r is the non-Newtonian index;

[0056] Preferably, the output formula of the red blood cell shear injury prediction model is:

[0057] ;

[0058] in, is the cumulative amount of shear damage; is the critical fracture stress;

[0059] is the characteristic function, when hour, is 1, otherwise it is 0;

[0060] B12: The number of times the blood collector's action status , Tourniquet usage time and cumulative shear damage Import the hemolysis quality prediction strategy to estimate the hemolysis quality parameters of the blood collection personnel's work behavior ;

[0061] The hemolysis quality estimation strategy is specifically as follows:

[0062] .

[0063] Specifically, the key quality feature set screening unit is configured with the following strategies:

[0064] B21: Obtain 12-dimensional multi-state raw data on the work behavior of blood drawers, establish a spatiotemporal feature matrix, and extract statistical features of mean, variance, and peak value through a sliding window of 5 seconds to form a feature vector containing temporal dynamic information.

[0065] B22: Analyze each element and hemolysis quality parameter in the 12-dimensional spatiotemporal feature matrix using the transfer entropy algorithm The directional influence between them is used to obtain the transfer entropy sequence, and the multi-state original data corresponding to the first 50% of the transfer entropy sequence are screened as the key quality feature set.

[0066] Specifically, the quality prediction module includes: a priori knowledge embedding unit, a model structure optimization unit and a quality result output unit;

[0067] The prior knowledge embedding unit is configured with the following operating strategy:

[0068] Construct a random forest model. During the construction of the decision tree of the random forest, the information gain calculation function is rewritten by reconstructing the weights. During the growth of the decision tree, the feature prior is split to obtain the knowledge-random forest model.

[0069] The split feature prior includes: when the node meets the condition that the blood vessel visibility is less than 0.3, the behavioral pressure of the blood sampling staff and the number of action states of the blood sampling staff are used as split features;

[0070] Preferably, the rewriting of the information gain calculation function specifically includes: when the input feature of the decision tree node includes blood vessel visibility, and the blood vessel visibility is less than 0.3, reconstructing the weights corresponding to the behavioral pressure of the blood collector and the number of action states of the blood collector to 1.2 times the initial weights;

[0071] The model structure optimization unit is configured with the following operating strategies:

[0072] C21: Add supplementary feature parameters to the key quality features to obtain 10-dimensional feature parameters. The supplementary feature parameters include: the average value of the behavior start angle during the historical work process of the current blood sampling staff, the variance of the behavior start angle, the standard deviation of the oscillation intensity, and the vascular elasticity parameter of the blood sampling staff;

[0073] C22: Expand the input layer of the knowledge-random forest model and add the calculation of the mechanical damage index to the output layer of the knowledge-random forest model;

[0074] The expansion of the input layer includes: dividing the 10-dimensional feature parameters into three feature categories: static, time series and event, and performing feature integration;

[0075] Preferably, the mechanical damage index The calculation strategy is:

[0076] ;

[0077] in, are the pressure damage coefficient and the oscillation change rate coefficient respectively;

[0078] behavioral pressures on phlebotomists;

[0079] Specifically, the quality result output unit is configured with the following operation strategy:

[0080] C31: The spatiotemporal feature vectors in the validation dataset are normalized in the same way as the training data;

[0081] Specifically, the validation dataset includes: true labels including hemolysis probability, coagulation markers, contamination level, and corresponding 10-dimensional spatiotemporal feature vectors;

[0082] C32: The preprocessed feature vector is input into the knowledge-random forest model, and based on the trained parameters and decision rules, the three-dimensional quality index prediction results and mechanical damage index are output.

[0083] Preferably, the three-dimensional quality index prediction results include: hemolysis probability, coagulation markers and contamination level;

[0084] Specifically, the quality warning module includes: a real-time hemolysis probability calculation unit and a treatment suggestion output unit;

[0085] The real-time hemolysis probability calculation unit is used to calculate the real-time hemolysis probability value of the work results under the multi-state work behavior of the blood sampling personnel;

[0086] Preferably, the real-time hemolysis probability value The calculation strategy is:

[0087] ;

[0088] in, The number of actions taken by the blood draw personnel; Deviation of the starting angle of the blood collection personnel's behavior;

[0089] It should be noted that. This demonstrates the amplification effect of rapid entry on the probability of hemolysis under low vascular visibility;

[0090] It should be noted that represents the mixing stability penalty, is the standard deviation of the oscillation intensity. The larger the value, the more unstable the mixing operation of the blood collector is.

