Specialized nurse-oriented blood purification personalized course recommendation method and system

Abstract

The invention discloses a specialized nurse-oriented blood purification personalized course recommendation method and system, and relates to the technical field of medical teaching. The method comprises the following steps: S1, collecting blood purification operation data, and preprocessing the blood purification operation data; s2, dividing the preprocessed blood purification operation data into four stages, and calculating corresponding operation characteristic values; S3, obtaining operation capability characterization parameters of each stage, constructing a nurse operation evaluation model, and outputting a blood purification operation evaluation value of a nurse; and S4, generating a personalized learning task package based on the hematopurification operation evaluation value and the weight coefficient of each stage, pushing the personalized learning task package to a nurse side, constructing an evaluation difference matrix before and after learning, and performing closed-loop updating of course recommendation. The problem that an existing nurse training system cannot achieve objective evaluation and precise course matching based on multi-dimensional data of the hematology operation process, and consequently learning content is general and insufficient in individuation is solved.

Description

Technical Field

[0001] This invention relates to the field of medical teaching technology, and in particular to a method and system for recommending personalized blood purification courses for specialist nurses. Background Technology

[0002] Blood purification technology is an important clinical treatment for renal failure, metabolic disorders, and certain poisoning diseases. The treatment process involves several key steps, including puncture, catheter flushing, connection fixation, and on-machine monitoring. The quality of operation at each step directly affects the safety and efficacy of dialysis treatment.

[0003] In blood purification nursing practice, specialist nurses undertake core tasks such as equipment preparation, patient connection, treatment monitoring, and handling of abnormalities. Because the blood purification procedure involves multiple monitoring parameters, the stability, standardization, and aseptic compliance of the procedure all have a significant impact on treatment outcomes.

[0004] Currently, the training of specialist nurses in blood purification operations mainly relies on centralized lectures, experience sharing, and periodic assessments. While this training model has some effectiveness, the following common problems still exist in clinical application: (1) The training content is mainly based on standardized procedures, which makes it difficult to reflect the individual differences of different nurses at each stage of the operation; (2) Operation evaluation mainly relies on manual observation and subjective scoring, lacking objective and quantitative operational data support; (3) Training feedback is often delayed, making it difficult to promptly address the shortcomings of nurses in specific operational steps; (4) Although blood purification equipment can collect multidimensional monitoring data, this data has not yet been effectively used for nursing operation behavior analysis and precision training.

[0005] Therefore, in response to the above problems, there is an urgent need for a method and system for recommending personalized blood purification courses for specialist nurses. Summary of the Invention

[0006] To address the technical problem that existing nurse training systems cannot achieve objective assessment and precise course matching based on multidimensional data from the blood purification procedure, resulting in generalized and insufficiently personalized learning content, this invention provides a method and system for recommending personalized blood purification courses for specialist nurses. The technical solution is as follows:

[0007] On the one hand, a personalized course recommendation method for blood purification for specialist nurses is provided. This method includes: S1, collecting blood purification operation data and performing time stamp uniform correction, noise suppression, anomaly removal, missing data compensation, and standardization on the blood purification operation data to obtain pre-processed blood purification operation data; S2, dividing the pre-processed blood purification operation data into four stages according to the machine log records: puncture, flushing, connection fixation, and machine monitoring, recording the time index of each stage and calculating the corresponding operation feature value; S3, obtaining the operation ability representation parameters of each stage based on the operation feature value, and extracting the historical blood purification operation data of each nurse to construct a nurse operation evaluation model, using the weight coefficients of each stage as input, and outputting the nurse's blood purification operation evaluation value; S4, generating a personalized learning task package based on each nurse's blood purification operation evaluation value and the weight coefficients of each stage and pushing it to the nurse's end, and after the nurse completes the learning task package, constructing the evaluation difference matrix before and after learning, and performing closed-loop update of course recommendations.

[0008] Optionally, blood purification operation data is collected, and the data undergoes time-stamp uniform correction, noise suppression, anomaly removal, missing data compensation, and standardization. The specific steps for obtaining preprocessed blood purification operation data are as follows: Real-time collection of blood purification operation data for each nurse during the blood purification process. This data includes arterial pressure, venous pressure, transmembrane pressure, blood pump flow rate, dialysate conductivity, dialysate temperature, vessel depth, needle-to-forearm angle, needle insertion speed, needle insertion depth, distance to the sterile area, wrist angular velocity, hand linear acceleration, cumulative flushing volume, and flushing pressure. The collected blood purification operation data undergoes time-base uniform correction using a timestamp resampling algorithm to ensure all data are aligned under the same time reference. A weighted moving average filtering algorithm is used for smoothing to suppress high-frequency noise and transient jitter. A 3σ outlier detection algorithm is used to initially remove abnormal sampling points, and a cubic spline interpolation algorithm is used to compensate for missing data. Finally, a Z-Score standardization algorithm is used for standardization to unify the numerical scale and eliminate dimensional differences.

[0009] Optionally, the pre-processed blood purification operation data is divided into four stages according to the machine log records: puncture, flushing, connection fixation, and machine monitoring. The specific steps for recording the time index of each stage and calculating the corresponding operation characteristic values ​​are as follows: The pre-processed blood purification operation data is read in timestamp order. Based on the operation event time identifiers recorded in the machine log and the operation trigger signals for needle insertion, flushing, connection, and machine monitoring, the entire operation process is divided into stages: puncture stage, flushing stage, connection fixation stage, and machine monitoring stage. The time index of each operation stage is recorded, and the operation characteristic values ​​of each operation stage are calculated: For the puncture stage, the time index of each operation stage is recorded, and the operation characteristic values ​​of each operation stage are calculated. Calculate the standard deviation and mean of needle insertion speed; for the flushing stage, calculate the standard deviation and mean of flushing pressure, and count the number of times the flushing pressure exceeds the flushing pressure threshold to obtain the number of flushing pressure exceeding the limit; for the connection fixation stage, calculate the root mean square of wrist angular velocity and hand linear acceleration, extract the minimum distance to the sterile zone, and count the number of times the distance to the sterile zone is lower than the sterile zone distance threshold to obtain the number of sterility exceeding the limit; for the machine monitoring stage, calculate the standard deviation and mean of arterial pressure, venous pressure, transmembrane pressure, blood pump flow rate, dialysate conductivity, and dialysate temperature, and count the number of times arterial pressure, venous pressure, and transmembrane pressure exceed the corresponding safety threshold to obtain the number of monitoring pressure exceeding the limit.

[0010] Optionally, the specific steps for obtaining operational capability representation parameters for each stage based on operational feature values, and constructing a nurse operation assessment model by extracting historical blood purification operation data for each nurse, using the weight coefficients of each stage as input, and outputting the nurse's blood purification operation assessment value are as follows: Extract four capability dimensions from the department's nurse training program: puncture accuracy, flushing standardization, aseptic compliance, and monitoring coordination stability. Weight the operational feature values ​​extracted for each stage and obtain the puncture weight coefficient, flushing weight coefficient, connection fixation weight coefficient, and on-machine monitoring weight coefficient, respectively. Extract historical blood purification operation data for each nurse, construct a nurse operation assessment model based on the gradient boosting tree algorithm, and train the model using standard blood purification data from the department as a supervision label. Input the puncture weight coefficient, flushing weight coefficient, connection fixation weight coefficient, and on-machine monitoring weight coefficient into the trained nurse operation assessment model, and output the blood purification operation assessment value for each nurse.

