Acute kidney injury risk assessment and early warning system

Abstract

The invention relates to the technical field of medical monitoring and early warning, in particular to an acute kidney injury risk assessment and early warning system which comprises a urine bioelectrical impedance analysis module, a data acquisition module, a data preprocessing module, a model construction module, a risk assessment module and an early warning module. Wherein the urine bioelectrical impedance analysis module is used for acquiring urine electrical impedance spectroscopy data; the data acquisition module is used for acquiring physiological parameters of a patient and laboratory inspection indexes; the data preprocessing module is used for extracting urine electrical impedance characteristic parameters, physiological characteristic parameters and laboratory characteristic parameters; the model construction module is used for constructing an acute kidney injury risk assessment model; the risk assessment module is used for calculating the risk score and the risk grade of the acute kidney injury of the patient. According to the system and the method, early-stage accurate evaluation and graded early warning of the acute kidney injury are realized through multi-source data fusion and intelligent analysis technologies, timely intervention of medical staff is facilitated, and the accuracy and response efficiency of risk detection are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of medical monitoring and early warning technology, and in particular to an acute kidney injury risk assessment and early warning system. Background Art

[0002] With the advancement of modern medicine, acute kidney injury (AKI) has become one of the common complications in clinical practice, with high morbidity and mortality, especially in intensive care patients. Early detection and warning of acute kidney injury are of great significance to the treatment and prognosis of patients.

[0003] Existing AKI detection methods mainly rely on biochemical indicators such as serum creatinine concentration and urine volume, but the changes in these indicators usually lag behind the actual occurrence of renal injury, making it difficult to effectively reflect early pathological changes, causing patients to miss the best time for intervention. In addition, the occurrence of acute kidney injury is related to multiple factors, including multi-dimensional data such as physiological characteristics, laboratory test indicators and urine characteristics. Existing evaluation methods lack the ability to comprehensively analyze these data, making it difficult to accurately identify high-risk patients.

[0004] Therefore, how to use advanced data processing and analysis techniques and combine multi-source data to improve the detection sensitivity and prediction accuracy of AKI has become a technical problem that needs to be solved urgently. Summary of the invention

[0005] Based on the above objectives, the present invention provides an acute kidney injury risk assessment and early warning system.

[0006] The acute kidney injury risk assessment and early warning system includes a urine bioelectrical impedance analysis module, a data acquisition module, a data preprocessing module, a model building module, a risk assessment module and an early warning module; among which:

[0007] Urine bioelectrical impedance analysis module: used to measure the bioelectrical impedance of the patient's urine sample and obtain urine electrical impedance spectrum data reflecting the physical and chemical properties of urine;

[0008] Data acquisition module: connected to the urine bioelectrical impedance analysis module, receiving the urine electrical impedance spectrum data, and collecting the patient's physiological parameters and laboratory test indicators, the physiological parameters include blood pressure, heart rate, blood oxygen saturation, and the laboratory test indicators include serum creatinine and urea nitrogen;

[0009] Data preprocessing module: connected to the data acquisition module, used to receive urine electrical impedance spectrum data, physiological parameters and laboratory test indicators, and perform denoising, filtering and standardization processing, and then extract urine electrical impedance characteristic parameters, physiological characteristic parameters and laboratory characteristic parameters;

[0010] Model building module: Based on the pre-established training data set, the support vector machine algorithm is used to build an acute kidney injury risk assessment model to identify the association between characteristic parameters and the risk of acute kidney injury;

[0011] Risk assessment module: connected with the model building module and the data preprocessing module, it calculates the risk score and risk level of acute kidney injury in patients based on the acute kidney injury risk assessment model and the characteristic parameters of each patient extracted by the data preprocessing module;

[0012] Early warning module: connected to the risk assessment module, used to generate corresponding early warning signals according to the risk level.

[0013] Optionally, the urine bioelectrical impedance analysis module includes an electrical impedance measurement unit, a signal processing unit, a data conversion unit and a calibration unit; wherein:

[0014] An electrical impedance measurement unit, used for applying a multi-frequency alternating current with a frequency range of 1 kHz to 1 MHz and a current intensity of 1 μA to 10 μA to a urine sample of a patient, and detecting a voltage response of the urine sample to obtain multi-frequency electrical impedance data;

[0015] A signal processing unit is connected to the electrical impedance measurement unit, and is used to receive the multi-frequency electrical impedance data, and to amplify the electrical impedance data by 50 times and perform bandpass filtering to eliminate background noise in the signal;

[0016] A data conversion unit is connected to the signal processing unit and is used to convert the amplified and filtered electrical impedance signal into a digital signal and generate electrical impedance spectrum data with a frequency resolution of 1 Hz;

[0017] Calibration unit: connected to the electrical impedance measurement unit, used for automatically calibrating before each measurement, setting the measurement error range within ±0.1%.

[0018] Optionally, the data acquisition module includes a physiological parameter acquisition unit, a laboratory index acquisition unit and a data transmission unit; wherein:

[0019] Physiological parameter acquisition unit: used to acquire the patient's physiological parameters in real time through a physiological monitoring device connected to the patient;

[0020] The physiological parameter acquisition unit comprises:

[0021] Blood pressure monitoring subunit: used to collect the patient's systolic and diastolic blood pressure values ​​once per minute in millimeters of mercury;

[0022] Heart rate monitoring subunit: used to collect the patient's heart rate value at a frequency of once per second, in units of beats per minute;

[0023] Blood oxygen saturation monitoring subunit: used to collect the patient's blood oxygen saturation value once per second in percentage;

[0024] Laboratory index collection unit: used to receive patient test index data recorded in the laboratory test system and automatically update it after each test of the patient is completed;

[0025] The laboratory index collection unit comprises:

[0026] Serum creatinine collection subunit: used to obtain the patient's serum creatinine level in micromoles per liter;

[0027] Urea nitrogen collection subunit: used to obtain the patient's urea nitrogen concentration in millimoles per liter;

[0028] Data transmission unit: connected to the physiological parameter acquisition unit and the laboratory index acquisition unit, used to transmit the physiological parameters and laboratory test indicators collected in real time to the data preprocessing module, and match them according to the patient's personal information to ensure the unique correspondence between the collected data and the patient's identity.

