Acute kidney injury risk early warning system

Abstract

The invention discloses an acute kidney injury risk early warning system, and relates to the technical field of clinical decision support systems. In order to solve the problems that in the prior art, due to the fact that a large amount of data preprocessing and model training need long-time data processing and model operation, the early warning opportunity is not accurate, the real-time performance is low, personalized early warning prompting cannot be carried out for patient groups, and early warning fatigue and early warning desensitization are likely to be caused. Through data preprocessing, model construction and risk analysis, real-time monitoring and accurate prediction of the AKI risk of a patient are realized, the early recognition rate is improved, the early warning decision unit generates specific early warning information according to risk scores and kidney function change rules and provides targeted treatment suggestions, and meanwhile, through an early warning threshold adjustment and approach matching module, the early warning decision unit can perform early warning on the AKI risk of the patient. The early warning strategy is optimized, the response speed of medical staff and the compliance of patients are improved, and the user interaction unit continuously optimizes the early warning system through a feedback mechanism.

Description

Technical Field

[0001] The invention discloses an acute kidney injury risk early warning system, and relates to the technical field of clinical decision support systems. Background Art

[0002] At present, the early identification and early warning of acute kidney injury (AKI) mainly rely on the observation and analysis of clinical indicators by medical staff. For example, the patent application with the publication number: CN118299054A discloses a warning system for the occurrence of acute kidney injury in critically ill patients based on machine learning, including a data acquisition and integration module, a feature screening and determination module, an optimal prediction model screening and determination module, a risk prediction result output module, and a model interpretability analysis module; the data acquisition and integration module collects the baseline characteristics of the patient when entering the ICU and performs preprocessing; the feature screening and determination module has four submodules embedded in it, namely, baseline feature analysis, univariate logistic regression analysis, Pearson / Spearman correlation analysis, and feature importance ranking; the optimal prediction model screening and determination module includes two submodules, namely, optimal variable number determination and optimal model determination, which are used to clarify the minimum number of variables required for each machine learning model to achieve the best predictive performance; the model interpretability analysis module explains the risk of AKI for each patient and the contribution of key features based on the SHAP framework.

[0003] Although the above patent can assess the risk of AKI in patients earlier and more accurately, this method has certain limitations. It requires a lot of data preprocessing and model training, and takes a long time for data processing and model calculation, which leads to inaccurate warning timing and low real-time performance. It is also impossible to provide personalized warning prompts for patient groups, which can easily cause warning fatigue and warning desensitization. Summary of the invention

[0004] The purpose of the present invention is to provide an acute kidney injury risk warning system, which significantly improves the early identification and intervention efficiency of AKI, reduces complications and mortality, improves clinical prognosis, optimizes the allocation of medical resources, promotes the continuous improvement of medical quality, and provides patients with safer and more efficient medical services to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solution, an acute kidney injury risk early warning system, comprising:

[0006] The data collection unit is used to collect the patient's basic information, medical history, medication status, laboratory test results, and biomarker data, assign a unique identifier to the patient based on the patient's basic information, and build the patient's electronic health record;

[0007] The data analysis unit is used to pre-process the acquired electronic health records of patients, build an AKI risk assessment model, extract and combine laboratory test results and biomarker data in the electronic health records of patients, and predict the risk of AKI in patients based on the AKI risk assessment model;

[0008] The early warning decision unit is used to generate corresponding AKI early warning information based on the risk prediction results, adjust the early warning threshold based on the characteristics of the patient group, and match the corresponding AKI early warning information and notification channels based on the preferences of the early warning audience;

[0009] The user interaction unit is used to display the AKI warning information and historical warning records output by the warning decision unit in real time, and to establish a feedback mechanism to optimize and adjust the AKI warning information based on the feedback mechanism.

[0010] Furthermore, the data acquisition unit comprises:

[0011] The multi-source data integration module is used to establish a multi-source heterogeneous data interface with the hospital's various electronic medical record systems, laboratory information systems, and medical imaging systems, extract key patient information based on the multi-source heterogeneous data interface, and standardize the data format of key information in different data sources;

[0012] A patient data packet generation module is used to classify and package the acquired key information based on the data source, generate a patient data packet, and assign a unique data packet identifier to each patient data packet, corresponding to the patient's unique identifier;

[0013] The electronic health record management module is used to match the patient's basic data, medical history, medication status, laboratory test results and biomarker data based on the data source; based on the above relationship, when a unique identifier is entered, the corresponding patient data package can be retrieved at the same time to build the patient's electronic health record.