[0091] The disposal suggestion output unit is configured with the following operation strategy:

[0092] D11: Extract the real-time hemolysis probability value, the blood vessel visibility and elasticity parameters of the blood sampling personnel, and the behavioral pressure change rate and action state number of the blood sampling personnel;

[0093] D12: Based on D11, conduct quality risk warning for the work results of blood collection personnel, including:

[0094] When the vascular visibility is less than 0.3 and the vascular elasticity parameter is less than 50ms, the warning threshold of the real-time hemolysis probability value is ignored and the blood collection personnel's work results are directly marked as yellow warning. The output is a disposal suggestion to prioritize the work results into the warehouse for quality inspection and increase the red blood cell morphology microscopy.

[0095] When the number of action states of the blood drawer is greater than or equal to 2 and the blood drawer's behavioral pressure change rate is greater than 1.5N / s, the warning threshold of the real-time hemolysis probability value is ignored, a red warning is directly triggered, and a disposal suggestion of re-blood collection is output;

[0096] When the real-time hemolysis probability value is less than 0.3 and there are no spatiotemporal anomaly characteristics, the work results of the blood collection personnel are directly marked as a green warning, and disposal recommendations for normal transportation are output.

[0097] Compared with the prior art, the technical effects of the present invention are as follows:

[0098] The present invention significantly improves the fusion accuracy and reliability of blood sampling operation parameters through synchronous multimodal data acquisition and three-dimensional space-time alignment, overcoming the timing deviation problem of traditional single-sensor monitoring.

[0099] The causal feature screening mechanism based on transferred entropy in the present invention effectively eliminates redundant parameter interference, identifies key causal chains such as "puncture force-vascular response" through dynamic mutual information calculation, and achieves 40% compression of feature dimensions on an embedded platform (STM32H7) while retaining 95% of effective information. The computing efficiency is greatly improved, ensuring efficient quality assessment of the rapid blood collection process. In the outdoor blood donation vehicle scenario, when the vascular elasticity parameter is detected to be lower than 50ms or the pressure mutation rate is >1.5N / s, a graded warning is immediately triggered and a disposal suggestion is generated (such as priority microscopic examination or re-blood collection). Compared with the traditional fixed threshold model, the false alarm rate is reduced, the on-site quality defect detection rate of blood samples is improved, and the rejection rate of blood stations is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0100] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them:

[0101] Figure 1 This is a structural diagram of an intelligent blood sampling personnel behavior monitoring system of the present invention. DETAILED DESCRIPTION

[0102] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0103] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0104] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0105] Example 1:

[0106] like Figure 1 As shown, an intelligent blood sampling personnel behavior monitoring system according to an embodiment of the present invention is as follows: Figure 1 As shown, it includes the following modules:

[0107] Multimodal data processing module, key quality parameter extraction module, quality prediction module and quality early warning module;

[0108] The multimodal data processing module collects multi-state raw data of blood collection, disinfection and mixing of blood collection personnel's working behavior through multi-sensor temporal and spatial synchronization, and generates a time series correlation data set containing vascular elasticity characteristics through DTW time series alignment and mutual information verification;

[0109] The multimodal data alignment module is configured with a data acquisition architecture construction unit, a multimodal data acquisition unit, a data preprocessing unit and a redundant data removal unit;

[0110] Preferably, the data acquisition architecture construction unit is used to run the following strategy:

[0111] A11: By adopting a master-slave clock synchronization architecture, a hardware timer (a high-precision timer built into the STM32H7 series microcontroller with a resolution of 1ns) generates a timestamp with a 1ms interval, and the synchronization pulse signal output by the clock subunit starts the multi-state raw data;

[0112] For example, in this embodiment, each sensor is connected to a synchronous clock signal via the SPI / I2C bus, and the camera (such as IMX219, supporting 120fps high-speed shooting) and the infrared imager (FLIR Lepton 3.5, 9Hz frame rate) use external trigger mode;

[0113] A12: Implement a timestamp embedding mechanism in the sensor driver layer. Specifically, set the data acquisition frequency of the angle sensor and pressure sensor to 10ms, add the current timestamp to the data frame header, and inject timestamps into the camera and infrared imager through the FPGA.