[0029] Optionally, the data preprocessing module includes a data denoising unit, a filtering unit, a standardization unit and a feature extraction unit; wherein:

[0030] Data denoising unit: used to receive urine electrical impedance spectrum data, physiological parameters and laboratory test indicators, and perform independent denoising on each data; specifically, for urine electrical impedance spectrum data, wavelet denoising method is used to remove high-frequency background noise; for physiological parameter data, moving average filtering algorithm is used to smooth short-term fluctuations; for laboratory test indicator data, threshold elimination method is used to exclude abnormal data points;

[0031] Filtering unit: connected to the data denoising unit, used to perform bandpass filtering on the denoised data to retain valid information within a predetermined frequency range; for urine electrical impedance spectrum data, retain electrical impedance signals within a frequency range of 1kHz to 1MHz; for physiological parameters, retain valid heart rate fluctuations within a range of 0.5Hz to 2.5Hz; laboratory test indicators, no filtering is required;

[0032] Standardization unit: connected to the filtering unit, used to standardize the denoised and filtered data and convert various types of data into a unified dimensionless form;

[0033] Feature extraction unit: connected to the standardization unit, used to extract urine electrical impedance feature parameters, physiological feature parameters and laboratory feature parameters.

[0034] Optionally, the feature extraction unit includes:

[0035] Urine electrical impedance feature extraction subunit: used to extract peak conductivity, average dielectric constant and phase angle change from urine electrical impedance spectrum data;

[0036] Physiological feature extraction subunit: used to extract heart rate variability, mean blood pressure and blood oxygen saturation fluctuation amplitude from physiological parameters;

[0037] Laboratory feature extraction unit: used to extract the rate of change of serum creatinine concentration and the fluctuation range of urea nitrogen concentration from laboratory test indicators.

[0038] Optionally, the model building module includes a data input unit, a data segmentation unit, a support vector machine training unit, a model verification unit and a model storage unit; wherein:

[0039] Data input unit: used to receive a pre-established training data set, wherein the training data set includes a plurality of sample data, each of which is composed of urine electrical impedance characteristic parameters, physiological characteristic parameters and laboratory characteristic parameters, and each of which is marked with a label indicating whether acute kidney injury has occurred;

[0040] Data segmentation unit: connected to the data input unit, used to divide the training data set into a training set and a validation set, wherein the training set accounts for 80% of the training data set and is used for parameter training of the model; the validation set accounts for 20% of the training data set and is used for performance verification of the model;

[0041] A support vector machine training unit is connected to the data segmentation unit, and based on the training set, adopts a support vector machine algorithm to establish a risk assessment model for distinguishing different acute kidney injury risk levels by optimizing the position and direction of the classification hyperplane;

[0042] Model verification unit: connected to the support vector machine training unit, used to verify the model using the verification set after the model training is completed, evaluate the accuracy, sensitivity and specificity of the model, and adjust the hyperparameters to optimize the model performance;

[0043] Model storage unit: used to store validated acute kidney injury risk assessment models for subsequent risk assessment module calls.

[0044] Optionally, the support vector machine training unit includes:

[0045] Data preprocessing: Normalize the characteristic parameters in the training set samples, convert the urine electrical impedance characteristic parameters, physiological characteristic parameters and laboratory characteristic parameters into a unified numerical range to ensure that the dimensions of each characteristic parameter are consistent during the algorithm training process;

[0046] Feature selection: Select feature parameters associated with the risk of acute kidney injury from the normalized training set features to form a feature vector;

[0047] Establish a support vector machine classification model: Set a linear support vector machine model and maximize the interval from the sample to the hyperplane by finding the classification hyperplane to achieve the best classification accuracy;

[0048] Solving the optimal hyperplane: By introducing Lagrange multipliers, the optimization problem is transformed into a dual problem to simplify the calculation process;

[0049] Generate classification hyperplane: Generate the optimal classification hyperplane based on the optimal Lagrange multiplier obtained, and then determine the parameters of the acute kidney injury risk assessment model.

[0050] Optionally, the model verification unit specifically includes:

[0051] Model prediction: Input the samples in the validation set into the trained AKI risk assessment model to obtain the prediction results for each sample, that is, the prediction of the AKI risk level for each sample;

[0052] Confusion matrix construction: According to the prediction results of the model and the true labels of the samples in the validation set, a confusion matrix is ​​constructed. The confusion matrix includes four indicators, namely true positives, false positives, true negatives and false negatives. Among them, true positives represent the number of samples in which the model correctly predicts acute kidney injury; false positives represent the number of samples in which the model incorrectly predicts acute kidney injury; true negatives represent the number of samples in which the model correctly predicts non-acute kidney injury; and false negatives represent the number of samples in which the model incorrectly predicts non-acute kidney injury.

[0053] Performance index calculation: Based on the confusion matrix, the accuracy, sensitivity and specificity of the model were calculated; accuracy represents the ratio of all correctly predicted samples to the total number of samples; sensitivity represents the ratio of the number of samples correctly predicted as acute kidney injury to the number of actual acute kidney injury samples; specificity represents the ratio of the number of samples correctly predicted as non-acute kidney injury to the number of actual non-acute kidney injury samples;

[0054] Hyperparameter tuning: Based on the results of accuracy, sensitivity, and specificity, hyperparameter optimization is performed.