[0014] Furthermore, the data analysis unit comprises:

[0015] The data preprocessing module is used to clean, normalize and standardize the acquired patient electronic health record data, and extract key features related to AKI risk assessment from the processed patient electronic health record data;

[0016] A model building module is used to input the extracted key features related to AKI risk assessment into a preset network model to obtain an AKI risk assessment model for each patient;

[0017] The risk analysis module is used to input the newly acquired patient electronic health record data into the AKI risk assessment model for risk identification and output the patient's risk prediction results.

[0018] Furthermore, the data analysis unit further includes:

[0019] Extracting the patient's laboratory test results and biomarker data from the patient's electronic health record, and extracting time series data of each data carried by the laboratory test results and biomarker data;

[0020] Determine the patient's renal function characteristics based on the time series data of each data, obtain the change of the patient's renal function characteristics in the electronic health record based on the time series data, and determine the patient's renal function change rules;

[0021] Compare the renal function change rules of each patient, and integrate the patients whose similarity is greater than or equal to the preset threshold based on the comparison results to generate a group of similar patients;

[0022] The renal function change rules of any patient in each patient group are taken as the representative risk characteristics of the patient group and confirmed as the patient group characteristics.

[0023] Furthermore, the risk analysis module is also used to:

[0024] Combine the patient's risk prediction results with the corresponding patient population characteristics to assess the patient's likelihood of developing AKI, and quantify the results to obtain the patient's risk score;

[0025] Comparing the patient's risk score with a preset risk threshold, identifying high-risk patients based on the comparison results, and marking the high-risk patients as high-risk;

[0026] Obtain the patient's historical risk identification results and analyze the long-term trends of the patient's renal function indicators. Long-term trend identification based on renal function indicators may lead to early signals of AKI.

[0027] Furthermore, the renal function change rules include: linear change rules: renal function indicators show a straight upward or downward trend, representing the development of chronic kidney disease or recovery after treatment; non-linear change rules: renal function indicators change irregularly, representing acute kidney injury or repeated inflammatory response; stage change rules: renal function indicators are stable in certain time periods, but change significantly in other time periods, which may be related to treatment or lifestyle changes; threshold trigger rules: when the renal function indicators exceed a specific threshold, it indicates that renal function may have changed significantly.

[0028] Furthermore, the early warning decision-making unit includes:

[0029] The early warning information generation module is used to receive the patient risk score and renal function change rules output by the risk analysis module, determine the early warning level according to the risk score, and match the corresponding AKI early warning information;

[0030] The warning threshold adjustment module is used to extract representative risk features from each patient group, analyze the risk distribution within each group, and adjust the warning threshold for each patient group based on the group characteristics and risk distribution;

[0031] The warning channel matching module is used to collect the AKI warning information receiving preferences of medical staff and patients, customize the content and format of AKI warning information according to the preferences, and match the corresponding notification channel according to the warning level.

[0032] Furthermore, the warning information generation module is also used to:

[0033] Analyze the key factors that may cause AKI, determine the specific cause of AKI, and match the corresponding laboratory tests and imaging examination items according to the specific cause;

[0034] Provide corresponding treatment recommendations based on laboratory tests, imaging tests, the patient's medical history, and medication status, and generate AKI warning information;

[0035] Medical resources are allocated based on the patient's AKI warning information, and the feedback data obtained through the user interaction unit is used to analyze treatment recommendations, identify potential problems in the medical process, and adjust treatment recommendations.

[0036] Further, the patient's renal function characteristics are determined based on the time series data of each data, and the changes of the patient's renal function characteristics in the electronic health record are obtained based on the time series data to determine the patient's renal function change rules, including:

[0037] Determine multimodal development-operation data of the patient's kidney in multiple time periods based on the time series data of each data;

[0038] Determine the patient's working label in each time period based on the multimodal development-working data based on a preset multimodal computing framework, and determine the expression level of the patient's renal function based on the working label;

[0039] Determine the patient's renal function characteristics according to the patient's renal function expression level, and obtain the changes of the patient's renal function characteristics in the electronic health record based on the time series data;

[0040] The urination frequency coefficient was defined as the design variable, the kidney health status was taken as the objective function, and the exercise index was taken as the constraint condition to establish a renal function gradient assessment model.