[0114] A sliding window algorithm is used to calibrate the time stamp deviation received by the data processing center in real time;

[0115] A13: The phlebotomist's work behavior is divided into four working states: standby, action (needle insertion), feedback (blood return), and end. The switching trigger conditions between the states are set, including: switching to the first working state, switching to the second working state, and switching to the third working state.

[0116] The triggering condition for switching the first working state includes: a falling edge of the pressure sensor data and a change in the blood color detected by the camera;

[0117] The triggering conditions for switching to the second working state include: the feedback state lasts for more than 3 seconds or the pressure sensor data is lower than the threshold of 0.5N;

[0118] The triggering condition for switching to the third working state includes: the blood sampling personnel's behavior enters the end state and all sensor data returns to the standby state;

[0119] A14: In each working state, establish a parameter collection queue, including:

[0120] Behavior onset angle, behavior onset pressure, and behavior pressure fluctuation rate in the action state;

[0121] Blood feedback rate and behavioral feedback pressure in feedback state;

[0122] the rate of change of the behavior end angle and the behavior end pressure in the end state;

[0123] For example, in this embodiment, it should be noted that, for the work process of the blood collection personnel, the action state is the state of needle insertion, and the action start angle reflects the angle change of the blood collection needle at each moment during the needle insertion process. Accurate angle information is crucial for evaluating the standardization and accuracy of the needle insertion operation; the action start pressure reflects the pressure exerted on the blood collection needle in the state of needle insertion, and the action pressure fluctuation rate reflects the speed of pressure change. In the state of needle insertion, the pressure and its pressure fluctuation rate can be used to determine whether the work behavior (needle insertion) of the blood collection personnel is normal, which helps the blood collection personnel to adjust the operation in time and provides a basis for monitoring the behavior of the blood collection personnel;

[0124] For example, in this embodiment, it should be noted that, for the working process of the blood drawer, the feedback state is the state of blood return, and the blood feedback rate can indicate whether the working behavior of the blood drawer is normal; the behavioral feedback pressure reflects the pressure of the blood drawer needle in the blood vessel, and the change in the pressure is related to factors such as the elasticity of the blood vessel wall and blood flow. By monitoring this data of the blood drawer's working behavior, it is possible to determine whether the blood vessel state and the blood drawer's working behavior are stable, providing a reference for quality assessment;

[0125] For example, in this embodiment, it should be noted that, for the work process of the blood collector, the end state is the state of needle removal, and the behavior end angle rate reflects the stability and standardization of the behavior of the blood collector at the end (needle removal) state. It should be noted that a reasonable angle change rate helps to reduce damage to blood vessels and surrounding tissues. For example, if the angle change rate is too large, it will cause complications such as blood vessel tearing or bleeding. In this embodiment, by monitoring this data, behavioral feedback can be provided for the behavior monitoring of the blood collector; the behavior end pressure reflects the size and change of the resistance encountered by the blood collector at the end (needle removal) state. The change in pressure indicates whether the blood collector's behavior will cause abnormal conditions such as tissue adhesion, which can help the blood collector judge whether the behavior of the end (needle removal) state is smooth.

[0126] Preferably, the multimodal data acquisition unit collects multi-state raw data of the working behavior of the blood sampling personnel through the intelligent hardware device, including: action (puncture) state data, standby (disinfection) state data, mixing state data and characteristic data of the blood sampling personnel;

[0127] The puncture status data includes: action angle, action pressure (puncture force), action pressure change rate, infrared blood vessel displacement data and action feedback time (blood return time);

[0128] The disinfection status data includes: calculation of disinfection coverage through a visual recognition algorithm;

[0129] The mixing status data includes: mixing times and shaking intensity;

[0130] The characteristic data of the person whose blood is being drawn include: blood vessel visibility;

[0131] The data pre-processing unit is used to run the following strategy:

[0132] A21: Jointly align the pressure curve with the angle sensor data and infrared vascular displacement data, normalize the behavior start angle, behavior pressure, and infrared vascular displacement data, and construct a three-dimensional feature vector. This 3D feature vector is then imported into the difference distance calculation strategy to obtain the difference distance of the 3D feature vector.

[0133] Preferably, the difference distance calculation strategy is specifically as follows:

[0134] ;

[0135] in, Represents the value of the k-th dimension feature at time i and j;

[0136] is the weight of the k-th dimension feature;

[0137] For example, in this embodiment, ;

[0138] is the standard deviation of historical data of the k-th dimension feature;

[0139] A22: Extract the difference distance of the three-dimensional feature vectors and construct a difference distance matrix. Use the dynamic time warping (DTW) path search algorithm to find the optimal path from the start point to the end point.