[0055] Optionally, the risk assessment module includes a feature input unit, a risk score calculation unit, and a risk level determination unit; wherein:

[0056] Feature input unit: used to receive various patient feature parameters extracted by the data preprocessing module, including urine electrical impedance feature parameters, physiological feature parameters and laboratory feature parameters, and input the feature parameters into the acute kidney injury risk assessment model for calculation;

[0057] A risk score calculation unit: connected to the feature input unit, and used to calculate the feature parameters of the patient according to the acute kidney injury risk assessment model to generate a risk score;

[0058] Risk level determination unit: connected to the risk score calculation unit, used to classify the patient's risk score into different risk levels according to a preset risk score threshold. The risk levels are divided according to the following rules:

[0059] Rule 1: When the risk score is less than the threshold T low When , the patient is judged to be low risk;

[0060] Rule 2: When the risk score is above the threshold T low and T When the risk is between 1 and 2, the patient is considered to be at medium risk;

[0061] Rule 3: When the risk score is greater than or equal to the threshold When the patient is judged to be high risk;

[0062] Among them, T low and are risk score thresholds preset based on clinical data, indicating the score limits of low risk and high risk respectively.

[0063] Optionally, the warning module includes a warning signal generating unit and a signal output unit; wherein:

[0064] Warning signal generation unit: used to generate corresponding warning signals according to the patient's risk level; specifically, when the risk level is low risk, a green warning signal will be generated to remind the patient that the current risk of acute kidney injury is low and no immediate intervention is required, but regular monitoring is recommended; when the risk level is medium risk, a yellow warning signal will be generated to remind medical staff to pay attention to the patient's condition and recommend increasing the monitoring frequency; when the risk level is high risk, a red warning signal will be generated to warn medical staff that the patient is at high risk of acute kidney injury and that immediate detailed examination and clinical intervention are recommended;

[0065] Signal output unit: connected to the warning signal generating unit, used to send the generated warning signal to the terminal device of the medical staff; the signal output unit also includes distinguishing different levels of warning signals by corresponding colors and prompt sounds.

[0066] Beneficial effects of the present invention:

[0067] The present invention, through multi-source data fusion and intelligent analysis technology, can effectively identify the risk of acute kidney injury in the early stage. The system combines urine bioimpedance characteristics, physiological parameters and laboratory test indicators, and uses machine learning algorithms to build a risk assessment model to achieve a comprehensive assessment of the patient's acute kidney injury risk, significantly improving the accuracy and sensitivity of risk prediction. Compared with the existing single-indicator detection method, it can more comprehensively reflect the patient's health status, capture early pathological changes, avoid the shortcomings of traditional detection lags, and provide efficient and accurate risk identification support for clinical practice.

[0068] The present invention automatically generates corresponding warning signals according to different risk levels through the warning module, which is convenient for medical staff to quickly identify the patient's acute kidney injury risk status. The system provides graded warnings between low risk, medium risk and high risk, combined with intuitive colors and prompt sounds, and pushes them in real time through the medical staff's terminal device, which helps medical staff to intervene in time and take necessary measures, thereby effectively managing patients in the early stages. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0070] Figure 1 Schematic diagram of an acute kidney injury risk assessment and early warning system according to an embodiment of the present invention;

[0071] Figure 2 Schematic diagram of a model building module according to an embodiment of the present invention. DETAILED DESCRIPTION

[0072] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0073] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0074] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0075] like Figure 1-Figure 2 As shown, the acute kidney injury risk assessment and early warning system includes a urine bioelectrical impedance analysis module, a data acquisition module, a data preprocessing module, a model building module, a risk assessment module and an early warning module; wherein:

[0076] Urine bioelectrical impedance analysis module: used to measure the bioelectrical impedance of the patient's urine sample and obtain urine electrical impedance spectrum data reflecting the physical and chemical properties of urine;

[0077] Data acquisition module: connected to the urine bioelectrical impedance analysis module, receives urine electrical impedance spectrum data, and collects the patient's physiological parameters and laboratory test indicators. The physiological parameters include blood pressure, heart rate, and blood oxygen saturation, and the laboratory test indicators include serum creatinine and urea nitrogen;

[0078] Data preprocessing module: connected to the data acquisition module, used to receive urine electrical impedance spectrum data, physiological parameters and laboratory test indicators, and perform denoising, filtering and standardization processing, and then extract urine electrical impedance characteristic parameters, physiological characteristic parameters and laboratory characteristic parameters;

[0079] Model building module: Based on the pre-established training data set, the support vector machine algorithm is used to build an acute kidney injury risk assessment model to identify the association between characteristic parameters and the risk of acute kidney injury;

[0080] Risk assessment module: connected with the model building module and the data preprocessing module, it calculates the risk score and risk level of acute kidney injury in patients based on the acute kidney injury risk assessment model and the characteristic parameters of each patient extracted by the data preprocessing module;

[0081] Early warning module: connected with the risk assessment module, used to generate corresponding early warning signals according to the risk level.