[0011] Optionally, the specific steps for obtaining the puncture weighting coefficient are as follows: Divide the absolute value of the difference between the angle between the needle body and the forearm and the standard puncture angle by the allowable angle deviation to obtain the normalized angle deviation value; divide the standard deviation of the needle insertion speed by the average needle insertion speed to obtain the speed stability ratio; subtract the vessel depth from the needle insertion depth and divide by the vessel depth to obtain the depth deviation ratio; square the angle deviation normalization value, the speed stability ratio, and the depth deviation ratio respectively, add them together, take the square root of the sum and then take the negative value as the exponent, calculate the power function value with base e to obtain the puncture accuracy attenuation term; divide the sum of the squares of the wrist angular velocity and the hand linear acceleration by the sum of the squares of the reference wrist angular velocity and the reference hand linear acceleration, take the cube root of the ratio, add one and take the reciprocal to obtain the operation stability correction term; multiply the puncture accuracy attenuation term by the operation stability correction term to obtain the puncture weighting coefficient.

[0012] Optionally, the specific steps for obtaining the flushing weight coefficient are as follows: Calculate the difference between the cumulative injection volume of each flush and the standard flush injection volume in the sampling order, and sum the absolute values; divide the sum by the product of the number of samplings in the flushing stage and the standard flush injection volume to obtain the flow deviation term; divide the number of times the flushing pressure exceeds the limit by the number of samplings in the flushing stage to obtain the pressure exceedance rate term; divide the standard deviation of the flushing pressure by the average value of the flushing pressure to obtain the pressure fluctuation ratio term; add the flow deviation term, the pressure exceedance rate term, and the pressure fluctuation ratio term, take the cube root, and then subtract the cube root from one to obtain the flushing weight coefficient.

[0013] Optionally, the specific steps for obtaining the connection fixed weight coefficient are as follows: Divide the minimum distance of the sterile area by the distance of the sterile area in each sampling, sum the resulting ratios and take the average to obtain the sterile area distance deviation term; Divide the number of sterile limit violations by the number of samplings in the connection fixed stage, add the sterile area distance deviation term to obtain the sterile control comprehensive term; Add the root mean square of the wrist angular velocity to the root mean square of the hand linear acceleration, divide by the sum of the reference wrist angular velocity and the reference hand linear acceleration to obtain the attitude fluctuation correction term; Multiply the sterile control comprehensive term by the attitude fluctuation correction term, add one to the product and take the natural logarithm to obtain the logarithmic gain term; Divide the logarithmic gain term by one and add the natural logarithm of e to obtain the normalization coefficient; Subtract the normalization coefficient from one to obtain the connection fixed weight coefficient.

[0014] Optionally, the specific steps for obtaining the on-machine monitoring weight coefficient are as follows: subtract venous pressure from arterial pressure to obtain the blood circuit pressure difference; subtract the ratio of average transmembrane pressure to average blood pump flow rate from the ratio of transmembrane pressure to blood pump flow rate to obtain the filtration flux deviation; multiply the blood circuit pressure difference by the filtration flux deviation and take the absolute value to obtain the pressure-flow coupling deviation term; add the absolute value of the difference between dialysate conductivity and standard dialysate conductivity and the absolute value of the difference between dialysate temperature and standard dialysate temperature to obtain the dialysate stability deviation term; multiply the dialysate stability deviation term by the number of monitoring pressure exceedances plus one, and take the square root of the product to obtain the stability adjustment term; divide the pressure-flow coupling deviation term by the stability adjustment term, take the negative value as the exponent, and calculate the power function value with base e to obtain the on-machine monitoring weight coefficient.

[0015] Optionally, based on each nurse's blood purification operation assessment value and the weight coefficients of each stage, a personalized learning task package is generated and pushed to the nurse's end. After the nurse completes the learning task package, an assessment difference matrix before and after learning is constructed to perform a closed-loop update of course recommendations. The specific steps are as follows: Real-time reading of each nurse's blood purification operation assessment value and the weight coefficients of each stage, and real-time comparison of the blood purification operation assessment value with the operation assessment threshold. When the blood purification operation assessment value is greater than the operation assessment threshold, the deviation ratio of each stage's weight coefficient to the corresponding threshold is calculated, and the operation stage corresponding to the largest deviation ratio is determined as the dominant imbalance stage; automatic retrieval of operation knowledge points and teaching videos associated with the dominant imbalance stage, generating a targeted learning task package and pushing it to the nurse's end; when the blood purification operation assessment value is less than or equal to the operation assessment threshold... The system retrieves operational knowledge points for each stage of the operation and constructs a multi-layered operational knowledge graph based on weight hierarchy, ranked from largest to smallest deviation ratio for each stage. Based on the graph hierarchy, it generates phased reinforcement learning packages, pushing the knowledge point index, instructional videos, and operational procedure demonstrations to the nurse's end. After the nurse completes the learning package, the system re-acquires the corresponding nurse's weight coefficients and blood purification operation assessment values ​​for each stage, and calculates the difference between the weight coefficients and blood purification operation assessment values ​​before and after learning, generating an assessment difference matrix. When the difference in weight coefficients for any stage is greater than or equal to the corresponding improvement threshold, the system updates the weight coefficients and time index for the corresponding stage in the nurse's operational assessment model. When the difference in weight coefficients for any stage is less than the improvement threshold, the system re-selects and matches knowledge points to generate a secondary learning package, which is then pushed to the nurse's end.

[0016] On the other hand, a personalized blood purification course recommendation system for specialist nurses is provided. This system is applied to a personalized blood purification course recommendation method for specialist nurses. The system includes: a data acquisition and preprocessing module, used to collect blood purification operation data and perform timestamp-based unified correction, noise suppression, anomaly removal, missing data compensation, and standardization on the blood purification operation data to obtain preprocessed blood purification operation data; a stage feature extraction module, used to divide the preprocessed blood purification operation data into four stages according to the machine log records: puncture, flushing, connection fixation, and on-machine monitoring, recording the time index of each stage and calculating the corresponding operation feature values; an operation assessment modeling module, used to obtain the operation capability representation parameters of each stage based on the operation feature values, and extract the historical blood purification operation data of each nurse to construct a nurse operation assessment model, using the weight coefficients of each stage as input, and outputting the nurse's blood purification operation assessment value; and a personalized recommendation update module, used to generate personalized learning task packages based on each nurse's blood purification operation assessment value and the weight coefficients of each stage, push them to the nurse's end, and after the nurse completes the learning task package, construct the assessment difference matrix before and after learning to perform closed-loop update of course recommendations.

[0017] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: (1) By combining machine logs and operation signals, the complex blood purification operation is automatically divided into multiple stages, and the core characteristic parameters of each stage are extracted through mathematical models. Based on indicators such as operation stability, standardization and compliance, the stage weight coefficients are calculated to achieve a comprehensive evaluation of multiple links and dimensions.

[0018] (2) By introducing the gradient boosting tree algorithm to construct a nurse operation evaluation model, and comprehensively analyzing the weight coefficients of each stage and historical operation data, the nurse's puncture accuracy, flushing standardization, aseptic compliance and monitoring coordination stability can be automatically evaluated, overcoming the problems of strong subjectivity and insufficient accuracy of traditional manual assessment.

[0019] (3) By analyzing the ratio of blood purification operation assessment values ​​to stage deviations in real time, the system automatically identifies the nurse's dominant imbalance stage, generates individualized learning task packages and multi-layered operation knowledge graphs, and pushes targeted learning content. The system can dynamically adjust the recommendation strategy based on assessment feedback to achieve adaptive optimization of the training process.