[0140] It should be noted that the optimal path is the mapping relationship between two time series, which indicates how to minimize the morphological difference between the two series through alignment (stretching / compression);

[0141] A23: Using the optimal path, a time mapping table is established between the behavior start pressure curve and the behavior feedback (blood recovery) state. Using a gradient detection algorithm, the local minimum point of the difference distance in the pressure sequence after path mapping is found and marked as the peak moment of the behavior start pressure curve.

[0142] For example, in this embodiment, it should be noted that the local minimum point of the difference distance corresponds to the maximum resistance point of the blood sampling staff's work behavior;

[0143] Calculate the vascular elasticity parameters of the person whose blood is being drawn through the time mapping table ;

[0144] ;

[0145] in, The starting moment of the blood collection personnel's behavior feedback (blood return) state, This is the peak moment of the pressure curve for the phlebotomist's behavior.

[0146] For example, in this embodiment, it should be noted that the DTW in the prior art only aligns a single pressure curve, but in the actual work process of blood collection personnel, there is a dynamic correlation between the behavior start pressure, behavior start angle and vascular displacement, and the vascular conditions of different patients vary greatly, and fixed parameters are prone to misjudgment. Therefore, three-dimensional feature vectors are used for joint alignment, which can identify some special situations of the work of blood collection personnel, such as excessive force but correct angle identification. For example, for blood collection personnel with harder blood vessels, their pressure curve may be normal, but the vascular displacement will be significantly reduced. Joint alignment through three-dimensional feature vectors can provide early warning of the work quality of blood collection personnel.

[0147] The redundant data removal unit is used to execute the following strategy:

[0148] A31: Quantify the correlation between disinfection coverage and vascular visibility using a probabilistic algorithm, and calculate the mutual information value between the two;

[0149] The calculation strategy of the mutual information value is:

[0150] ;

[0151] in, They are disinfection coverage and blood vessel visibility;

[0152] Indicates the time window (30 seconds) and The joint probability density of

[0153] Respectively represent the time window , The marginal probability density of ;

[0154] A32: Extract the mutual information value and compare it with the preset correlation threshold. When the mutual information value is greater than the preset correlation threshold, it is determined that the disinfection coverage is highly correlated with the vascular visibility. Before extracting the key quality parameters, redundant parameters are eliminated.

[0155] For example, in this embodiment, a redundant data elimination strategy is provided. Specifically, when the calculated mutual information value is greater than 0.3, and it is determined that the reflection in the disinfected area has reduced the visibility of the blood vessels, the core feature reflecting the blood vessel condition of the person whose blood is being drawn, namely the blood vessel visibility, is retained first, and the strongly correlated disinfection coverage is discarded, thereby reducing data redundancy and improving the calculation efficiency of the system.

[0156] The key quality parameter extraction module is used to extract multi-dimensional quality parameters from time series correlation data, build a hemolysis quality model, output hemolysis quality parameters, and construct a key quality feature set through spatiotemporal feature matrix and transfer entropy screening;

[0157] The key quality parameter extraction module includes: a hemolysis quality estimation unit and a key quality feature set screening unit;

[0158] The hemolysis quality estimation unit is configured with the following strategy:

[0159] B11: Construct a red blood cell shear damage prediction model, calculate the shear stress on red blood cells during the mixing operation by the blood collector through the shear stress calculation strategy, input the shear stress into the red blood cell shear damage prediction model to output the cumulative shear damage ;

[0160] Preferably, the shear stress calculation strategy is:

[0161] ;

[0162] in, The shear stress on red blood cells caused by blood collection personnel during mixing operations;

[0163] K is the blood viscosity of the person whose blood is being drawn (characterizing the viscosity characteristics of the fluid); It is the intensity of the shaking during the mixing operation of the blood collector; is the apparent viscosity of blood (related to G); r is the non-Newtonian index;

[0164] For example, in this embodiment, ;

[0165] Preferably, the output formula of the red blood cell shear injury prediction model is:

[0166] ;

[0167] in, is the cumulative amount of shear damage; is the critical fracture stress;

[0168] is the characteristic function, when hour, is 1, otherwise it is 0;

[0169] B12: The number of times the blood collector's action status , Tourniquet usage time and cumulative shear damage Import the hemolysis quality prediction strategy to estimate the hemolysis quality parameters of the blood collection personnel's work behavior ;

[0170] The hemolysis quality estimation strategy is specifically as follows:

[0171] .