[0082] The urine bioelectrical impedance analysis module includes an electrical impedance measurement unit, a signal processing unit, a data conversion unit and a calibration unit; wherein:

[0083] An electrical impedance measurement unit, used for applying a multi-frequency alternating current with a frequency range of 1 kHz to 1 MHz and a current intensity of 1 μA to 10 μA to a urine sample of a patient, and detecting a voltage response of the urine sample to obtain multi-frequency electrical impedance data, wherein the electrical impedance data reflects the electrical conductivity and dielectric properties of the urine;

[0084] Signal processing unit: connected to the electrical impedance measurement unit, used to receive multi-frequency electrical impedance data, and perform 50 times amplification and bandpass filtering on the electrical impedance data to eliminate background noise in the signal and ensure the accuracy of the data;

[0085] Data conversion unit: connected to the signal processing unit, used to convert the amplified and filtered electrical impedance signal into a digital signal, and generate electrical impedance spectrum data with a frequency resolution of 1 Hz to reflect the physical and chemical properties of urine;

[0086] Calibration unit: connected to the impedance measurement unit, used for automatic calibration before each measurement, setting the measurement error range within ±0.1%, ensuring the accuracy consistency of the impedance measurement device and the reliability of the measurement results; through the mutual cooperation of the above units, the urine bioimpedance analysis module can perform accurate multi-frequency impedance measurement of patient urine samples in the frequency range of 1kHz to 1MHz, and generate high-precision impedance spectrum data reflecting urine conductivity and dielectric constant, providing scientific and reliable basic data support for early risk assessment of acute kidney injury.

[0087] The data acquisition module includes a physiological parameter acquisition unit, a laboratory index acquisition unit and a data transmission unit; wherein:

[0088] Physiological parameter acquisition unit: used to acquire the patient's physiological parameters in real time through a physiological monitoring device connected to the patient;

[0089] The physiological parameter acquisition unit includes:

[0090] Blood pressure monitoring subunit: used to collect the patient's systolic and diastolic blood pressure values ​​once per minute in millimeters of mercury (mmHg);

[0091] Heart rate monitoring subunit: used to collect the patient's heart rate value at a frequency of once per second, in units of beats per minute;

[0092] Blood oxygen saturation monitoring subunit: used to collect the patient's blood oxygen saturation value once per second, in percentage (%);

[0093] Laboratory index collection unit: used to receive patient test index data recorded in the laboratory test system and automatically update it after each test of the patient is completed;

[0094] The laboratory index collection unit includes:

[0095] Serum creatinine collection subunit: used to obtain the patient's serum creatinine level in micromoles per liter (μmol / L);

[0096] Urea nitrogen collection subunit: used to obtain the patient's urea nitrogen concentration in millimoles per liter (mmol / L);

[0097] Data transmission unit: connected to the physiological parameter collection unit and the laboratory index collection unit, used to transmit the real-time collected physiological parameters and laboratory test indicators to the data preprocessing module, and match them according to the patient's personal information to ensure the unique correspondence between the collected data and the patient's identity; through the cooperation of the above units, the data acquisition module can stably and accurately collect the patient's physiological parameters and laboratory test indicator data, and automatically transmit them to the data preprocessing module in the form of matching tags, providing basic data support for subsequent risk assessment and realizing real-time monitoring of the patient's health status.

[0098] The data preprocessing module includes a data denoising unit, a filtering unit, a normalization unit and a feature extraction unit; wherein:

[0099] Data denoising unit: used to receive urine electrical impedance spectrum data, physiological parameters and laboratory test indicators, and perform independent denoising on each data; specifically for urine electrical impedance spectrum data, wavelet denoising method is used to remove high-frequency background noise; for physiological parameter data, moving average filtering algorithm is used to smooth short-term fluctuations; for laboratory test indicator data, threshold elimination method is used to exclude abnormal data points to ensure data accuracy;

[0100] Filtering unit: connected to the data denoising unit, used to perform bandpass filtering on the denoised data to retain valid information within a predetermined frequency range; for urine electrical impedance spectrum data, retain electrical impedance signals within a frequency range of 1kHz to 1MHz; for physiological parameters, retain valid heart rate fluctuations within a range of 0.5Hz to 2.5Hz; laboratory test indicators, without filtering, are directly output to the next unit;

[0101] Standardization unit: connected to the filtering unit, used to standardize the denoised and filtered data, converting various types of data into a unified dimensionless form for subsequent analysis; urine electrical impedance spectrum data is standardized by the maximum-minimum normalization method; physiological parameters are standardized by Z-score; laboratory test indicators are scaled according to the conventional numerical range;

[0102] Feature extraction unit: connected to the standardization unit, used to extract urine electrical impedance characteristic parameters, physiological characteristic parameters and laboratory characteristic parameters; through the sequential processing of the above units, the data preprocessing module can denoise, filter and standardize the urine electrical impedance spectrum data, physiological parameters and laboratory test indicators, and extract characteristic parameters reflecting the patient's health status, providing consistency and accuracy of basic data support for subsequent risk assessment.

[0103] The feature extraction unit includes:

[0104] Urine electrical impedance feature extraction subunit: used to extract peak conductivity, average dielectric constant and phase angle change from urine electrical impedance spectrum data;

[0105] Assume the peak conductivity is G max , the calculation formula is Among them, I max is the peak value of the AC current in the electrical impedance spectrum, V min is the minimum value of voltage;

[0106] Assume the average dielectric constant is ε avg , the calculation formula is: Among them, ε i represents the dielectric constant at frequency i, and n is the number of frequency points;

[0107] Assume that the phase angle change is Δθ, and the calculation formula is: Δθ=θ max -θ min , where θ max and θ min are the maximum and minimum phase angles, respectively, used to analyze the changes in electrochemical properties in urine samples;

[0108] Physiological feature extraction subunit: used to extract heart rate variability, mean blood pressure and blood oxygen saturation fluctuation amplitude from physiological parameters;

[0109] Assuming that heart rate variability is HRV, based on the heart rate sequence, the standard deviation formula is used for calculation: in, is the jth heart rate value, is the average heart rate, N is the total number of samples;

[0110] Assume the average blood pressure is The calculation formula is: in, is the blood pressure measurement value per minute, and M is the number of measurements;

[0111] Assume that the oxygen saturation fluctuation range is The calculation formula is: in, and are the maximum and minimum values ​​of blood oxygen saturation respectively;