[0041] Determine the time series data difference according to the change of the renal function characteristics in the electronic health record, and determine the patient's urination related parameters and physical status related parameters according to the time series data;

[0042] Substituting urination-related parameters and physical status-related parameters into the renal function gradient assessment model to determine the patient's renal function gradient change parameters;

[0043] Determine the patient's renal function gradient qualitative boundary range according to the gradient change parameters, and determine the patient's renal function level decline trend and renal function level level trend according to the gradient qualitative boundary range and the parameters within the range;

[0044] According to the downward trend of renal function level and the leveling trend of renal function level, the quantitative change interval of the patient's renal function is determined by the preset logical quantitative rules;

[0045] The threshold decreasing trend is determined according to the quantitative change threshold in the quantitative change interval, and the patient's renal function change rule is determined according to the threshold decreasing trend.

[0046] Furthermore, after constructing the patient's electronic health record, it also includes:

[0047] Collect multiple static grayscale kidney images of the patient and select the target kidney image with the best quality, and import the target kidney image into the preset 3D graphics software to extract the kidney morphological features;

[0048] Determine a selection operator according to the kidney morphological characteristics, and determine the health non-zero coefficients of multiple characteristic points of the kidney morphological characteristics through the selection operator;

[0049] The degree of renal function damage of each patient is calculated based on the non-zero health coefficients of multiple feature points:

[0050]

[0051] Among them, A i It is expressed as the degree of renal function damage of the i-th patient, D i is the mental state index of the i-th patient, Ni is the number of collected feature points of the i-th patient, j is the j-th feature point, P j It is represented as the healthy non-zero coefficient of the j-th feature point, β is represented as the test factor, e is represented as a natural constant with a value of 2.72, and F i It is represented as the historical renal performance index of the i-th patient;

[0052] Determine a renal function label for each patient according to the degree of renal function damage of the patient, and associate the renal function label with the electronic health record of the patient, wherein the renal function label includes: no renal damage, slight renal damage, general renal damage, and severe renal damage;

[0053] The risk of AKI in patients was predicted and assessed based on each patient's renal function signature.

[0054] Through the present invention, an acute kidney injury risk warning system is disclosed. Through data preprocessing, model construction and risk analysis, real-time monitoring and accurate prediction of patients' AKI risks are achieved, and the early recognition rate is improved. The warning decision unit generates specific warning information according to the risk score and renal function change rules, and provides targeted treatment suggestions. At the same time, through the warning threshold adjustment and pathway matching modules, the warning strategy is optimized, the response speed of medical staff and the compliance of patients are improved, and the user interaction unit continuously optimizes the warning system through the feedback mechanism.

[0055] The beneficial effects of the technical solution of the present invention are:

[0056] The data acquisition unit integrates multi-source heterogeneous data to build a complete patient electronic health record, ensuring the accuracy and traceability of the information. The data analysis unit realizes real-time monitoring and accurate prediction of the patient's AKI risk through data preprocessing, model construction and risk analysis, and improves the early identification rate. Furthermore, through time series data analysis, it realizes accurate identification of the patient's renal function characteristics and change rules, providing an important data basis for AKI risk assessment and early warning, which helps to improve the early identification rate and intervention efficiency of AKI. The early warning decision unit generates specific early warning information according to the risk score and renal function change rules, and provides targeted treatment recommendations. At the same time, through the early warning threshold adjustment and pathway matching modules, the early warning strategy is optimized, false alarms and missed alarms are reduced, and the response speed of medical staff and patient compliance are improved. The user interaction unit realizes real-time display of warning information and historical record analysis, and continuously optimizes the early warning system through the feedback mechanism.

[0057] By calculating the degree of renal function damage of each patient based on their daily mental state and renal performance, the renal health of each patient can be determined quickly, objectively and accurately, and patients with unhealthy kidneys can be quickly screened out for priority AKI assessment, thereby improving work efficiency and the accuracy of determining renal function damage.

[0058] With the aging of the population and the increase in the number of patients with chronic diseases, the incidence of AKI is on the rise. Early identification and intervention are of great significance for reducing mortality and improving patient prognosis. The country attaches great importance to the construction of medical informatization and has issued a series of policies to support the development of medical informatization, providing a good policy environment for the promotion of the present invention. The present invention can achieve real-time monitoring and accurate prediction, which can effectively improve the early identification rate and treatment success rate of AKI, solve the problems of inaccurate warning timing, low real-time performance, lack of personalized warning, etc. in the prior art, meet clinical needs, and have broad market prospects.