[0020] (4) By combining blood purification operation data before and after learning to construct an evaluation difference matrix, a quantitative monitoring model for changes in operational ability was established. The system automatically updates the weight parameters and evaluation model based on the difference results, realizing dynamic verification of training results and generation of secondary learning tasks, thereby forming a continuously optimized course recommendation closed loop and improving the pertinence and long-term effectiveness of training. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart of a personalized blood purification course recommendation method for specialist nurses provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the personalized blood purification course recommendation system for specialist nurses provided in an embodiment of the present invention; Figure 3 This is a distribution chart of nurses' blood purification operation assessment values ​​and dominant imbalance stages provided in an embodiment of the present invention; Figure 4 This is a flowchart of a personalized blood purification course recommendation system for specialist nurses provided in an embodiment of the present invention. Detailed Implementation

[0023] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0024] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0025] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0026] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0027] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0028] This invention provides a method for recommending personalized blood purification courses for specialist nurses, such as... Figure 1 The flowchart shown illustrates a personalized blood purification curriculum recommendation method for specialist nurses. This method's process may include the following steps: S1. Collect blood purification operation data, and perform time stamp uniform correction, noise suppression, anomaly removal, missing data compensation and standardization on the blood purification operation data to obtain preprocessed blood purification operation data. S2. According to the machine log, the pre-treated blood purification operation data is divided into four stages: puncture, flushing, connection fixation, and machine monitoring. The time index of each stage is recorded and the corresponding operation characteristic value is calculated. S3. Based on the operational feature values, obtain the operational capability representation parameters for each stage, and extract the historical blood purification operation data of each nurse to construct a nurse operation evaluation model. Use the weight coefficients of each stage as input and output the nurse's blood purification operation evaluation value. S4. Based on each nurse's blood purification operation assessment value and the weight coefficient of each stage, a personalized learning task package is generated and pushed to the nurse's end. After the nurse completes the learning task package, an assessment difference matrix before and after learning is constructed to perform a closed-loop update of course recommendations.

[0029] Optionally, blood purification operation data is collected, and the blood purification operation data is subjected to unified timestamp correction, noise suppression, anomaly removal, missing data compensation and standardization. The specific steps to obtain the pre-processed blood purification operation data are as follows: Blood purification operation data of each nurse during the blood purification operation is collected in real time. During the collection process, operation parameters are acquired synchronously based on the built-in sensor interface of the blood purification equipment and the multi-parameter motion monitoring device worn by the nurse. Blood purification operation data includes arterial pressure, venous pressure, transmembrane pressure, blood pump flow rate, dialysate conductivity, dialysate temperature, vascular depth, needle-to-forearm angle, needle insertion speed, needle insertion depth, sterile area distance, wrist angular velocity, hand linear acceleration, cumulative flushing volume and flushing pressure. Among them, arterial pressure is collected in real time by the arterial end pressure sensing channel of the blood purification equipment, reflecting the dynamic changes in blood pressure entering the circuit; venous pressure is detected by the venous end pressure sensing channel to assess the resistance to blood return; transmembrane pressure is obtained by the blood purification equipment automatically calculating the pressure difference across the dialyzer, reflecting the characteristics of changes in filtration membrane permeability; blood pump flow rate is monitored in real time by the built-in speed encoder and converted into flow rate per unit time; dialysate conductivity is measured by the dialysate flow path conductivity detection probe to characterize solute concentration fluctuations; dialysate temperature is recorded by the dialysate temperature control sensor to monitor temperature control accuracy; vessel depth is measured in real time by an ultrasound probe worn on the back of the nurse's hand to obtain the vertical depth from the needle tip to the center of the vessel; the angle between the needle body and the forearm is measured by a posture sensing wristband worn on the forearm. The data is obtained using a gyroscope; the needle insertion speed is obtained by collecting the displacement change of the needle tip using a wearable position tracking device and calculating the time gradient; the needle insertion depth is calculated by subtracting the puncture initiation position from the maximum displacement of the needle tip trajectory, reflecting the puncture control accuracy; the sterile zone distance is measured in real time by an infrared ranging ring worn on the back of the hand, representing the minimum spatial distance between the operator's hand and the boundary of the sterile protection zone; the wrist angular velocity is collected by an inertial measurement strap worn on the wrist, used to reflect operational stability; the hand linear acceleration is detected by a linear acceleration collector worn on the back of the fingers, reflecting the intensity of subtle hand tremors; the cumulative injection volume during flushing is statistically recorded in real time by a flow meter inside the device, recording the total amount of fluid injected during the flushing stage; the flushing pressure is measured by a pressure sensor built into the flushing tubing, reflecting the pressure fluctuation characteristics during the flushing stage. The sampling frequency is maintained within the high-resolution sampling range throughout the entire blood purification process, ensuring that all blood purification operation data form a continuous time-series data stream within the complete operation cycle, ensuring the integrity and synchronization of time domain characteristics. The collected blood purification operation data is uniformly time-base corrected using a timestamp resampling algorithm to ensure that all blood purification operation data are aligned on a unified time axis, achieving synchronization across signal channels. A weighted moving average filtering algorithm is used to smooth the blood purification operation data, suppressing high-frequency noise, transient jitter, and mechanical interference introduced by wearable devices, thus maintaining the true trend of the operation behavior curve.An outlier detection algorithm (3σ outlier detection algorithm) was used to remove outlier sampling points from the blood purification operation data. Dynamic statistical thresholds were set for each parameter; when a sampling point deviated from the mean by more than three standard deviations, it was removed to eliminate non-physiological abnormal signals. For the time gaps caused by outlier removal, a cubic spline interpolation algorithm was used to compensate for missing data in the blood purification operation data, restoring the complete time sequence based on the gradient continuity and curve smoothness of adjacent sampling points. After interpolation compensation, the Z-Score normalization algorithm was used to standardize all blood purification operation data, using zero mean and unit variance as normalization targets. This unified the numerical scale and eliminated dimensional differences, ensuring that arterial pressure, venous pressure, transmembrane pressure, blood pump flow rate, dialysate conductivity, dialysate temperature, vessel depth, needle-to-forearm angle, needle insertion speed, needle insertion depth, sterile zone distance, wrist angular velocity, hand linear acceleration, cumulative flushing volume, and flushing pressure were comparable in the feature space. This provided a stable and consistent data input foundation for subsequent stage division and operational feature analysis.

[0030] In this implementation plan, by uniformly collecting and standardizing the preprocessing of arterial pressure, venous pressure, transmembrane pressure, blood pump flow rate, dialysate conductivity, dialysate temperature, vessel depth, needle-to-forearm angle, needle insertion speed, needle insertion depth, sterile zone distance, wrist angular velocity, hand linear acceleration, cumulative flushing volume, and flushing pressure, the operational data of the entire blood purification process are synchronized in time series and consistent in numerical scale. After unified time base correction, noise suppression, anomaly removal, missing data compensation, and standardization, the stability, continuity, and analyzability of the data are improved, avoiding evaluation errors caused by time offsets and dimensional differences between different data sources, and providing reliable data support for subsequent operational stage division and feature extraction.