[0172] The key quality feature set screening unit is configured with the following strategies:

[0173] B21: Obtain 12-dimensional multi-state raw data on the work behavior of blood drawers, establish a spatiotemporal feature matrix, and extract statistical features of mean, variance, and peak value through a sliding window of 5 seconds to form a feature vector containing temporal dynamic information.

[0174] For example, in this embodiment, the 12-dimensional multi-state raw data of the blood collection personnel's work behavior includes: the number of behavioral actions, the starting angle of the behavior, the disinfection coverage rate, the tourniquet use time, the behavioral pressure change rate, the mixing and oscillation intensity; the blood vessel visibility, the infrared blood vessel displacement, the age of the blood collection personnel, the BMI index, the blood collection needle diameter, and the blood collection personnel's working environment temperature;

[0175] B22: Analyze each element and hemolysis quality parameter in the 12-dimensional spatiotemporal feature matrix using the transfer entropy algorithm The directional influence between them is used to obtain the transfer entropy sequence, and the multi-state original data corresponding to the first 50% of the transfer entropy sequence are screened as the key quality feature set.

[0176] For example, in this embodiment, a transfer entropy calculation strategy is provided as follows:

[0177] ;

[0178] It should be noted that t is the current time point; X and Y are two data variables that need to be analyzed for correlation in the 12-dimensional multi-state original data of the work behavior of blood drawers;

[0179] Represents the state of the data variable X at time t; Represent the states of data variable Y at time t and time t+1 respectively;

[0180] is the joint probability, which means and The probability of three states appearing simultaneously;

[0181] is the conditional probability, given the historical state of Y and the current state of X When predicting the state of Y in the future ; It should be noted that if X has a causal effect on Y, the probability should be higher than the predictive ability of using only Y's own history.

[0182] is the conditional probability, indicating that only the historical state of Y is used Predict the future state of Y (i.e., autoregressive forecasting).

[0183] The quality prediction module is used to build a knowledge-random forest model, integrate key quality feature sets into model training, and output work product quality prediction results and mechanical damage index;

[0184] The quality prediction module includes: a priori knowledge embedding unit, a model structure optimization unit and a quality result output unit;

[0185] The prior knowledge embedding unit is configured with the following operating strategy:

[0186] Construct a random forest model. During the construction of the decision tree of the random forest, the information gain calculation function is rewritten by reconstructing the weights. During the growth of the decision tree, the feature prior is split to obtain the knowledge-random forest model.

[0187] The split feature prior includes: when the node meets the condition that the blood vessel visibility is less than 0.3, the behavioral pressure of the blood sampling staff and the number of action states of the blood sampling staff are used as split features;

[0188] Preferably, the rewriting of the information gain calculation function specifically includes: when the input feature of the decision tree node includes blood vessel visibility, and the blood vessel visibility is less than 0.3, reconstructing the weights corresponding to the behavioral pressure of the blood collector and the number of action states (puncture times) of the blood collector to 1.2 times the initial weights;

[0189] The model structure optimization unit is configured with the following operating strategies:

[0190] C21: Add supplementary feature parameters to the key quality features to obtain 10-dimensional feature parameters. The supplementary feature parameters include: the average value of the behavior start angle during the historical work process of the current blood sampling staff, the variance of the behavior start angle, the standard deviation of the oscillation intensity, and the vascular elasticity parameter of the blood sampling staff;

[0191] C22: Expand the input layer of the knowledge-random forest model and add the calculation of the mechanical damage index to the output layer of the knowledge-random forest model;

[0192] The expansion of the input layer includes: dividing the 10-dimensional feature parameters into three feature categories: static, time series and event, and performing feature integration;

[0193] Preferably, the mechanical damage index The calculation strategy is:

[0194] ;

[0195] in, are the pressure damage coefficient and the oscillation change rate coefficient respectively;

[0196] behavioral pressures on phlebotomists;

[0197] For example, in this embodiment, it should be noted that the formula for red blood cell membrane strain energy is: , the strain energy (causing rupture) is proportional to the square of the pressure, so the square of the behavioral pressure is used to characterize nonlinear damage;

[0198] Here, U is the strain energy of the red blood cell membrane, which refers to the elastic potential energy stored in the red blood cell membrane due to deformation (stretching, compression, or shear). This is the energy source that can damage or even rupture the membrane structure. It should be noted that when red blood cells pass through narrow blood vessels (such as capillaries) or are subjected to fluid shear forces, the membrane deforms and accumulates strain energy. If the energy exceeds the membrane's tolerance threshold, the membrane will rupture (e.g., hemolysis).

