Construction method of ICU nursing decision support system based on case reasoning

Abstract

The invention relates to the technical field of decision support, in particular to a method for constructing an ICU nursing decision support system based on case reasoning. The method comprises the following steps: carrying out structured time sequence sign change extraction according to ICU nursing historical case data to obtain structured time sequence sign change data; tolerance impairment interpolation processing is carried out on the structured time sequence sign change data, nursing intervention efficiency analysis is carried out, then physiological sign index multi-dimensional dynamic evaluation is carried out, and physiological sign index multi-dimensional dynamic data is obtained; carrying out complication occurrence probability deduction analysis on the physiological sign index multi-dimensional dynamic data, and carrying out abnormal decision deduction to obtain symptom nursing abnormal decision deduction data; and performing ICU nursing decision support scheme construction on the symptom nursing abnormality decision deduction data based on a random forest algorithm to obtain an ICU nursing decision support scheme. According to the method, the decision support technology is optimized, so that the decision support technology is more perfect.

Description

Technical Field

[0001] The present invention relates to the technical field of decision support, and particularly to a construction method of an ICU nursing decision support system based on case-based reasoning. Background Art

[0002] In the ICU environment, patients usually have complex conditions and rapid state changes, and nursing staff face the challenge of making quick and accurate decisions. Traditional decision-making methods often rely on standardized processes and experience accumulation, but they are insufficient when facing changing conditions. The nursing decision system based on case-based reasoning collects and analyzes a large amount of historical case data and the diagnosis and treatment processes and nursing experiences of previous similar patients to establish a case base. When the system receives new conditions, it can retrieve similar cases by comparing with historical cases and provide suggestions to assist nursing staff in quickly judging the best nursing plan. This system not only helps to improve the accuracy and timeliness of nursing decisions, but also reduces the cognitive burden of nursing staff. The CBR-based method has the characteristics of self-learning and dynamic update, and can continuously optimize the case base according to new nursing data and case information to improve the reliability and adaptability of the system. However, there is a problem in the traditional construction method of an ICU nursing decision support system based on case-based reasoning that the analysis of patients' physical signs is inaccurate, resulting in large nursing decision errors. Summary of the Invention

[0003] Based on this, it is necessary to provide a construction method of an ICU nursing decision support system based on case-based reasoning to solve at least one of the above technical problems.

[0004] To achieve the above object, a construction method of an ICU nursing decision support system based on case-based reasoning, the method includes the following steps: Step S1: Obtain ICU nursing historical case data; extract structured time-series physical sign changes according to the ICU nursing historical case data to obtain structured time-series physical sign change data; Step S2: Perform tolerance reduction interpolation processing on the structured time-series physical sign change data to obtain symptom treatment tolerance interpolation data; perform nursing intervention efficacy analysis on the symptom treatment tolerance interpolation data to obtain symptom nursing intervention efficacy data; perform multi-dimensional dynamic assessment of physiological sign indicators according to the symptom nursing intervention efficacy data to obtain multi-dimensional dynamic data of physiological sign indicators; Step S3: Perform complication occurrence probability reduction analysis on the multi-dimensional dynamic data of physiological sign indicators to obtain complication occurrence probability reduction data; perform abnormal decision deduction on the complication occurrence probability reduction data to obtain symptom nursing abnormal decision deduction data; Step S4: Based on the random forest algorithm, construct an ICU nursing decision support plan for the symptom nursing abnormal decision deduction data to obtain an ICU nursing decision support plan.