[0031] Optionally, the pre-processed blood purification operation data is divided into four stages—puncture, flushing, connection fixation, and on-machine monitoring—according to the machine log records. The specific steps for recording the time index of each stage and calculating the corresponding operation characteristic values ​​are as follows: The pre-processed blood purification operation data is read in timestamp order. Based on the operation event time identifiers recorded in the machine log, combined with the needle insertion trigger signal, flushing start signal, connection fixation operation trigger signal, and on-machine monitoring start signal, the entire blood purification process is divided into operation stages. Each signal is aligned with the equipment clock reference through a time synchronization interface to ensure that the time boundaries of the stage division accurately correspond to the actual operation nodes. The blood purification operation data is divided into the puncture stage, flushing stage, connection fixation stage, and on-machine monitoring stage. Within each stage, the corresponding time index interval is extracted to establish a stage time index table, which is used to record the start time, end time, and data sampling range of each stage. During the puncture phase, time-series curves of the angle between the needle and forearm, needle insertion speed, and needle depth are constructed. The standard deviation and mean of the needle insertion speed are calculated. The standard deviation reflects the degree of speed fluctuation during needle insertion, while the mean is used to characterize the overall operational stability. Together, they are used to assess the smoothness and control precision of the puncture procedure. During the flushing phase, based on continuous sampling data of flushing pressure and cumulative injection volume, the standard deviation and mean of flushing pressure are calculated. A safe threshold for flushing pressure is extracted, and the number of times the flushing pressure exceeds this threshold is counted to obtain the number of flushing pressure over-limits, reflecting the nurse's pressure control stability and flow regulation accuracy during the flushing process. During the connection fixation phase, based on continuous time series of wrist angular velocity and hand linear acceleration, the root mean square of wrist angular velocity and the root mean square of hand linear acceleration are calculated to characterize the intensity of hand posture fluctuations. The minimum distance to the sterile area is extracted to reflect the spatial safety boundary control capability during the operation. A sterile area distance threshold is set, and the number of times the sterile area distance falls below the threshold is counted to obtain the number of sterile area over-limits, used to assess aseptic operation compliance and spatial movement accuracy. During the on-board monitoring phase, based on continuous sampling data of arterial pressure, venous pressure, transmembrane pressure, blood pump flow rate, dialysate conductivity, and dialysate temperature, the standard deviation and mean of each parameter are calculated to characterize the stability of equipment operation and parameter fluctuations during monitoring. Safe threshold ranges for arterial pressure, venous pressure, and transmembrane pressure are set, and the number of times each parameter exceeds the corresponding threshold is counted to obtain the number of monitoring pressure exceedances, reflecting the hemodynamic control level and operational safety during the on-board monitoring phase.

[0032] In this implementation plan, the pre-processed blood purification operation data is accurately divided into four stages: puncture, flushing, connection fixation, and monitoring. This ensures clear time boundaries and data correspondences for each stage. Extraction of operational characteristic values ​​for each stage allows for the independent quantification of key parameters such as needle insertion speed, flushing pressure, wrist angular velocity, hand linear acceleration, sterile zone distance, arterial pressure, venous pressure, transmembrane pressure, blood pump flow rate, dialysate conductivity, and dialysate temperature at different operational stages. Through the characteristic processing of operational behaviors at each stage, a structured decomposition and quantifiable analysis of the entire blood purification operation process are achieved, providing precise stage-specific data support for subsequent weighting coefficient calculations and nurse performance evaluation.

[0033] Optionally, based on operational feature values, operational capability representation parameters for each stage are obtained, and historical blood purification operation data for each nurse is extracted to construct a nurse operation assessment model. The specific steps for outputting the nurse's blood purification operation assessment value, using the weight coefficients of each stage as input, are as follows: While maintaining consistency between stage data and time index, firstly, a mapping list between capability dimensions and stage features is established, clarifying the feature source, statistical caliber, and dimensional benchmark for each capability dimension, unifying the value direction and normalization interval of features, and ensuring that different features can be superimposed and compared on the same scale during weight calculation; then, four capability dimensions—puncture accuracy, flushing standardization, aseptic compliance, and monitoring coordination stability—are extracted from the department's nurse training program. The operational feature values ​​extracted from each stage are weighted and summarized to obtain puncture weight coefficients, flushing weight coefficients, connection fixation weight coefficients, and on-machine monitoring weight coefficients, respectively; during the generation of weight coefficients, for each... One capability dimension sets weight boundaries and numerical stability constraints, and selects weighting factors through cross-validation to ensure the repeatability of weights across different batches of data. The time labels for weight generation are recorded for subsequent tracking. Next, historical blood purification operation data for each nurse are extracted, and a nurse operation evaluation model is constructed based on the gradient boosting tree algorithm. The model is trained using standard blood purification data from the department as supervised labels. Before training, the training and validation sets are divided and the class ratios are aligned. A grid search method is used to determine the learning rate, the number of base learners, the maximum tree depth, and the subsample ratio. An early stopping strategy is used to avoid overfitting, and the consistency between the weight coefficients and the capability dimension is verified using feature importance evaluation results. The standard blood purification data from the department, as well as the reference wrist angular velocity, reference hand linear acceleration, standard dialysate conductivity, standard dialysate temperature, and standard flushing volume involved in this invention, are all derived from a benchmark dataset validated through standardized operations. This benchmark data is collected by senior specialist nurses with extensive blood purification experience under standardized clinical conditions. All sampling processes use a uniform wearable data acquisition device and the same time sampling frequency. After time alignment, outlier removal, and root mean square (RMS) calculation of the collected data, the median of multiple operation results is taken as the corresponding reference value. Each reference value is calibrated and indexed with a time index to ensure consistency in the numerical values ​​of each parameter, traceability of source, and comparability of results in subsequent calculations. The puncture weight coefficient, flushing weight coefficient, connection fixation weight coefficient, and on-machine monitoring weight coefficient are input into the trained nurse operation assessment model, outputting a blood purification operation assessment value for each nurse. During the output stage, the assessment values ​​are calibrated within a numerical range, and threshold suggestions and time benchmarks based on historical reference quantiles are provided. The correspondence between the blood purification operation assessment value and the input weight coefficients, along with the time index, are retained to facilitate traceability verification during subsequent learning task generation and closed-loop updates.

[0034] In this implementation plan, weighted coefficients for puncture, flushing, connection fixation, and on-board monitoring are established based on operational feature values. A gradient boosting tree algorithm-based operational evaluation model is constructed by combining this with nurses' historical blood purification operation data. This allows the evaluation results to comprehensively reflect nurses' overall performance across four capability dimensions: puncture accuracy, flushing standardization, aseptic compliance, and monitoring coordination stability. By introducing standard blood purification data as a supervisory label, this method achieves objectivity and quantification of the model training process, ensuring the comparability and traceability of the evaluation results. Through the corresponding calculation of the weighted inputs and outputs, the differences in nurses' capabilities at each stage of the operation can be accurately identified, providing a reliable evaluation basis for the generation of subsequent personalized learning tasks.

[0035] Optionally, the specific steps for obtaining the puncture weighting coefficient are as follows: Within the time index range of the puncture phase, extract continuous sampling data of the angle between the needle and forearm, needle insertion speed, needle insertion depth, vessel depth, wrist angular velocity, and hand linear acceleration. Align the sequences according to a unified time reference to ensure that each parameter corresponds to the same operational action at the same sampling time. Divide the absolute value of the difference between the angle between the needle and forearm and the standard puncture angle by the allowable angle deviation to obtain the angle deviation normalized value, which is used to measure the degree of deviation between the needle insertion posture and the standard angle. Divide the standard deviation of the needle insertion speed by the average needle insertion speed to obtain the speed stability ratio, which reflects the fluctuation and uniformity of the needle insertion speed. Subtract the vessel depth from the needle insertion depth and divide by the vessel depth to obtain the depth deviation ratio, which characterizes the deviation amplitude between the needle insertion endpoint and the target vessel depth. The normalized angle deviation, velocity stability ratio, and depth deviation ratio are squared and summed. The square root of the sum is then taken as the negative exponent, and a power function with base e is calculated to obtain the puncture accuracy attenuation term. This attenuation term describes the attenuation of puncture accuracy due to accumulated operational deviations. The sum of the squares of the wrist angular velocity and the hand linear acceleration is divided by the sum of the squares of the reference wrist angular velocity and the reference hand linear acceleration. The cube root of the ratio is then added to one and the reciprocal is taken to obtain the operational stability correction term, which reflects the stability level of the nurse's posture control during puncture. The puncture accuracy attenuation term is multiplied by the operational stability correction term to obtain the puncture weight coefficient, and the corresponding calculation time index is recorded for subsequent weight trend analysis and operational evaluation model input. To prevent abnormal operational data from causing the calculation results to deviate from the reasonable range, a boundary protection mechanism is set for the puncture weight coefficient and the tube flushing weight coefficient, connection fixing weight coefficient and on-machine monitoring weight coefficient involved in this invention. After the calculation is completed, the results are restricted according to the statistical range of historical data distribution and standard operation samples. When the calculated value exceeds the boundary, it is automatically truncated or mapped and corrected to keep the weight coefficients of each stage within the effective range that reflects the actual operation capability.