[0199] is the stiffness coefficient of the red blood cell membrane, which characterizes the elastic properties of the red blood cell membrane and reflects the ability of the membrane to resist deformation; is the relative deformation of the membrane; its relationship to the force per unit area acting on the membrane is , according to Hooke's law, within the elastic range, strain and stress are proportional, so the strain energy (causing rupture) is proportional to the square of the pressure;

[0200] For example, in this embodiment, it should be noted that the rapidly changing shear rate (high ) causes a sudden drop in local viscosity, making cells more easily torn, which will reduce the quality of the blood collection staff's work results;

[0201] The quality warning module is used to calculate the real-time hemolysis probability value based on edge computing, dynamically adjust the work result quality warning threshold, perform graded quality warning of work results through hardware interaction, and generate work result disposal suggestions.

[0202] The quality result output unit is configured with the following operation strategy:

[0203] C31: The spatiotemporal feature vectors in the validation dataset are normalized in the same way as the training data;

[0204] Specifically, the validation dataset includes: true labels including hemolysis probability, coagulation markers, contamination level, and corresponding 10-dimensional spatiotemporal feature vectors;

[0205] C32: The preprocessed feature vector is input into the knowledge-random forest model, and based on the trained parameters and decision rules, the three-dimensional quality index prediction results and mechanical damage index are output.

[0206] Preferably, the three-dimensional quality index prediction results include: hemolysis probability, coagulation markers and contamination level;

[0207] Illustratively, in this embodiment, the labels in the knowledge-random forest model include: hemolysis probability, coagulation markers, and contamination level;

[0208] The hemolysis probability ranges from [0, 1], indicating the possibility of hemolysis occurring in the blood sample;

[0209] The coagulation marker is a binary variable with a value of normal or abnormal, and is used to determine the coagulation status of the blood sample;

[0210] The contamination level is a three-category variable with values of low, medium, and high, reflecting the degree of contamination of the blood sample;

[0211] The quality warning module includes: a real-time hemolysis probability calculation unit and a treatment suggestion output unit;

[0212] The real-time hemolysis probability calculation unit is used to calculate the real-time hemolysis probability value of the work results under the multi-state work behavior of the blood sampling personnel;

[0213] Preferably, the real-time hemolysis probability value The calculation strategy is:

[0214] ;

[0215] in, The number of actions taken by the blood draw personnel; Deviation of the starting angle of the blood collection personnel's behavior;

[0216] It should be noted that. This demonstrates the amplification effect of rapid entry on the probability of hemolysis under low vascular visibility;

[0217] It should be noted that represents the mixing stability penalty, is the standard deviation of the oscillation intensity. The larger the value, the more unstable the mixing operation of the blood collector is.

[0218] The disposal suggestion output unit is configured with the following operation strategy:

[0219] D11: Extract the real-time hemolysis probability value, the blood vessel visibility and elasticity parameters of the blood sampling personnel, and the behavioral pressure change rate and action state number of the blood sampling personnel;

[0220] D12: Based on D11, conduct quality risk warning for the work results of blood collection personnel, including:

[0221] When the vascular visibility is less than 0.3 and the vascular elasticity parameter is less than 50ms, the warning threshold of the real-time hemolysis probability value is ignored and the blood collection personnel's work results are directly marked as yellow warning. The output is a disposal suggestion to prioritize the work results into the warehouse for quality inspection and increase the red blood cell morphology microscopy.

[0222] When the number of action states of the blood drawer is greater than or equal to 2 and the blood drawer's behavioral pressure change rate is greater than 1.5N / s, the warning threshold of the real-time hemolysis probability value is ignored, a red warning is directly triggered, and a disposal suggestion of re-blood collection is output;

[0223] When the real-time hemolysis probability value is less than 0.3 and there are no spatiotemporal anomaly characteristics, the work results of the blood collection personnel are directly marked as a green warning, and disposal recommendations for normal transportation are output.

[0224] Example 2:

[0225] This embodiment provides an electronic device, including: a processor and a memory, wherein the memory stores a computer program that can be called by the processor;

[0226] The processor runs the above-mentioned intelligent blood sampling personnel behavior monitoring system by calling the computer program stored in the memory.