[0059] The present invention can identify and intervene in AKI at an early stage, reduce the use of intensive care unit (ICU) beds, the consumption of medical resources such as dialysis, automate early warning and personalized treatment recommendations, effectively reduce the treatment costs and hospitalization time of AKI patients, reduce medical costs, improve the work efficiency of medical staff, reduce complications and mortality, and thus generate significant economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a diagram of the acute kidney injury risk warning system in the present invention. DETAILED DESCRIPTION

[0061] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0062] In order to solve the technical problems in the existing technology, which require a lot of data preprocessing and model training, a long time of data processing and model operation, resulting in inaccurate warning timing and low real-time performance, and the inability to provide personalized warning prompts for patient groups, which easily leads to warning fatigue and warning desensitization, please refer to Figure 1 , this embodiment provides the following technical solutions:

[0063] An acute kidney injury risk early warning system, comprising:

[0064] The data collection unit is used to collect the patient's basic information (such as name, ID number, etc.), medical history, medication status, laboratory test results (such as blood creatinine, urine volume, etc.) and biomarker data (such as NGAL, KIM-1, etc.), assign a unique identifier to the patient based on the patient's basic information, and build the patient's electronic health record, and record the patient's medical history, medication status, laboratory test results and biomarker data in the patient's electronic health record;

[0065] The data analysis unit is used to pre-process the acquired electronic health records of patients, build an AKI risk assessment model, extract and combine laboratory test results and biomarker data in the electronic health records of patients, predict the risk of AKI in patients based on the AKI risk assessment model, monitor the dynamic changes of patients' renal function in real time, automatically identify the trend of renal function decline, and trigger the early warning mechanism;

[0066] The early warning decision unit is used to generate corresponding AKI early warning information based on the risk prediction results, including the warning level, possible causes, recommended examination items and preliminary treatment suggestions, and adjust the warning threshold based on the characteristics of the patient group. The corresponding AKI early warning information and notification channels are matched based on the preferences of the warning audience, so as to optimize the early warning strategy and reduce false alarms and missed alarms.

[0067] The user interaction unit is used to display the AKI warning information and historical warning records output by the warning decision unit in real time, including past warning events, treatment measures and patient outcomes, and to establish a feedback mechanism to optimize and adjust the AKI warning information based on the feedback mechanism. According to the feedback mechanism, the effects of different intervention measures on reducing the risk of AKI are evaluated, and the risk scores before and after the intervention are compared to measure the treatment effect, monitor the changes in the incidence of AKI, and evaluate the effectiveness of prevention strategies.

[0068] In this embodiment, by constructing the patient's electronic health record, not only a rich data basis is provided for the assessment of AKI risk, but also the integrity and traceability of patient information are ensured, which helps doctors make more accurate decisions during diagnosis and treatment, build an efficient and accurate AKI risk assessment model, accurately predict the risk of AKI in patients, and monitor the dynamic changes of renal function in real time to improve the early recognition rate, which can significantly improve the efficiency of early identification and intervention of AKI, reduce the complications and mortality caused by AKI in patients, and improve clinical prognosis. In addition, according to the characteristics of the patient group and the preferences of the warning audience, the warning threshold and notification channel can be adjusted to reduce false alarms and missed alarms. The personalized warning mechanism can ensure that medical staff obtain key information in the shortest time, so as to take timely and effective intervention measures, which will help promote the development and application of clinical decision support systems and improve the overall quality and level of medical services.

[0069] In this embodiment, the data acquisition unit includes:

[0070] The multi-source data integration module is used to establish a multi-source heterogeneous data interface with the hospital's various electronic medical record systems, laboratory information systems, and medical imaging systems. Based on the multi-source heterogeneous data interface, the module extracts the patient's key information, including basic information, medical history, medication status, laboratory test results, and biomarker data, and standardizes the data format of key information in different data sources.

[0071] The patient data packet generation module is used to classify and package the acquired key information based on the data source to generate patient data packets, each of which contains relevant information of the patient in different systems; each patient data packet is assigned a unique data packet identifier corresponding to the patient's unique identifier to ensure the traceability of the data packet;

[0072] The electronic health record management module is used to match the patient's basic data, medical history, medication status, laboratory test results and biomarker data based on the data source; based on the above relationship, when a unique identifier is entered, the corresponding patient data package can be retrieved at the same time to build the patient's electronic health record.