[0036] The specific formula for calculating the puncture weighting coefficient is as follows: ; In the formula, This represents the puncture weighting coefficient. Indicates the angle between the needle body and the forearm. Indicates the standard puncture angle. Indicates the allowable angular deviation. This indicates the standard deviation of the needle insertion speed. This represents the average needle insertion speed. Indicates the depth of needle insertion. Indicates the depth of the blood vessel. Indicates the wrist angular velocity. Indicates the linear acceleration of the hand. Indicates the reference wrist angular velocity. This indicates the linear acceleration of the reference hand.

[0037] In this implementation plan, a quantitative evaluation mechanism for the puncture stage is established through the joint calculation of parameters such as the angle between the needle and forearm, needle insertion speed, needle insertion depth, blood vessel depth, wrist angular velocity, and hand linear acceleration. This allows for a comprehensive representation of angular deviation, speed stability, and depth control capabilities during puncture operations using a mathematical model. By constructing a puncture accuracy attenuation term and an operational stability correction term and fusing them exponentially, continuous mapping and dynamic evaluation of puncture operation errors are achieved, ensuring that the puncture weight coefficients accurately reflect the nurse's posture control and operational precision during the puncture process. This method transforms puncture data characteristics into calculable quantitative indicators, improving the objectivity and sensitivity of the evaluation and providing a highly discriminative input basis for subsequent multi-stage weighted analysis and comprehensive evaluation model training.

[0038] Optionally, the specific steps for obtaining the flushing weight coefficient are as follows: Within the time index range of the flushing stage, read the pre-processed flushing pressure and cumulative flushing volume data sequentially according to the timestamp order to ensure that the sampling points of each parameter are aligned under a unified time reference, avoiding calculation deviations caused by data displacement. Calculate the difference between the cumulative flushing volume and the standard flushing volume in the sampling order, sum the absolute values, and record the time index and difference amplitude for each calculation for subsequent flow deviation trend analysis. Divide the sum by the product of the number of samplings in the flushing stage and the standard flushing volume to obtain the flow deviation term, which reflects the stability of the nurse's injection volume control during the flushing operation. Divide the number of times the flushing pressure exceeds the limit by the number of samplings in the flushing stage to obtain the pressure exceedance rate term, which characterizes the ability to control the flushing pressure within a safe range. Divide the standard deviation of the flushing pressure by the average flushing pressure to obtain the pressure fluctuation ratio term, which is used to evaluate the uniformity and consistency of pressure output during the flushing process. The flow deviation term, pressure exceedance rate term, and pressure fluctuation ratio term are added together, and the cube root is taken. The result represents the overall deviation of the flushing operation. Subtracting the cube root from one yields the flushing weight coefficient. To ensure data traceability, the sampling time index and data source identifier corresponding to the flushing weight coefficient are recorded after calculation for subsequent input and calibration of the evaluation model.

[0039] The specific formula for calculating the punching weight coefficient is as follows: ; In the formula, This represents the punching weight coefficient. Indicates the number of samples taken during the flushing stage. Indicates the standard flushing injection volume. This represents the cumulative injection volume during the k-th sampling. Indicates the standard flushing injection volume. This indicates the number of times the flushing pressure exceeds the limit. Indicates the standard deviation of the flushing pressure. This represents the average pressure applied during flushing.

[0040] This implementation plan achieves a comprehensive assessment of the accuracy of injection volume control and pressure stability during the flushing stage by quantifying key data such as cumulative flushing volume, standard deviation of flushing pressure, average flushing pressure, and number of times flushing pressure exceeds limits. By constructing flow deviation, pressure exceedance rate, and pressure fluctuation ratio terms, and obtaining flushing weighting coefficients through nonlinear function transformation, the flow balance and pressure safety during the flushing operation are uniformly measured. This method effectively reflects the nurse's operational standardization and control ability during the flushing stage, providing stable and discriminative stage-specific input parameters for training subsequent operational evaluation models.

[0041] Optionally, the specific steps for obtaining the connection fixation weight coefficient are as follows: Within the time index range of the connection fixation phase, extract continuous sampling data of sterile zone distance, wrist angular velocity, and hand linear acceleration, ensuring that the data corresponds consistently with the timestamps. Divide the minimum sterile zone distance by the sterile zone distance of each sampling, sum the resulting ratios, and take the average to obtain the sterile zone distance deviation term, which characterizes the stability of the nurse's maintenance of the sterile zone boundary during the connection fixation operation. Divide the number of sterile zone violations by the number of samplings in the connection fixation phase, and add the sterile zone distance deviation term to obtain the sterile control comprehensive term, which reflects the degree of coordination between hand position control and sterile distance maintenance during the operation. Add the root mean square of the wrist angular velocity to the root mean square of the hand linear acceleration, and divide by the sum of the reference wrist angular velocity and the reference hand linear acceleration to obtain the posture fluctuation correction term, which is used to correct control deviations caused by posture changes during the operation. Multiplying the aseptic control comprehensive term by the attitude fluctuation correction term, adding one to the product, and taking the natural logarithm yields the logarithmic gain term, which describes the gain characteristics of aseptic control during the dynamic process of operation. Dividing the logarithmic gain term by one plus the natural logarithm of e yields the normalization coefficient, ensuring that the results among different nurses have uniform dimensions and comparability. Subtracting the normalization coefficient from one yields the connection fixed weight coefficient, and the corresponding calculation time index is recorded for subsequent input and calibration analysis of the operational evaluation model.

[0042] The specific formula for calculating the fixed weighting coefficient of the connection is as follows: ; In the formula, This indicates the fixed weighting coefficient for the circuit connection. Indicates the number of times the sterility limit has been exceeded. Indicates the number of samples taken during the fixed phase of the connection. This represents the minimum distance to the sterile area. This represents the distance to the sterile area during the i-th sampling. This represents the root mean square of the wrist angular velocity. This represents the root mean square of the linear acceleration of the hand. Indicates the reference wrist angular velocity. This indicates the linear acceleration of the reference hand.

[0043] In this implementation plan, a mathematical correlation between operational stability and aseptic control capability during the connection fixation phase is established by quantifying the distance to the sterile area, wrist angular velocity, and hand linear acceleration. By constructing a sterile area distance deviation term, a comprehensive aseptic control term, and a posture fluctuation correction term, and introducing logarithmic gain and normalization, the connection fixation weight coefficient accurately reflects the nurse's comprehensive ability to maintain sterile boundaries, control hand posture, and perform connection fixation procedures. This method improves the accuracy and stability of operational assessment and provides representative, highly reliable input parameters for individualized assessment models.

[0044] Optionally, the specific steps for obtaining the weighting coefficients for on-machine monitoring are as follows: Within the time index range of the on-machine monitoring phase, extract continuous sampling data for arterial pressure, venous pressure, transmembrane pressure, blood pump flow rate, dialysate conductivity, dialysate temperature, and the number of monitoring pressure exceedances, ensuring that all monitoring parameters are aligned under a unified time reference. Subtract venous pressure from arterial pressure to obtain the blood circuit pressure differential, used to characterize the hemodynamic load balance during blood purification. Subtract the ratio of average transmembrane pressure to average blood pump flow rate from the ratio of transmembrane pressure to blood pump flow rate to obtain the filtration flux deviation, used to reflect the fluid transfer stability between membrane flux and pump speed. Multiply the blood circuit pressure differential by the filtration flux deviation and take the absolute value to obtain the pressure-flow coupling deviation term, used to assess the synergy between blood flow rate and transmembrane pressure changes. Add the absolute value of the difference between dialysate conductivity and standard dialysate conductivity to the absolute value of the difference between dialysate temperature and standard dialysate temperature to obtain the dialysate stability deviation term, used to describe the stability of dialysate physical properties during operation. Multiply the dialysate stability deviation term by the number of monitoring pressure exceedances plus one, and take the square root of the product to obtain the stability adjustment term, which is used to correct the combined deviation caused by dialysate characteristic fluctuations and monitoring abnormalities. Divide the pressure-flow coupling deviation term by the stability adjustment term and take the negative value as the exponent to calculate the power function value with base e, which yields the on-machine monitoring weight coefficient. Record the corresponding calculation time index to ensure the traceability of the results and the consistency of the input to the subsequent evaluation model.