[0112] Laboratory feature extraction unit: used to extract the change rate of serum creatinine concentration and the fluctuation range of urea nitrogen concentration from laboratory test indicators;

[0113] Assume that the rate of change of serum creatinine concentration is The calculation formula is: Among them, C t is the current serum creatinine concentration, is the baseline serum creatinine concentration, and t is the detection time interval;

[0114] Assuming the fluctuation range of urea nitrogen concentration is ΔBUN, the calculation formula is: in, and They are the maximum and minimum values ​​of urea nitrogen concentration respectively; through the combination of the above sub-units, characteristic parameters reflecting the patient's health status can be extracted from urine electrical impedance spectroscopy data, physiological parameters and laboratory test indicators, providing multi-dimensional quantitative indicators for subsequent risk assessment modules.

[0115] The model building module includes a data input unit, a data segmentation unit, a support vector machine training unit, a model verification unit and a model storage unit; wherein:

[0116] Data input unit: used to receive a pre-established training data set, the training data set includes a plurality of sample data, each of which is composed of urine electrical impedance characteristic parameters, physiological characteristic parameters and laboratory characteristic parameters, and each of which is marked with a label indicating whether acute kidney injury has occurred;

[0117] Data segmentation unit: connected to the data input unit, used to divide the training data set into a training set and a validation set. The training set accounts for 80% of the training data set and is used for model parameter training; the validation set accounts for 20% of the training data set and is used for model performance verification to ensure the generalization ability of the model.

[0118] Support vector machine training unit: connected to the data segmentation unit, based on the training set, using the support vector machine algorithm, by optimizing the position and direction of the classification hyperplane, to establish a risk assessment model for distinguishing different acute kidney injury risk levels. The support vector machine training unit finds the best separation boundary according to the characteristic values ​​of various characteristic parameters in the training set samples to achieve high-precision classification effect;

[0119] Model validation unit: connected to the support vector machine training unit, used to validate the model using the validation set after model training is completed, evaluate the accuracy, sensitivity and specificity of the model, and adjust the hyperparameters to optimize the model performance to ensure the accuracy of the model's prediction of the risk of acute kidney injury;

[0120] Model storage unit: used to store verified acute kidney injury risk assessment models for subsequent risk assessment modules to call and achieve rapid assessment of patients' acute kidney injury risk.

[0121] The support vector machine training unit includes:

[0122] Data preprocessing: Normalize the characteristic parameters in the training set samples, convert the urine electrical impedance characteristic parameters, physiological characteristic parameters and laboratory characteristic parameters into a unified numerical range to ensure that the dimensions of each characteristic parameter are consistent during the algorithm training process;

[0123] Feature selection: From the normalized training set features, select feature parameters associated with the risk of acute kidney injury to form a feature vector. During the process, statistical analysis is performed to ensure that the selected features contribute significantly to the classification effect.

[0124] Establish a support vector machine classification model: Set a linear support vector machine model and maximize the interval from the sample to the hyperplane by finding the classification hyperplane to achieve the best classification accuracy;

[0125] Solving the optimal hyperplane: By introducing Lagrange multipliers, the optimization problem is transformed into a dual problem to simplify the calculation process;

[0126] Generate classification hyperplane: Generate the optimal classification hyperplane based on the optimal Lagrange multiplier obtained, and then determine the parameters of the acute kidney injury risk assessment model.

[0127] The calculation steps for constructing the acute kidney injury risk assessment model are as follows:

[0128] First, the characteristic parameters in the training set samples were normalized, and the urine electrical impedance characteristic parameters, physiological characteristic parameters, and laboratory characteristic parameters were converted into a unified numerical range to ensure the dimensional consistency of the characteristic parameters during the algorithm training process; the normalization formula is: Among them, x is the original eigenvalue, x min is the minimum eigenvalue, x max is the maximum value of the eigenvalue, and x′ is the normalized eigenvalue;

[0129] Next, based on the normalized training set features, the characteristic parameters associated with the risk of acute kidney injury are selected to form a feature vector, which is expressed as: x = [x1, x2, ..., x n ], where xi is the i-th feature selected, and n is the total number of features;

[0130] Then, a linear support vector machine model is set to maximize the interval from the sample to the hyperplane by finding the classification hyperplane w·x+b=0 to achieve the best classification accuracy. The optimization problem formula is: Under the constraint y i (w·x i +b)≥1, where y i is the category label of the i-th sample, x i is the feature vector of the i-th sample, w is the normal vector of the hyperplane, and b is the offset of the hyperplane;

[0131] Then, the Lagrange multiplier α is introduced by the Lagrange multiplier method. i , transform the optimization problem into a dual problem to simplify the calculation; the solution formula of the dual problem is:

[0132] Under the constraints And 0≤α i ≤C, where m is the number of training samples and C is the penalty parameter used to control the tolerance to misclassification;

[0133] Finally, the optimal Lagrange multiplier α is obtained i Substitute into the hyperplane equation to generate the optimal classification hyperplane, and then obtain the model parameters w and b; the classification decision function is: Among them, f(x) is the classification result used to distinguish the risk level of acute kidney injury, sign represents the sign function, which is used to determine the category of the sample; through the above steps, the support vector machine training unit can build a high-precision acute kidney injury risk assessment model based on the training set using the support vector machine algorithm, generate the optimal classification hyperplane, and use the model parameters for subsequent patient risk assessment.