[0005] By obtaining ICU nursing historical case data, the present invention can systematically collect the clinical information of patients in intensive care. These data usually include patients' physiological indicators, medical history, treatment plans and their responses, etc. By analyzing these data, the extraction of structured time-series sign changes can transform these complex time-series data into an easy-to-process structured format. This process not only helps to identify the changing trends of patients' states, but also provides a basis for subsequent data analysis and processing, enabling nursing staff to more intuitively understand the changes in patients' conditions and make timely nursing decisions accordingly. For the structured time-series sign change data, tolerance reduction interpolation processing is performed to fill in the blanks caused by data missing or interruption, ensuring the continuity and integrity of the data. Through this interpolation method, the tolerance of patients to symptom treatment can be more accurately evaluated. In addition, by analyzing the nursing intervention efficacy of the interpolated data, the actual effects of different nursing measures in improving patients' symptoms can be revealed. This process not only helps nursing staff understand the effectiveness of the nursing measures adopted, but also provides data support for further optimizing the nursing plan, thereby improving the overall nursing quality and safety of patients. By analyzing the symptom nursing intervention efficacy data, a multi-dimensional dynamic assessment of physiological sign indicators is carried out. This step focuses on the comprehensive analysis of patients' physiological indicators, such as the combined changes of multiple indicators such as heart rate, blood pressure, and respiratory rate. This assessment can be carried out by constructing a multi-dimensional model and using data analysis tools to deeply analyze the correlation between various physiological indicators and their dynamic changing trends. This not only helps to identify the short-term and long-term effects of nursing intervention measures, but also provides a scientific basis for formulating personalized nursing plans clinically. Finally, this series of analyses and assessments will greatly improve the accuracy and personalization of intensive care, thereby improving the treatment effect and quality of life of patients. First, a complication occurrence probability reduction analysis is carried out on the multi-dimensional dynamic data of physiological sign indicators. The key to this process lies in analyzing the relationship between the changes of different physiological indicators of patients in the ICU environment and the occurrence of complications through statistical and machine learning methods, identifying potential risk factors, and calculating the probability of complication occurrence in different situations. This kind of analysis can help the clinical team identify high-risk patients in advance, so as to take preventive measures and reduce the incidence of complications. Next, an abnormal decision deduction is carried out on the complication occurrence probability reduction data, deeply mining the results based on probability analysis, and identifying the key factors leading to nursing abnormalities. This deduction can not only reveal the potential risk points in the nursing process, but also provide empirical evidence for subsequent clinical decisions, ensuring that patients receive timely and effective interventions, and ultimately improving the safety and effect of nursing. The random forest algorithm is used to analyze the symptom nursing abnormal decision deduction data to construct an ICU nursing decision support scheme. The random forest algorithm is a powerful ensemble learning method that can handle high-dimensional data and effectively identify the complex relationship between features and results.In this step, the algorithm comprehensively considers various variables in historical cases and identifies key factors related to patient symptoms, complication risks, and nursing measures. Through this analysis, scientific decision-making support can be provided for ICU nursing staff, including recommended nursing measures, strategies for preventing complications, and personalized nursing plans. This can not only improve the accuracy and effectiveness of nursing, but also help clinical staff make wise decisions quickly in complex situations, thereby improving the overall nursing experience and prognosis of patients. This data-driven decision-making support solution will significantly improve the nursing quality and efficiency in the ICU. Therefore, the present invention makes an improvement to the construction method of a traditional case-based reasoning ICU nursing decision support system, solves the problem that the construction method of a traditional case-based reasoning ICU nursing decision support system has inaccurate analysis of patient signs, resulting in large nursing decision errors, improves the accuracy of patient sign analysis, and reduces nursing decision errors.

[0006] Preferably, step S1 includes the following steps: Step S11: Obtain ICU nursing historical case data; Step S12: Classify the ICU nursing historical case data by subject to obtain the subject data of ICU nursing cases; Step S13: Fill in the missing values of the nursing content in the subject data of ICU nursing cases to obtain the filled subject nursing content data; Step S14: Extract the structured temporal sign changes based on the filled subject nursing content data to obtain the structured temporal sign change data.

[0007] The acquisition of ICU nursing historical case data in the present invention is the foundation of the entire process. This step involves extracting detailed nursing information of patients in the intensive care environment from various medical record systems and databases, including physiological indicators, treatment plans, complications, patient responses, etc. By systematically collecting these historical case data, researchers and clinicians can establish a comprehensive perspective on patient care, laying a solid foundation for subsequent data analysis and processing. This dataset not only provides a basis for nursing quality assessment but also reveals potential patterns and trends in ICU nursing, thereby providing scientific guidance for improving patient care. Classifying the acquired ICU nursing historical case data by subject can subdivide the data into different medical subjects, such as respiration, heart, nerves, etc. Through such classification, specialized data analysis can be provided for nursing practices in different fields, thereby enabling more targeted nursing intervention measures. The advantage of this classification is that it can help nursing staff and researchers quickly locate and compare the nursing quality and effects among different subjects, identify common problems and advantages in specific subjects, and thus improve the overall nursing level of each subject and the treatment effect of patients. By filling in the missing values of the nursing content data of the subjects to which the ICU nursing cases belong, the integrity and effectiveness of the data can be ensured. This process usually adopts various techniques, such as mean imputation, regression filling, or interpolation method, to fill in the missing data in the nursing content caused by various reasons. Filling in the missing values is crucial for the accuracy of the analysis because missing data can lead to biased analysis results or misleading conclusions. By filling in the missing values, nursing staff can obtain more complete nursing data, and thus make the subsequent data analysis and decision support processes more reliable and scientific, thereby providing better nursing services for patients. Based on the filled nursing content data of the subjects, structured time-series sign changes are extracted. This process extracts key physiological indicators and clinical features from time-series data and transforms them into a structured format for further analysis. This structured data can clearly show the trend of the patient's physiological state changing over time, providing an important basis for clinical decision-making. Through this extraction process, nursing staff can timely identify the patient's condition changes and responses, thereby achieving early warning and intervention, and ultimately improving the overall efficiency and safety of ICU nursing.