[0045] The specific formula for calculating the weighting coefficient of on-site monitoring is as follows: ; In the formula, This represents the weighting coefficient for on-machine monitoring. Arterial pressure, Indicates venous pressure. Indicates transmembrane pressure, Indicates blood pump flow rate, This represents the average transmembrane pressure. This represents the average blood pump flow rate. Indicates the conductivity of the dialysate. Indicates the conductivity of the standard dialysate. Indicates the temperature of the dialysate. Indicates the standard dialysate temperature. This indicates the number of times the monitoring pressure exceeded the limit.

[0046] This implementation plan achieves a quantitative assessment of hemodynamic stability and dialysate control precision during the on-machine monitoring phase by comprehensively calculating parameters such as arterial pressure, venous pressure, transmembrane pressure, blood pump flow rate, dialysate conductivity, dialysate temperature, and the number of monitoring pressure exceedances. By constructing blood circuit pressure difference, filtration flux deviation, pressure-flow coupling deviation, dialysate stability deviation, and stability adjustment terms, and introducing an exponential function calculation, the on-machine monitoring weight coefficients accurately reflect the nurse's control ability and stability maintenance level during on-machine monitoring. This method improves the sensitivity and accuracy of data assessment during the monitoring phase, providing a scientific data foundation for dynamic feedback and course recommendation in the operational assessment model.

[0047] Optionally, based on each nurse's blood purification operation assessment value and the weight coefficients of each stage, a personalized learning task package is generated and pushed to the nurse's end. After the nurse completes the learning task package, an assessment difference matrix before and after learning is constructed, and the specific steps for closed-loop updating of course recommendations are as follows: Under a unified time benchmark, the blood purification operation assessment value and the weight coefficients of each stage are read in real time. Consistency verification is performed by combining the time index and data source identifier, and the blood purification operation assessment value is compared with the operation assessment threshold in real time. When the blood purification operation assessment value is greater than the operation assessment threshold, the deviation ratio of each stage weight coefficient to the corresponding threshold is calculated. The calculation caliber and decimal precision of the deviation ratio are clarified, and the operation stage corresponding to the largest deviation ratio is determined as the dominant imbalance stage. The time index and weight value of the dominant imbalance stage are recorded. The operation knowledge points and teaching videos associated with the dominant imbalance stage are automatically retrieved. Based on the mapping relationship between knowledge points and weight coefficients, a targeted learning task package is generated and pushed to the nurse's end. At the same time, a learning duration suggestion and completion judgment criteria are given, and the push time and receipt confirmation record are retained. When the blood purification operation assessment value is less than or equal to the operation assessment threshold, the specific steps are as follows: During the process, operational knowledge points for each stage are retrieved, and a multi-layered operational knowledge graph based on weight hierarchy is constructed according to the deviation ratio of each stage from largest to smallest. Based on the graph hierarchy, a phased reinforcement learning package is generated, and the knowledge point index, teaching videos, and operational process demonstrations are pushed to the nurse's end, along with the phased completion order and assessment requirements. After the nurse completes the learning task package, the weight coefficients and blood purification operation assessment values ​​for each stage are re-acquired for the corresponding nurse using the same sampling frequency and statistical caliber. The difference between the weight coefficients and blood purification operation assessment values ​​before and after learning is calculated, generating an assessment difference matrix, and marking the direction, magnitude, and time index of the difference. When the weight coefficient difference for any stage is greater than or equal to the corresponding improvement threshold, the weight coefficients and time index for the corresponding stage in the nurse's operational assessment model are updated, and the update batch and parameter version number are recorded. When the weight coefficient difference for any stage is less than the improvement threshold, knowledge points are re-selected and matched based on the assessment difference matrix to generate a secondary learning package, which is then pushed to the nurse's end. Simultaneously, a retesting time window and a passing threshold are set for the secondary learning package to ensure the traceability and stable convergence of the closed-loop update process.

[0048] In this embodiment, Table 1 is a data table for evaluating nurses' blood purification operations, listing the weight coefficients for the puncture stage, flushing stage, connection fixation stage, and on-machine monitoring stage, as well as the corresponding blood purification operation evaluation results for five nurses. Specific data are as follows: For nurse A001, the weight coefficient for the puncture stage is 0.86, the weight coefficient for the flushing stage is 0.79, the weight coefficient for the connection fixation stage is 0.73, and the weight coefficient for the on-machine monitoring stage is 0.88, resulting in a blood purification operation evaluation value of 0.82; for nurse A002, the weight coefficients for the puncture stage are 0.89, the weight coefficients for the flushing stage are 0.82, the weight coefficients for the connection fixation stage are 0.78, and the weight coefficients for the on-machine monitoring stage are 0.91, resulting in a blood purification operation evaluation value of 0.85; for nurse A003, the weight coefficients for the puncture stage are 0.80, the weight coefficients for the flushing stage are 0.75, and the weight coefficients for the connection fixation stage are 0.73, 0.88, and 0.91, resulting in a blood purification operation evaluation value of 0.85; for nurse A003, the weight coefficients for the puncture stage are 0.80, the weight coefficients for the flushing stage are 0.75, and the weight coefficients for the connection fixation stage are 0.73, 0.88, and 0.91, resulting in a blood purification operation evaluation value of 0.85. The weighting coefficient for the fixation stage is 0.70, the weighting coefficient for the on-machine monitoring stage is 0.84, and the corresponding blood purification operation assessment value is 0.77. For nurse A004, the weighting coefficient for the puncture stage is 0.91, the weighting coefficient for the flushing stage is 0.85, the weighting coefficient for the connection fixation stage is 0.80, and the weighting coefficient for the on-machine monitoring stage is 0.93, and the corresponding blood purification operation assessment value is 0.87. For nurse A005, the weighting coefficient for the puncture stage is 0.84, the weighting coefficient for the flushing stage is 0.78, the weighting coefficient for the connection fixation stage is 0.75, and the weighting coefficient for the on-machine monitoring stage is 0.86, and the corresponding blood purification operation assessment value is 0.80.

[0049] Table 1. Nurses' Blood Purification Procedure Assessment Data like Figure 3The figure shows the distribution of blood purification operation assessment values ​​for five nurses in the same blood purification procedure scenario, reflecting the overall execution level and stage-specific ability differences among nurses throughout the entire operation. The figure uses a bar chart to plot the blood purification operation assessment results for each nurse, with the vertical axis representing the blood purification operation assessment value and the horizontal axis representing the nurse's number. A yellow dashed line is set as the operation assessment threshold to distinguish whether the nurse's overall operational ability meets the standard operation requirements. The bar colors differentiate the assessment result levels: blue bars represent nurses whose blood purification operation assessment values ​​exceed the operation assessment threshold, i.e., those whose overall operational performance is above the standard level, and the dominant imbalance stage is marked above the bar to indicate the key areas for optimization during their operation; red bars represent nurses whose blood purification operation assessment values ​​do not meet the operation assessment threshold, i.e., those whose operational standardization is insufficient, and the blood purification operation assessment value is marked above the bar. As shown in the figure, nurses A001, A002, and A004 have blood purification operation assessment values ​​higher than the operation assessment threshold, and need to focus on optimizing their learning in the flushing, connection fixation, and puncture stages, respectively; nurses A003 and A005 have blood purification operation assessment values ​​that do not exceed the operation assessment threshold, and need to undergo intensive training in stages. Figure 3 This invention visually demonstrates the nurse operation assessment results based on stage-based weight analysis, which can effectively identify individual operational shortcomings and provide data support for personalized learning task generation and course recommendation, thereby achieving continuous improvement and refined training of nurses' operational abilities.