[0073] In this embodiment, by uniformly accessing and processing various systems in the hospital, key information of patients can be extracted from multi-source heterogeneous data more effectively, and key information such as disease name, symptom description, treatment plan, etc. can be extracted from unstructured or semi-structured text in EMR. For structured data in LIS and PACS, required data items such as blood creatinine value, urine volume data, imaging diagnosis results, etc. can be directly extracted, and associations between data can be established. The medical history in EMR and the laboratory test results in LIS are associated with the patient ID to build a complete patient electronic health record, providing a solid data foundation for subsequent AKI risk assessment and early warning.

[0074] In this embodiment, the data analysis unit includes:

[0075] The data preprocessing module is used to clean, normalize and standardize the acquired patient electronic health record data, and extract key features related to AKI risk assessment from the processed patient electronic health record data, including time series features, renal function features, etc.;

[0076] A model building module is used to input the extracted key features related to AKI risk assessment into a preset network model to obtain an AKI risk assessment model for each patient;

[0077] The risk analysis module is used to input the newly acquired patient electronic health record data into the AKI risk assessment model for risk identification and output the patient's risk prediction results, including risk level, risk probability and other forms of assessment results. It is also used to:

[0078] Combine the patient's risk prediction results with the corresponding patient population characteristics to assess the patient's likelihood of developing AKI, and quantify the results to obtain the patient's risk score;

[0079] Comparing the patient's risk score with a preset risk threshold, identifying high-risk patients based on the comparison results, and marking the high-risk patients as high-risk;

[0080] Obtain the patient's historical risk identification results and analyze the long-term trend of the patient's renal function indicators. Based on the long-term trend of renal function indicators, early signs of AKI may be identified, such as a gradual decline in renal function.

[0081] In this embodiment, by combining the risk prediction results and patient population characteristics, the possibility of AKI in patients is evaluated and quantified into a risk score. Medical staff take corresponding preventive measures according to the score, which helps medical staff quickly identify potential high-risk patients, ensuring that high-risk patients can receive priority attention and timely intervention, thereby possibly reducing the occurrence of AKI, improving the early detection rate of AKI, and monitoring the long-term trend of renal function indicators, so as to better understand the disease progression of patients, provide a basis for personalized treatment, and improve the quality and efficiency of medical services;

[0082] In this embodiment, the data analysis unit further includes:

[0083] Extracting the patient's laboratory test results and biomarker data from the patient's electronic health record, and extracting time series data of each data carried by the laboratory test results and biomarker data;

[0084] Determine the patient's renal function characteristics based on the time series data of each data, including stable renal function, gradual decline in renal function, acute renal function injury, renal function fluctuation, etc., and obtain the changes of the patient's renal function characteristics in the electronic health record based on the time series data to determine the patient's renal function change rules;

[0085] In this embodiment, the renal function change rules include: linear change rule: the renal function index shows a straight upward or downward trend, which represents the development of chronic kidney disease or recovery after treatment; nonlinear change rule: the renal function index changes irregularly, which represents acute kidney injury or repeated inflammatory response; stage change rule: the renal function index is stable in certain time periods, but changes significantly in other time periods, which may be related to treatment or lifestyle changes; threshold trigger rule: when the renal function index exceeds a specific threshold, it indicates that the renal function may have changed significantly;

[0086] Compare the renal function change rules of each patient, and integrate the patients whose similarity is greater than or equal to the preset threshold based on the comparison results to generate a group of similar patients;

[0087] The renal function change rules of any patient in each patient group are taken as the representative risk characteristics of the patient group and confirmed as the patient group characteristics.

[0088] In this embodiment, by real-time monitoring of the patient's renal function indicators, vital signs and other key data, an AKI risk prediction model is constructed to conduct a comprehensive analysis of the patient's various indicators, thereby achieving intelligent and precise prediction of AKI risk, which not only improves the early identification ability of medical institutions for AKI, but also provides strong support for clinical decision-making, thereby significantly reducing the incidence and mortality of AKI. By analyzing the rules of changes in patient renal function and classifying patient groups, an in-depth understanding of the patient group and the change pattern of patient renal function is achieved, which provides a basis for personalized treatment, optimizes treatment strategies, improves the pertinence and efficiency of treatment, and promotes the development of personalized medicine.