[0134] The model verification unit specifically includes:

[0135] Model prediction: Input the samples in the validation set into the trained AKI risk assessment model to obtain the prediction results for each sample, that is, the prediction of the AKI risk level for each sample;

[0136] Confusion matrix construction: Based on the prediction results of the model and the true labels of the samples in the validation set, a confusion matrix is ​​constructed. The confusion matrix includes four indicators, namely true positives, false positives, true negatives, and false negatives. Among them, true positives represent the number of samples in which the model correctly predicts acute kidney injury; false positives represent the number of samples in which the model incorrectly predicts acute kidney injury; true negatives represent the number of samples in which the model correctly predicts non-acute kidney injury; and false negatives represent the number of samples in which the model incorrectly predicts non-acute kidney injury.

[0137] Performance index calculation: Based on the confusion matrix, the accuracy, sensitivity and specificity of the model were calculated; accuracy represents the ratio of all correctly predicted samples to the total number of samples; sensitivity represents the ratio of the number of samples correctly predicted as acute kidney injury to the number of actual acute kidney injury samples; specificity represents the ratio of the number of samples correctly predicted as non-acute kidney injury to the number of actual non-acute kidney injury samples;

[0138] Hyperparameter adjustment: Based on the results of accuracy, sensitivity, and specificity, hyperparameter optimization is performed. Specifically, the grid search method is used to select the best combination of penalty parameters and kernel width parameters in the support vector machine model within the preset parameter range.

[0139] The specific calculation steps for using the validation set to validate the model are as follows:

[0140] The accuracy of the model is calculated based on the confusion matrix, which is defined as the ratio of the number of correctly predicted samples to the total number of samples. The accuracy formula is: Among them, TP, FP, TN and FN are true positives, false positives, true negatives and false negatives respectively;

[0141] The sensitivity of the model was calculated based on the confusion matrix, which was defined as the ratio of the number of samples correctly predicted to be acute kidney injury to the number of actual acute kidney injury samples. The sensitivity formula was:

[0142]

[0143] The specificity of the model was calculated based on the confusion matrix, which was defined as the ratio of the number of samples correctly predicted as non-acute kidney injury to the number of actual non-acute kidney injury samples. The specificity formula was:

[0144]

[0145] Based on the calculation results of accuracy, sensitivity and specificity, hyperparameter optimization is performed; specifically, the grid search method is used to select the best combination of penalty parameter C and kernel width parameter γ in the support vector machine model within the preset parameter range; the parameter range C∈[C min , C max] and γ∈[γ min , γ max ], and perform traversal testing with fixed step sizes ΔC and Δγ within this range. The specific steps are as follows:

[0146] Step 1: For each pair of parameter combinations (C, γ), the accuracy, sensitivity, and specificity of the model are calculated using the validation set, and the results are compared with the performance of the current best parameter combination.

[0147] Step 2: If the model performance under the current combination (C, γ) is better than the previous best combination, the best combination is updated to the current combination;

[0148] Step 3: The optimal parameter combination is finally determined to meet the following conditions: (C * , γ * ) = arg max C,γ (Accuracy×Sensitivity×Specificity), where C * and γ * represents the optimal penalty parameter and kernel width parameter; through the above steps, the model validation unit can validate the acute kidney injury risk assessment model based on the validation set, evaluate its accuracy, sensitivity and specificity, and optimize the model performance through hyperparameter adjustment, thereby improving the reliability and accuracy of the model in predicting the risk of acute kidney injury.

[0149] The risk assessment module includes a feature input unit, a risk score calculation unit, and a risk level determination unit; wherein:

[0150] Feature input unit: used to receive various patient feature parameters extracted by the data preprocessing module, including urine electrical impedance feature parameters, physiological feature parameters and laboratory feature parameters, and input the feature parameters into the acute kidney injury risk assessment model for calculation;

[0151] Risk score calculation unit: connected to the feature input unit, used to calculate the patient's feature parameters according to the acute kidney injury risk assessment model to generate a risk score; the specific calculation formula is as follows: Among them, Risk Score represents the patient's risk score, w i is the i-th characteristic parameter x in the acute kidney injury risk assessment model i The weight of , n is the total number of characteristic parameters, and b is the model offset; the risk score is used to quantify the risk level of acute kidney injury in patients;

[0152] Risk level determination unit: connected to the risk score calculation unit, used to classify the patient's risk score into different risk levels according to the preset risk score threshold. The risk levels are divided according to the following rules:

[0153] Rule 1: When the risk score is less than the threshold T low When , the patient is judged to be low risk;

[0154] Rule 2: When the risk score is above the threshold T low and When the risk is between 1 and 2, the patient is considered to be at medium risk;

[0155] Rule 3: When the risk score is greater than or equal to the threshold When the patient is judged to be high risk;

[0156] Among them, T low and The risk score thresholds are preset according to clinical data, indicating the score limits of low risk and high risk respectively. Through the synergistic effect of the above units, the risk assessment module can accurately calculate the patient's risk score based on the acute kidney injury risk assessment model and the patient's characteristic parameters, and divide it into corresponding risk levels, thereby providing a reliable risk judgment basis for clinical intervention.

[0157] The early warning module includes an early warning signal generating unit and a signal output unit; wherein:

[0158] Warning signal generation unit: used to generate corresponding warning signals according to the patient's risk level; specifically, when the risk level is low risk, a green warning signal will be generated to remind the patient that the current risk of acute kidney injury is low and no immediate intervention is required, but regular monitoring is recommended; when the risk level is medium risk, a yellow warning signal will be generated to remind medical staff to pay attention to the patient's condition and recommend increasing the monitoring frequency; when the risk level is high risk, a red warning signal will be generated to warn medical staff that the patient is at high risk of acute kidney injury and that immediate detailed examination and clinical intervention are recommended;

[0159] Signal output unit: connected to the warning signal generation unit, used to send the generated warning signal to the terminal device of medical staff; the signal output unit also includes distinguishing different levels of warning signals through corresponding colors and prompt sounds, so that medical staff can quickly identify the patient's acute kidney injury risk level and respond in time.