[0050] In this implementation plan, by comparing blood purification operation assessment values ​​with operation assessment thresholds in real time and analyzing the deviation ratio of weight coefficients at each stage, the system achieves automatic diagnosis of nurses' operational capabilities and generates targeted learning tasks. This method can automatically match corresponding operational knowledge points and instructional videos based on the dominant imbalance stage, ensuring a high degree of correspondence between learning content and actual skill gaps. By constructing a pre- and post-learning assessment difference matrix and dynamically updating the weight coefficients and blood purification operation assessment values, the course recommendation process forms an adaptive closed-loop optimization mechanism, thereby achieving continuous tracking and precise feedback on nurses' skill improvement.

[0051] Furthermore, embodiments of the present invention also provide a personalized blood purification course recommendation system for specialist nurses, such as... Figure 2The diagram shows the structure of a personalized blood purification course recommendation system for specialist nurses. This system, applied to a personalized blood purification course recommendation method for specialist nurses, includes: a data acquisition and preprocessing module for collecting blood purification operation data and performing timestamp-based correction, noise suppression, anomaly removal, missing data compensation, and standardization to obtain preprocessed blood purification operation data; a stage feature extraction module for dividing the preprocessed blood purification operation data into four stages—puncture, flushing, connection fixation, and on-machine monitoring—according to machine log records, recording the time index of each stage and calculating the corresponding operation feature values; an operation assessment modeling module for obtaining operation capability representation parameters for each stage based on the operation feature values, extracting historical blood purification operation data for each nurse to construct a nurse operation assessment model, using the weight coefficients of each stage as input, and outputting the nurse's blood purification operation assessment value; and a personalized recommendation update module for generating personalized learning task packages based on each nurse's blood purification operation assessment value and the weight coefficients of each stage, pushing them to the nurse's end, and constructing an assessment difference matrix before and after learning after the nurse completes the learning task package, performing a closed-loop update of the course recommendation. The specific implementation process of the system is as follows: Figure 4 As shown.

[0052] In this implementation plan, a data acquisition and preprocessing module achieves high-precision acquisition and standardized processing of blood purification operation data, ensuring the temporal consistency and integrity of data input. A stage feature extraction module divides the preprocessed data into four stages: puncture, flushing, connection fixation, and on-machine monitoring, extracting and quantifying key operational feature values ​​to achieve a structured expression of the entire process. An operation assessment modeling module establishes a nurse operation assessment model based on a gradient boosting tree algorithm, using the correlation between the weight coefficients of each stage and historical blood purification operation data for training, achieving a comprehensive evaluation of puncture accuracy, flushing standardization, aseptic compliance, and monitoring coordination stability. A personalized recommendation and update module automatically generates personalized learning task packages based on the dynamic changes in blood purification operation assessment values ​​and stage weight coefficients, and uses the pre- and post-learning assessment difference matrix to achieve closed-loop optimization and updates, enabling continuous improvement and adaptive characteristics in the nurse training process. Through data-driven quantitative modeling and intelligent recommendation, this system automates, refines, and personalizes the assessment and course recommendation of specialist nurses' blood purification operation skills, significantly improving training effectiveness and clinical operation quality.

[0053] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0054] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0055] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0056] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply 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 invention.

[0057] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0058] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0059] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0060] The units described as separate components may or may not be physically separate. The 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 the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0061] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0062] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0063] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for recommending personalized blood purification courses for specialist nurses, characterized in that, The method includes: S1. Collect blood purification operation data, and perform time stamp uniform correction, noise suppression, anomaly removal, missing data compensation and standardization on the blood purification operation data to obtain preprocessed blood purification operation data. S2. According to the machine log, the pre-treated blood purification operation data is divided into four stages: puncture, flushing, connection fixation, and machine monitoring. The time index of each stage is recorded and the corresponding operation characteristic value is calculated. S3. Based on the operational feature values, obtain the operational capability representation parameters for each stage, and extract the historical blood purification operation data of each nurse to construct a nurse operation evaluation model. Use the weight coefficients of each stage as input and output the nurse's blood purification operation evaluation value. S4. Based on each nurse's blood purification operation assessment value and the weight coefficient of each stage, a personalized learning task package is generated and pushed to the nurse's end. After the nurse completes the learning task package, an assessment difference matrix before and after learning is constructed to perform a closed-loop update of course recommendations.

2. The method for recommending personalized blood purification courses for specialist nurses according to claim 1, characterized in that, The specific steps for collecting blood purification operation data and performing time stamp correction, noise suppression, anomaly removal, missing data compensation, and standardization on the blood purification operation data to obtain preprocessed blood purification operation data are as follows: Real-time data collection of blood purification procedures for each nurse during the blood purification process. Blood purification procedure data includes arterial pressure, venous pressure, transmembrane pressure, blood pump flow rate, dialysate conductivity, dialysate temperature, vessel depth, needle-to-forearm angle, needle insertion speed, needle insertion depth, distance to sterile area, wrist angular velocity, hand linear acceleration, cumulative flushing volume, and flushing pressure. The collected blood purification operation data are uniformly time-base corrected using a timestamp resampling algorithm to ensure that all data are aligned under the same time base; smoothed using a weighted moving average filtering algorithm to suppress high-frequency noise and transient jitter; outlier sampling points are initially removed using a 3σ outlier detection algorithm, and missing data are compensated using a cubic spline interpolation algorithm; and standardized using a Z-Score standardization algorithm to unify the numerical scale and eliminate dimensional differences.

3. The method for recommending personalized blood purification courses for specialist nurses according to claim 1, characterized in that, The specific steps for dividing the pre-processed blood purification operation data into four stages—puncture, flushing, connection fixation, and on-machine monitoring—according to the machine log records, and recording the time index of each stage and calculating the corresponding operation characteristic values ​​are as follows: The preprocessed blood purification operation data is read in timestamp order. Based on the operation event time markers recorded in the machine log and the operation trigger signals for needle insertion, flushing, connection, and machine setup, the entire operation process is divided into stages: puncture stage, flushing stage, connection fixation stage, and machine monitoring stage. The time index of each operation stage is recorded, and the operation characteristic value of each operation stage is calculated. For the puncture phase, calculate the standard deviation and mean of the needle insertion speed; For the flushing stage, calculate the standard deviation and average value of the flushing pressure, count the number of times the flushing pressure exceeds the flushing pressure threshold, and obtain the number of times the flushing pressure exceeds the limit. During the connection fixation phase, the root mean square of the wrist angular velocity and hand linear acceleration is calculated, the minimum distance of the sterile area is extracted, and the number of times the distance of the sterile area is lower than the sterile area distance threshold is counted to obtain the number of sterile area violations. During the on-machine monitoring phase, the standard deviation and mean of arterial pressure, venous pressure, transmembrane pressure, blood pump flow rate, dialysate conductivity, and dialysate temperature are calculated, and the number of times arterial pressure, venous pressure, and transmembrane pressure exceed the corresponding safety threshold is counted to obtain the number of monitoring pressure exceeding the limit.