[0227] The electronic device may vary significantly due to different configurations or performance, and may include one or more processors (Central Processing Units, CPUs) and one or more memories, wherein the memories store at least one computer program, which is loaded and executed by the processor to implement an intelligent blood sampling personnel behavior monitoring system provided by the above-mentioned method embodiment. The electronic device may also include other components for implementing the device functions. For example, the electronic device may also have components such as wired or wireless network interfaces and input and output interfaces for data input and output. This embodiment will not be described in detail here.

[0228] Example 3:

[0229] This embodiment provides a computer-readable storage medium having a rewritable computer program stored thereon;

[0230] When the computer program runs on the computer device, the computer device runs the above-mentioned intelligent blood sampling personnel behavior monitoring system.

[0231] For example, computer-readable storage media can be read-only memory (ROM), random access memory (RAM), compact disc read-only memory (CD-ROM), magnetic tape, floppy disk, and optical data storage device.

[0232] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0233] It should be understood that determining B based on A does not mean determining B only based on A. B can also be determined based on A and / or other information.

[0234] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. A computer program product comprises one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the processes or functions according to the embodiments of the present invention are fully or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, computer instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via a wired network or / and a wireless network. The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0235] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0236] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0237] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only one type. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0238] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0239] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0240] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0241] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent blood sampling personnel behavior monitoring system, characterized in that: The system includes: a multimodal data processing module, a key quality parameter extraction module, a quality prediction module and a quality early warning module; The multimodal data processing module collects multi-state raw data of blood collection, disinfection and mixing of blood collection personnel's work behavior through multi-sensor temporal and spatial synchronization, and generates a time series correlation data set through DTW time series alignment and mutual information verification; The key quality parameter extraction module is used to extract multi-dimensional quality parameters from time series correlation data, build a hemolysis quality model, output hemolysis quality parameters, and construct a key quality feature set through spatiotemporal feature matrix and transfer entropy screening; The key quality parameter extraction module includes: a hemolysis quality estimation unit and a key quality feature set screening unit; The hemolysis quality estimation unit is configured with the following strategy: B11: Construct a red blood cell shear damage prediction model, calculate the shear stress on red blood cells during the mixing operation by the blood collector through the shear stress calculation strategy, input the shear stress into the red blood cell shear damage prediction model to output the cumulative shear damage ; B12: The number of times the blood collector's action status , Tourniquet usage time and cumulative shear damage Import the hemolysis quality prediction strategy to estimate the hemolysis quality parameters of the blood collection personnel's work behavior ; The hemolysis quality estimation strategy is specifically as follows: ； in, Tourniquet usage time 1.2 power; The quality prediction module is used to build a knowledge-random forest model, integrate key quality feature sets into model training, and output work product quality prediction results and mechanical damage index; The quality warning module is used to calculate the real-time hemolysis probability value based on edge computing, dynamically adjust the work result quality warning threshold, perform graded quality warning of work results through hardware interaction, and generate work result disposal suggestions.

2. The intelligent blood sampling personnel behavior monitoring system according to claim 1, characterized in that: The multimodal data alignment module is configured with a data acquisition architecture construction unit, a multimodal data acquisition unit, a data preprocessing unit and a redundant data removal unit; The data pre-processing unit is used to run the following strategy: A21: Jointly align the pressure curve with the angle sensor data and infrared vascular displacement data, normalize the behavior start angle, behavior pressure, and infrared vascular displacement data, and construct a three-dimensional feature vector. This 3D feature vector is then imported into the difference distance calculation strategy to obtain the difference distance of the 3D feature vector. A22: Extract the difference distance of the three-dimensional feature vectors and construct a difference distance matrix. Use the dynamic time warping path search algorithm to find the optimal path from the starting point to the end point. A23: Using the optimal path, a time mapping table is established between the behavior start pressure curve and the behavior feedback state. Using a gradient detection algorithm, the local minimum point of the difference distance is found in the pressure sequence after path mapping and marked as the peak moment of the behavior start pressure curve. Calculate the vascular elasticity parameters of the person whose blood is being drawn through the time mapping table ; ； in, is the starting moment of the blood sampling personnel’s behavior feedback state, This is the peak moment of the pressure curve for the phlebotomist's behavior.