[0089] In this embodiment, the early warning decision unit includes:

[0090] The early warning information generation module is used to receive the patient risk score and renal function change rules output by the risk analysis module, determine the early warning level according to the risk score, match the corresponding AKI early warning information, and is also used to:

[0091] Analyze the key factors that may cause AKI, such as drug use, surgery, infection, dehydration, etc., determine the specific cause of AKI, so as to take targeted treatment measures, and match the corresponding laboratory tests and imaging examination items according to the specific cause;

[0092] Provide corresponding treatment recommendations based on laboratory tests and imaging examination items, the patient's medical history and medication status, such as adjusting medication, fluid replacement, dialysis, etc., and generate AKI warning information;

[0093] Allocate medical resources, such as intensive care unit (ICU) beds and dialysis resources, based on the patient's AKI warning information. Analyze treatment recommendations through feedback data obtained from the user interaction unit, identify potential problems in the medical process, and adjust treatment recommendations to promote quality improvement measures.

[0094] An early warning threshold adjustment module is used to extract representative risk features from each patient group, analyze the risk distribution within each group, and adjust the early warning threshold for each patient group based on the group characteristics and risk distribution. For example, for high-risk groups, the threshold is lowered to increase sensitivity, and for low-risk groups, the threshold is increased to reduce false positives;

[0095] The warning channel matching module is used to collect the AKI warning information receiving preferences of medical staff and patients, customize the content and format of AKI warning information according to the preferences, and match the corresponding notification channel according to the warning level.

[0096] In this embodiment, specific AKI warning information is generated based on the patient's real-time data and risk score, the key factors of AKI are analyzed and targeted treatment suggestions are provided, thereby ensuring that the patient receives timely and effective intervention. By dynamically adjusting the warning threshold and personalized matching notification channels, the problems of warning fatigue and warning desensitization are effectively solved, the response speed of medical staff to warning information and the compliance of patients are improved, the early identification rate and treatment success rate of AKI are significantly improved, the health risks of patients are reduced, the allocation of medical resources is optimized, and the continuous improvement of medical quality is promoted, providing patients with safer and more efficient medical services.

[0097] In one embodiment, the patient's renal function characteristics are determined based on the time series data of each data, and the change of the patient's renal function characteristics in the electronic health record is obtained based on the time series data to determine the patient's renal function change rule, including:

[0098] Determine multimodal development-operation data of the patient's kidney in multiple time periods based on the time series data of each data;

[0099] Determine the patient's working label in each time period based on the multimodal development-working data based on a preset multimodal computing framework, and determine the expression level of the patient's renal function based on the working label;

[0100] Determine the patient's renal function characteristics according to the patient's renal function expression level, and obtain the changes of the patient's renal function characteristics in the electronic health record based on the time series data;

[0101] The urination frequency coefficient was defined as the design variable, the kidney health status was taken as the objective function, and the exercise index was taken as the constraint condition to establish a renal function gradient assessment model.

[0102] Determine the time series data difference according to the change of the renal function characteristics in the electronic health record, and determine the patient's urination related parameters and physical status related parameters according to the time series data;

[0103] Substituting urination-related parameters and physical status-related parameters into the renal function gradient assessment model to determine the patient's renal function gradient change parameters;

[0104] Determine the patient's renal function gradient qualitative boundary range according to the gradient change parameters, and determine the patient's renal function level decline trend and renal function level level trend according to the gradient qualitative boundary range and the parameters within the range;

[0105] According to the downward trend of renal function level and the leveling trend of renal function level, the quantitative change interval of the patient's renal function is determined by the preset logical quantitative rules;

[0106] The threshold decreasing trend is determined according to the quantitative change threshold in the quantitative change interval, and the patient's renal function change rule is determined according to the threshold decreasing trend.

[0107] The beneficial effects of the above technical solution are: by determining the decreasing trend of the patient's renal function level and the leveling trend of the renal function level, the patient's renal working level can be intuitively determined, and then the quantitative threshold change interval of the patient's renal function can be quickly determined according to the quantitative rules, and the renal function change rules can be determined according to the data change trend, thereby improving the judgment efficiency and accuracy.