[0160] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0161] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An acute kidney injury risk assessment and early warning system, characterized in that: It includes urine bioelectrical impedance analysis module, data acquisition module, data preprocessing module, model building module, risk assessment module and early warning module; among which: Urine bioelectrical impedance analysis module: used to measure the bioelectrical impedance of the patient's urine sample and obtain urine electrical impedance spectrum data reflecting the physical and chemical properties of urine; Data acquisition module: connected to the urine bioelectrical impedance analysis module, receiving the urine electrical impedance spectrum data, and collecting the patient's physiological parameters and laboratory test indicators, the physiological parameters include blood pressure, heart rate, blood oxygen saturation, and the laboratory test indicators include serum creatinine and urea nitrogen; Data preprocessing module: connected to the data acquisition module, used to receive urine electrical impedance spectrum data, physiological parameters and laboratory test indicators, and perform denoising, filtering and standardization processing, and then extract urine electrical impedance characteristic parameters, physiological characteristic parameters and laboratory characteristic parameters; Model building module: Based on the pre-established training data set, the support vector machine algorithm is used to build an acute kidney injury risk assessment model to identify the association between characteristic parameters and the risk of acute kidney injury; Risk assessment module: connected with the model building module and the data preprocessing module, it calculates the risk score and risk level of acute kidney injury in patients based on the acute kidney injury risk assessment model and the characteristic parameters of each patient extracted by the data preprocessing module; Early warning module: connected to the risk assessment module, used to generate corresponding early warning signals according to the risk level.

2. The acute kidney injury risk assessment and early warning system according to claim 1, characterized in that: The urine bioelectrical impedance analysis module includes an electrical impedance measurement unit, a signal processing unit, a data conversion unit and a calibration unit; wherein: An electrical impedance measurement unit, used for applying a multi-frequency alternating current with a frequency range of 1 kHz to 1 MHz and a current intensity of 1 μA to 10 μA to a urine sample of a patient, and detecting a voltage response of the urine sample to obtain multi-frequency electrical impedance data; A signal processing unit is connected to the electrical impedance measurement unit, and is used to receive the multi-frequency electrical impedance data, and to amplify the electrical impedance data by 50 times and perform bandpass filtering to eliminate background noise in the signal; A data conversion unit is connected to the signal processing unit and is used to convert the amplified and filtered electrical impedance signal into a digital signal and generate electrical impedance spectrum data with a frequency resolution of 1 Hz; Calibration unit: connected to the electrical impedance measurement unit, used for automatically calibrating before each measurement, setting the measurement error range within ±0.1%.

3. The acute kidney injury risk assessment and early warning system according to claim 1, characterized in that: The data acquisition module includes a physiological parameter acquisition unit, a laboratory index acquisition unit and a data transmission unit; wherein: Physiological parameter acquisition unit: used to acquire the patient's physiological parameters in real time through a physiological monitoring device connected to the patient; The physiological parameter acquisition unit comprises: Blood pressure monitoring subunit: used to collect the patient's systolic and diastolic blood pressure values ​​once per minute in millimeters of mercury; Heart rate monitoring subunit: used to collect the patient's heart rate value at a frequency of once per second, in units of beats per minute; Blood oxygen saturation monitoring subunit: used to collect the patient's blood oxygen saturation value once per second in percentage; Laboratory index collection unit: used to receive patient test index data recorded in the laboratory test system and automatically update it after each test of the patient is completed; The laboratory index collection unit comprises: Serum creatinine collection subunit: used to obtain the patient's serum creatinine level in micromoles per liter; Urea nitrogen collection subunit: used to obtain the patient's urea nitrogen concentration in millimoles per liter; Data transmission unit: connected to the physiological parameter acquisition unit and the laboratory index acquisition unit, used to transmit the physiological parameters and laboratory test indicators collected in real time to the data preprocessing module, and match them according to the patient's personal information to ensure the unique correspondence between the collected data and the patient's identity.

4. The acute kidney injury risk assessment and early warning system according to claim 1, characterized in that: The data preprocessing module includes a data denoising unit, a filtering unit, a standardization unit and a feature extraction unit; wherein: Data denoising unit: used to receive urine electrical impedance spectrum data, physiological parameters and laboratory test indicators, and perform independent denoising on each data; specifically, for urine electrical impedance spectrum data, wavelet denoising method is used to remove high-frequency background noise; for physiological parameter data, moving average filtering algorithm is used to smooth short-term fluctuations; for laboratory test indicator data, threshold elimination method is used to exclude abnormal data points; Filtering unit: connected to the data denoising unit, used to perform bandpass filtering on the denoised data to retain valid information within a predetermined frequency range; for urine electrical impedance spectrum data, retain electrical impedance signals within a frequency range of 1kHz to 1MHz; for physiological parameters, retain valid heart rate fluctuations within a range of 0.5Hz to 2.5Hz; laboratory test indicators, no filtering is required; Standardization unit: connected to the filtering unit, used to standardize the denoised and filtered data and convert various types of data into a unified dimensionless form; Feature extraction unit: connected to the standardization unit, used to extract urine electrical impedance feature parameters, physiological feature parameters and laboratory feature parameters.

5. The acute kidney injury risk assessment and early warning system according to claim 4, characterized in that: The feature extraction unit comprises: Urine electrical impedance feature extraction subunit: used to extract peak conductivity, average dielectric constant and phase angle change from urine electrical impedance spectrum data; Physiological feature extraction subunit: used to extract heart rate variability, mean blood pressure and blood oxygen saturation fluctuation amplitude from physiological parameters; Laboratory feature extraction unit: used to extract the rate of change of serum creatinine concentration and the fluctuation range of urea nitrogen concentration from laboratory test indicators.