3. The intelligent blood sampling personnel behavior monitoring system according to claim 2, characterized in that: The redundant data removal unit is used to execute the following strategy: A31: Quantify the correlation between disinfection coverage and vascular visibility using a probabilistic algorithm, and calculate the mutual information value between the two; The calculation strategy of the mutual information value is: ； in, They are disinfection coverage and blood vessel visibility; Indicates the time window and The joint probability density of Respectively represent the time window , The marginal probability density of ; A32: Extract the mutual information value and compare it with the preset correlation threshold. When the mutual information value is greater than the preset correlation threshold, it is determined that the disinfection coverage is highly correlated with the vascular visibility. Before extracting the key quality parameters, redundant parameters are eliminated.

4. The intelligent blood sampling personnel behavior monitoring system according to claim 3, characterized in that: The key quality feature set screening unit is configured with the following strategies: B21: Obtain 12-dimensional multi-state raw data on the work behavior of blood drawers, establish a spatiotemporal feature matrix, and extract statistical features of mean, variance, and peak value through a sliding window of 5 seconds to form a feature vector containing temporal dynamic information. B22: Analyze each element and hemolysis quality parameter in the 12-dimensional spatiotemporal feature matrix using the transfer entropy algorithm The directional influence between them is used to obtain the transfer entropy sequence, and the multi-state original data corresponding to the first 50% of the transfer entropy sequence are screened as the key quality feature set.

5. The intelligent blood sampling personnel behavior monitoring system according to claim 4, characterized in that: The quality prediction module includes: a priori knowledge embedding unit, a model structure optimization unit and a quality result output unit; The prior knowledge embedding unit is configured with the following operating strategy: Construct a random forest model. During the construction of the decision tree of the random forest, the information gain calculation function is rewritten by reconstructing the weights. During the growth of the decision tree, the feature prior is split to obtain the knowledge-random forest model. The split feature prior includes: when the node meets the condition that the blood vessel visibility is less than 0.3, the behavioral pressure of the blood sampling staff and the number of action states of the blood sampling staff are used as split features; The model structure optimization unit is configured with the following operating strategies: C21: Add supplementary feature parameters to the key quality features to obtain 10-dimensional feature parameters. The supplementary feature parameters include: the average value of the behavior start angle during the historical work process of the current blood sampling staff, the variance of the behavior start angle, the standard deviation of the oscillation intensity, and the vascular elasticity parameter of the blood sampling staff; C22: Expand the input layer of the knowledge-random forest model and add the calculation of the mechanical damage index to the output layer of the knowledge-random forest model; The expansion of the input layer includes: dividing the 10-dimensional feature parameters into three feature categories: static, time series and event, and performing feature integration.

6. The intelligent blood sampling personnel behavior monitoring system according to claim 5, characterized in that: The quality result output unit is configured with the following operation strategy: C31: The spatiotemporal feature vectors in the validation dataset are normalized in the same way as the training data; C32: The preprocessed feature vector is input into the knowledge-random forest model, and based on the trained parameters and decision rules, the three-dimensional quality index prediction results and mechanical damage index are output.

7. The intelligent blood sampling personnel behavior monitoring system according to claim 6, characterized in that: The quality warning module includes: a real-time hemolysis probability calculation unit and a treatment suggestion output unit; The real-time hemolysis probability calculation unit is used to calculate the real-time hemolysis probability value of the work results under the multi-state work behavior of the blood sampling personnel; The disposal suggestion output unit is configured with the following operation strategy: D11: Extract the real-time hemolysis probability value, the blood vessel visibility and elasticity parameters of the blood sampling personnel, and the behavioral pressure change rate and action state number of the blood sampling personnel; D12: Based on D11, conduct quality risk warning for the work results of blood collection personnel, including: When the vascular visibility is less than 0.3 and the vascular elasticity parameter is less than 50ms, the warning threshold of the real-time hemolysis probability value is ignored and the blood collection personnel's work results are directly marked as yellow warning. The output is a disposal suggestion to prioritize the work results into the warehouse for quality inspection and increase the red blood cell morphology microscopy. When the number of action states of the blood drawer is greater than or equal to 2 and the blood drawer's behavioral pressure change rate is greater than 1.5N / s, the warning threshold of the real-time hemolysis probability value is ignored, a red warning is directly triggered, and a disposal suggestion of re-blood collection is output; When the real-time hemolysis probability value is less than 0.3 and there are no spatiotemporal anomaly characteristics, the work results of the blood collection personnel are directly marked as a green warning, and disposal recommendations for normal transportation are output.

Images