[0108] In one embodiment, after constructing the patient's electronic health record, the method further includes:

[0109] Collect multiple static grayscale kidney images of the patient and select the target kidney image with the best quality, and import the target kidney image into the preset 3D graphics software to extract the kidney morphological features;

[0110] Determine a selection operator according to the kidney morphological characteristics, and determine the health non-zero coefficients of multiple characteristic points of the kidney morphological characteristics through the selection operator;

[0111] The degree of renal function damage of each patient is calculated based on the non-zero health coefficients of multiple feature points:

[0112]

[0113] Among them, A i It is expressed as the degree of renal function damage of the i-th patient, D i is the mental state index of the i-th patient, Ni is the number of collected feature points of the i-th patient, j is the j-th feature point, P j It is represented as the healthy non-zero coefficient of the j-th feature point, β is represented as the test factor, e is represented as a natural constant with a value of 2.72, and F i It is represented as the historical renal performance index of the i-th patient;

[0114] Determine a renal function label for each patient according to the degree of renal function damage of the patient, and associate the renal function label with the electronic health record of the patient, wherein the renal function label includes: no renal damage, slight renal damage, general renal damage, and severe renal damage;

[0115] The risk of AKI in patients was predicted and assessed based on each patient's renal function signature.

[0116] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. An acute kidney injury risk warning system, characterized in that: include: The data collection unit is used to collect the patient's basic information, medical history, medication status, laboratory test results, and biomarker data, assign a unique identifier to the patient based on the patient's basic information, and build the patient's electronic health record; The data analysis unit is used to pre-process the acquired electronic health records of patients, build an AKI risk assessment model, extract and combine laboratory test results and biomarker data in the electronic health records of patients, and predict the risk of AKI in patients based on the AKI risk assessment model; The early warning decision unit is used to generate corresponding AKI early warning information based on the risk prediction results, adjust the early warning threshold based on the characteristics of the patient group, and match the corresponding AKI early warning information and notification channels based on the preferences of the early warning audience; The user interaction unit is used to display the AKI warning information and historical warning records output by the warning decision unit in real time, and to establish a feedback mechanism to optimize and adjust the AKI warning information based on the feedback mechanism.

2. The acute kidney injury risk warning system according to claim 1, characterized in that: The data acquisition unit comprises: The multi-source data integration module is used to establish a multi-source heterogeneous data interface with the hospital's various electronic medical record systems, laboratory information systems, and medical imaging systems, extract key patient information based on the multi-source heterogeneous data interface, and standardize the data format of key information in different data sources; A patient data packet generation module is used to classify and package the acquired key information based on the data source, generate a patient data packet, and assign a unique data packet identifier to each patient data packet, corresponding to the patient's unique identifier; The electronic health record management module is used to match the patient's basic data, medical history, medication status, laboratory test results and biomarker data based on the data source; based on the above relationship, when a unique identifier is entered, the corresponding patient data package can be retrieved at the same time to build the patient's electronic health record.

3. The acute kidney injury risk warning system according to claim 1, characterized in that: The data analysis unit comprises: The data preprocessing module is used to clean, normalize and standardize the acquired patient electronic health record data, and extract key features related to AKI risk assessment from the processed patient electronic health record data; A model building module is used to input the extracted key features related to AKI risk assessment into a preset network model to obtain an AKI risk assessment model for each patient; The risk analysis module is used to input the newly acquired patient electronic health record data into the AKI risk assessment model for risk identification and output the patient's risk prediction results.

4. The acute kidney injury risk warning system according to claim 1, characterized in that: The data analysis unit also includes: Extracting the patient's laboratory test results and biomarker data from the patient's electronic health record, and extracting time series data of each data carried by the laboratory test results and biomarker data; Determine the patient's renal function characteristics based on the time series data of each data, obtain the change of the patient's renal function characteristics in the electronic health record based on the time series data, and determine the patient's renal function change rules; Compare the renal function change rules of each patient, and integrate the patients whose similarity is greater than or equal to the preset threshold based on the comparison results to generate a group of similar patients; The renal function change rules of any patient in each patient group are taken as the representative risk characteristics of the patient group and confirmed as the patient group characteristics.

5. The acute kidney injury risk warning system according to claim 1, characterized in that: The risk analysis module is also used to: Combine the patient's risk prediction results with the corresponding patient population characteristics to assess the patient's likelihood of developing AKI, and quantify the results to obtain the patient's risk score; Comparing the patient's risk score with a preset risk threshold, identifying high-risk patients based on the comparison results, and marking the high-risk patients as high-risk; Obtain the patient's historical risk identification results and analyze the long-term trends of the patient's renal function indicators. Long-term trend identification based on renal function indicators may lead to early signals of AKI.