6. The acute kidney injury risk assessment and early warning system according to claim 1, characterized in that: The model building module includes a data input unit, a data segmentation unit, a support vector machine training unit, a model verification unit and a model storage unit; wherein: Data input unit: used to receive a pre-established training data set, wherein the training data set includes a plurality of sample data, each of which is composed of urine electrical impedance characteristic parameters, physiological characteristic parameters and laboratory characteristic parameters, and each of which is marked with a label indicating whether acute kidney injury has occurred; Data segmentation unit: connected to the data input unit, used to divide the training data set into a training set and a validation set, wherein the training set accounts for 80% of the training data set and is used for parameter training of the model; the validation set accounts for 20% of the training data set and is used for performance verification of the model; A support vector machine training unit is connected to the data segmentation unit, and based on the training set, adopts a support vector machine algorithm to establish a risk assessment model for distinguishing different acute kidney injury risk levels by optimizing the position and direction of the classification hyperplane; Model verification unit: connected to the support vector machine training unit, used to verify the model using the verification set after the model training is completed, evaluate the accuracy, sensitivity and specificity of the model, and adjust the hyperparameters to optimize the model performance; Model storage unit: used to store validated acute kidney injury risk assessment models for subsequent risk assessment module calls.

7. The acute kidney injury risk assessment and early warning system according to claim 6, characterized in that: The support vector machine training unit comprises: Data preprocessing: Normalize the characteristic parameters in the training set samples, convert the urine electrical impedance characteristic parameters, physiological characteristic parameters and laboratory characteristic parameters into a unified numerical range to ensure that the dimensions of each characteristic parameter are consistent during the algorithm training process; Feature selection: Select feature parameters associated with the risk of acute kidney injury from the normalized training set features to form a feature vector; Establish a support vector machine classification model: Set a linear support vector machine model and maximize the interval from the sample to the hyperplane by finding the classification hyperplane to achieve the best classification accuracy; Solving the optimal hyperplane: By introducing Lagrange multipliers, the optimization problem is transformed into a dual problem to simplify the calculation process; Generate classification hyperplane: Generate the optimal classification hyperplane based on the optimal Lagrange multiplier obtained, and then determine the parameters of the acute kidney injury risk assessment model.

8. The acute kidney injury risk assessment and early warning system according to claim 7, characterized in that: The model verification unit specifically includes: Model prediction: Input the samples in the validation set into the trained AKI risk assessment model to obtain the prediction results for each sample, that is, the prediction of the AKI risk level for each sample; Confusion matrix construction: According to the prediction results of the model and the true labels of the samples in the validation set, a confusion matrix is ​​constructed. The confusion matrix includes four indicators, namely true positives, false positives, true negatives and false negatives. Among them, true positives represent the number of samples in which the model correctly predicts acute kidney injury; false positives represent the number of samples in which the model incorrectly predicts acute kidney injury; true negatives represent the number of samples in which the model correctly predicts non-acute kidney injury; and false negatives represent the number of samples in which the model incorrectly predicts non-acute kidney injury. Performance index calculation: Based on the confusion matrix, the accuracy, sensitivity and specificity of the model were calculated; accuracy represents the ratio of all correctly predicted samples to the total number of samples; sensitivity represents the ratio of the number of samples correctly predicted as acute kidney injury to the number of actual acute kidney injury samples; specificity represents the ratio of the number of samples correctly predicted as non-acute kidney injury to the number of actual non-acute kidney injury samples; Hyperparameter tuning: Based on the results of accuracy, sensitivity, and specificity, hyperparameter optimization is performed.

9. The acute kidney injury risk assessment and early warning system according to claim 1, characterized in that: The risk assessment module includes a feature input unit, a risk score calculation unit, and a risk level determination unit; wherein: Feature input unit: used to receive various patient feature parameters extracted by the data preprocessing module, including urine electrical impedance feature parameters, physiological feature parameters and laboratory feature parameters, and input the feature parameters into the acute kidney injury risk assessment model for calculation; A risk score calculation unit: connected to the feature input unit, and used to calculate the feature parameters of the patient according to the acute kidney injury risk assessment model to generate a risk score; Risk level determination unit: connected to the risk score calculation unit, used to classify the patient's risk score into different risk levels according to a preset risk score threshold. The risk levels are divided according to the following rules: Rule 1: When the risk score is less than the threshold T low When , the patient is judged to be low risk; Rule 2: When the risk score is above the threshold T low and T high When the risk is between 1 and 2, the patient is considered to be at medium risk; Rule 3: When the risk score is greater than or equal to the threshold T high When the patient is judged to be high risk; Among them, T low and T high are risk score thresholds preset based on clinical data, indicating the score limits of low risk and high risk respectively.

10. The acute kidney injury risk assessment and early warning system according to claim 1, characterized in that: The warning module includes a warning signal generating unit and a signal output unit; wherein: Warning signal generation unit: used to generate corresponding warning signals according to the patient's risk level; specifically, when the risk level is low risk, a green warning signal will be generated to remind the patient that the current risk of acute kidney injury is low and no immediate intervention is required, but regular monitoring is recommended; when the risk level is medium risk, a yellow warning signal will be generated to remind medical staff to pay attention to the patient's condition and recommend increasing the monitoring frequency; when the risk level is high risk, a red warning signal will be generated to warn medical staff that the patient is at high risk of acute kidney injury and that immediate detailed examination and clinical intervention are recommended; Signal output unit: connected to the warning signal generating unit, used to send the generated warning signal to the terminal device of the medical staff; the signal output unit also includes distinguishing different levels of warning signals by corresponding colors and prompt sounds.

Images