4. The method for recommending personalized blood purification courses for specialist nurses according to claim 3, characterized in that, The specific steps for obtaining operational capability characterization parameters for each stage based on operational feature values, constructing a nurse operational assessment model by extracting historical blood purification operation data for each nurse, and outputting the nurse's blood purification operation assessment value using the weight coefficients of each stage as input are as follows: Four competency dimensions were extracted from the department nurse training program: puncture accuracy, flushing standardization, aseptic compliance, and monitoring coordination stability. The operational feature values ​​extracted at each stage were weighted and summarized to obtain puncture weight coefficient, flushing weight coefficient, connection fixation weight coefficient, and on-board monitoring weight coefficient, respectively. Historical blood purification operation data of each nurse were extracted, and a nurse operation evaluation model was constructed based on the gradient boosting tree algorithm. The model was trained using standard blood purification data of the department as supervision labels. The weight coefficients of puncture, flushing, connection fixation and on-machine monitoring were input into the trained nurse operation evaluation model, and the blood purification operation evaluation value of each nurse was output.

5. The method for recommending personalized blood purification courses for specialist nurses according to claim 4, characterized in that, The specific steps for obtaining the puncture weighting coefficient are as follows: The angle deviation is normalized by dividing the absolute value of the difference between the angle between the needle body and the forearm and the standard puncture angle by the allowable angle deviation. The speed stability ratio is obtained by dividing the standard deviation of the needle insertion speed by the average needle insertion speed. The depth deviation ratio is obtained by subtracting the blood vessel depth from the needle insertion depth and then dividing by the blood vessel depth. The angle deviation normalization value, the speed stability ratio, and the depth deviation ratio are squared and summed. The square root of the sum is then taken as the exponent, and a power function with base e is calculated to obtain the puncture accuracy attenuation term. The operation stability correction term is obtained by dividing the sum of the squares of the wrist angular velocity and the hand linear acceleration by the sum of the squares of the reference wrist angular velocity and the reference hand linear acceleration, taking the cube root of the ratio, adding one, and taking the reciprocal. The puncture accuracy attenuation term is multiplied by the operation stability correction term to obtain the puncture weighting coefficient.

6. The method for recommending personalized blood purification courses for specialist nurses according to claim 4, characterized in that, The specific steps for obtaining the punching weight coefficient are as follows: Calculate the difference between the cumulative injection volume of each flushing and the standard flushing injection volume in the sampling order, take the absolute value and sum them; divide the sum by the product of the number of samplings in the flushing stage and the standard flushing injection volume to obtain the flow deviation term; divide the number of times the flushing pressure exceeds the limit by the number of samplings in the flushing stage to obtain the pressure exceedance rate term; divide the standard deviation of the flushing pressure by the average flushing pressure to obtain the pressure fluctuation ratio term; add the flow deviation term, the pressure exceedance rate term and the pressure fluctuation ratio term, take the cube root, and then subtract the cube root from one to obtain the flushing weight coefficient.

7. The method for recommending personalized blood purification courses for specialist nurses according to claim 4, characterized in that, The specific steps for obtaining the fixed weighting coefficient of the connection are as follows: Divide the minimum distance to the sterile zone by the distance to the sterile zone in each sampling, sum the resulting ratios, and take the average to obtain the sterile zone distance deviation term. Divide the number of sterile zone violations by the number of samplings in the connection fixing phase, and add the sterile zone distance deviation term to obtain the sterile control comprehensive term. Add the root mean square of the wrist angular velocity to the root mean square of the hand linear acceleration, and divide by the sum of the reference wrist angular velocity and the reference hand linear acceleration to obtain the attitude fluctuation correction term. Multiply the sterile control comprehensive term by the attitude fluctuation correction term, add one to the product, and take the natural logarithm to obtain the logarithmic gain term. Divide the logarithmic gain term by one and add the natural logarithm of e to obtain the normalization coefficient. Subtract the normalization coefficient from one to obtain the connection fixing weight coefficient.

8. The method for recommending personalized blood purification courses for specialist nurses according to claim 4, characterized in that, The specific steps for obtaining the on-site monitoring weight coefficient are as follows: The blood flow pressure differential is obtained by subtracting venous pressure from arterial pressure; the filtration flux deviation is obtained by subtracting the ratio of average transmembrane pressure to average blood pump flow from the ratio of transmembrane pressure to blood pump flow rate; the absolute value of the blood flow pressure differential multiplied by the filtration flux deviation is obtained as the pressure-flow coupling deviation term; the absolute value of the difference between dialysate conductivity and standard dialysate conductivity, and the absolute value of the difference between dialysate temperature and standard dialysate temperature, are added to obtain the dialysate stability deviation term; the dialysate stability deviation term is multiplied by the number of monitoring pressure exceedances plus one, and the square root of the product is obtained as the stability adjustment term; the negative value of the pressure-flow coupling deviation term divided by the stability adjustment term is used as the exponent to calculate the power function value with base e, which is the on-machine monitoring weight coefficient.

9. The method for recommending personalized blood purification courses for specialist nurses according to claim 1, characterized in that, The specific steps for generating personalized learning task packages based on each nurse's blood purification operation assessment value and the weight coefficients of each stage, pushing them to the nurse's end, and constructing an assessment difference matrix before and after learning to perform closed-loop updates of course recommendations are as follows: The system reads each nurse's blood purification operation assessment value and the weight coefficient of each stage in real time, and compares the blood purification operation assessment value with the operation assessment threshold in real time. When the blood purification operation assessment value is greater than the operation assessment threshold, the system calculates the deviation ratio between the weight coefficient of each stage and the corresponding threshold, and determines the operation stage corresponding to the largest deviation ratio as the dominant imbalance stage. The system automatically retrieves operation knowledge points and teaching videos associated with the dominant imbalance stage, generates targeted learning task packages, and pushes them to the nurse's terminal. When the blood purification operation assessment value is less than or equal to the operation assessment threshold, the operation knowledge points of each operation stage are retrieved, and a multi-level operation knowledge graph based on the weight hierarchy is constructed according to the deviation ratio of each stage from large to small. Based on the graph hierarchy, a phased consolidation learning package is generated, and the knowledge point index, teaching videos and operation process demonstrations are pushed to the nurse's terminal. After the nurse completes the learning task package, the weight coefficients and blood purification operation assessment values ​​for each stage of the corresponding nurse are retrieved again, and the difference between the weight coefficients and blood purification operation assessment values ​​before and after learning is calculated to generate an assessment difference matrix. When the difference in the weight coefficient of any stage is greater than or equal to the corresponding improvement threshold, the weight coefficient and time index of the corresponding stage in the nurse operation assessment model are updated. When the difference in the weight coefficient of any stage is less than the improvement threshold, the knowledge points are re-selected and matched to generate a secondary learning package and pushed to the nurse's end.

10. A personalized blood purification course recommendation system for specialist nurses, the system being applied to the personalized blood purification course recommendation method for specialist nurses as described in any one of claims 1-9, characterized in that, The system includes: The data acquisition and preprocessing module is used to collect blood purification operation data and perform time stamp uniform correction, noise suppression, anomaly removal, missing data compensation and standardization on the blood purification operation data to obtain preprocessed blood purification operation data. The stage feature extraction module is used to divide the pre-processed blood purification operation data into four stages according to the machine log records: puncture, flushing, connection fixation, and machine monitoring. It records the time index of each stage and calculates the corresponding operation feature values. The operation assessment modeling module is used to obtain the operation capability representation parameters of each stage based on the operation feature value, and to extract the historical blood purification operation data of each nurse to construct the nurse operation assessment model. The weight coefficients of each stage are used as input, and the output is the nurse's blood purification operation assessment value. The personalized recommendation update module is used to generate personalized learning task packages based on each nurse's blood purification operation assessment value and the weight coefficients of each stage, and push them to the nurse's end. After the nurse completes the learning task package, an assessment difference matrix before and after learning is constructed to perform a closed-loop update of course recommendations.

Images