6. The acute kidney injury risk warning system according to claim 1, characterized in that: The rules for changes in renal function include: Linear change rule: renal function indicators show a straight upward or downward trend, which represents the development of chronic kidney disease or recovery after treatment; Nonlinear change rules: Irregular changes in renal function indicators represent acute kidney injury or repeated inflammatory reactions; Periodic rule: Kidney function indicators are stable in some time periods and vary significantly in other time periods, which may be related to treatment or lifestyle changes; Threshold trigger rule: When the renal function index exceeds a specific threshold, it indicates that significant changes in renal function may have occurred.

7. The acute kidney injury risk warning system according to claim 1, characterized in that: The early warning decision unit comprises: The early warning information generation module is used to receive the patient risk score and renal function change rules output by the risk analysis module, determine the early warning level according to the risk score, and match the corresponding AKI early warning information; The warning threshold adjustment module is used to extract representative risk features from each patient group, analyze the risk distribution within each group, and adjust the warning threshold for each patient group based on the group characteristics and risk distribution; The warning channel matching module is used to collect the AKI warning information receiving preferences of medical staff and patients, customize the content and format of AKI warning information according to the preferences, and match the corresponding notification channel according to the warning level.

8. The acute kidney injury risk warning system according to claim 1, characterized in that: The warning information generation module is also used for: Analyze the key factors that may cause AKI, determine the specific cause of AKI, and match the corresponding laboratory tests and imaging examination items according to the specific cause; Provide corresponding treatment recommendations based on laboratory tests, imaging tests, the patient's medical history, and medication status, and generate AKI warning information; Medical resources are allocated based on the patient's AKI warning information, and the feedback data obtained through the user interaction unit is used to analyze treatment recommendations, identify potential problems in the medical process, and adjust treatment recommendations.

9. The acute kidney injury risk warning system according to claim 1, characterized in that: Determine the patient's renal function characteristics based on the time series data of each data, obtain the changes of the patient's renal function characteristics in the electronic health record based on the time series data, and determine the patient's renal function change rules, including: Determine multimodal development-operation data of the patient's kidney in multiple time periods based on the time series data of each data; Determine the patient's working label in each time period based on the multimodal development-working data based on a preset multimodal computing framework, and determine the expression level of the patient's renal function based on the working label; Determine the patient's renal function characteristics according to the patient's renal function expression level, and obtain the changes of the patient's renal function characteristics in the electronic health record based on the time series data; The urination frequency coefficient was defined as the design variable, the kidney health status was taken as the objective function, and the exercise index was taken as the constraint condition to establish a renal function gradient assessment model. Determine the time series data difference according to the change of the renal function characteristics in the electronic health record, and determine the patient's urination related parameters and physical status related parameters according to the time series data; Substituting urination-related parameters and physical status-related parameters into the renal function gradient assessment model to determine the patient's renal function gradient change parameters; Determine the patient's renal function gradient qualitative boundary range according to the gradient change parameters, and determine the patient's renal function level decline trend and renal function level level trend according to the gradient qualitative boundary range and the parameters within the range; According to the downward trend of renal function level and the leveling trend of renal function level, the quantitative change interval of the patient's renal function is determined by the preset logical quantitative rules; The threshold decreasing trend is determined according to the quantitative change threshold in the quantitative change interval, and the patient's renal function change rule is determined according to the threshold decreasing trend.

10. The acute kidney injury risk warning system according to claim 1, characterized in that: After building the patient's electronic health record, it also includes: Collect multiple static grayscale kidney images of the patient and select the target kidney image with the best quality, and import the target kidney image into the preset 3D graphics software to extract the kidney morphological features; Determine a selection operator according to the kidney morphological characteristics, and determine the health non-zero coefficients of multiple characteristic points of the kidney morphological characteristics through the selection operator; The degree of renal function damage of each patient is calculated based on the non-zero health coefficients of multiple feature points: Among them, A i It is expressed as the degree of renal function damage of the i-th patient, D i is the mental state index of the i-th patient, Ni is the number of collected feature points of the i-th patient, j is the j-th feature point, P j It is represented as the healthy non-zero coefficient of the j-th feature point, β is represented as the test factor, e is represented as a natural constant with a value of 2.72, and F i It is represented as the historical renal performance index of the i-th patient; Determine a renal function label for each patient according to the degree of renal function damage of the patient, and associate the renal function label with the electronic health record of the patient, wherein the renal function label includes: no renal damage, slight renal damage, general renal damage, and severe renal damage; The risk of AKI in patients was predicted and assessed based on each patient's renal function signature.

Images