[0008] Preferably, step S2 includes the following steps: Step S21: Analyze the symptom treatment trend of the structured time-series sign change data to obtain symptom treatment trend data; Step S22: Perform tolerance reduction interpolation processing on the symptom treatment trend data to obtain symptom treatment tolerance interpolation data; Step S23: Analyze the nursing intervention efficacy of the symptom treatment tolerance interpolation data based on case reasoning and ICU nursing historical case data to obtain symptom nursing intervention efficacy data; Step S24: Perform a multi-dimensional dynamic assessment of the physiological sign indicators on the structured time-series sign change data based on the symptom care intervention efficacy data to obtain the multi-dimensional dynamic data of the physiological sign indicators.

[0009] Through the analysis of the symptom treatment trend of the structured time-series sign change data, the present invention can systematically evaluate the responses of patients to different treatment plans and their changing trends. This analysis will use statistical methods and data visualization techniques to identify the improvement or deterioration trends of symptoms, and then provide important decision-making basis for clinical nursing staff. Such trend analysis not only helps to understand the effectiveness of specific treatment measures, but also can reveal potential treatment patterns and effects, laying a foundation for formulating personalized nursing plans. By timely discovering the trend of symptom changes, nursing staff can quickly adjust the treatment plan, thereby improving the accuracy and effect of patient care and enhancing the pertinence of clinical management. The purpose of performing tolerance reduction interpolation processing on the symptom treatment trend data is to fill the gaps caused by data missing or discontinuity to ensure the continuity and integrity of the data. This process involves various interpolation techniques, such as linear interpolation or polynomial interpolation, aiming to predict the symptom treatment tolerance of patients within a specific time period. Through the tolerance interpolation processing, nursing staff can more accurately evaluate the changes in patients' responses to symptoms during the treatment process, and then optimize nursing strategies and intervention measures. Ensuring the accuracy of the tolerance data will make the subsequent nursing efficacy analysis more reliable, provide strong data support for clinical decision-making, and ultimately improve the safety and treatment experience of patients. Based on case-based reasoning and ICU nursing historical case data, perform a nursing intervention efficacy analysis on the symptom treatment tolerance interpolation data. This analysis aims to evaluate the actual effects of different nursing measures in improving patients' symptoms. By comparing the interpolated tolerance data with the effective nursing measures in historical cases, nursing staff can identify best practices and feasible intervention measures. The results of this process will provide scientific basis for nursing staff to optimize the existing nursing process, ensure that patients receive the most suitable nursing intervention, thereby reducing the risk of complications and improving the overall nursing quality. According to the symptom care intervention efficacy data, perform a multi-dimensional dynamic assessment of the physiological sign indicators on the structured time-series sign change data. Through this assessment, it is possible to comprehensively analyze the changes in patients' physiological indicators during the nursing process, covering multiple key physiological parameters such as heart rate, blood pressure, and respiratory rate. This multi-dimensional assessment not only helps to identify the short-term and long-term effects of nursing interventions, but also can reveal the interrelationships and dynamic change trends among different physiological indicators, providing a more comprehensive perspective for clinical decision-making. Through this comprehensive data analysis, nursing staff can more effectively monitor patients' conditions, timely adjust the nursing plan to better meet the personalized needs of patients, and ultimately improve the overall efficiency and effect of ICU nursing.

[0010] Preferably, step S22 includes the following steps: Step S221: Analyze the clinical response improvement trend of the symptom treatment trend data to obtain the clinical response improvement trend data; Step S222: Analyze the temporal discontinuity difference of the clinical response improvement trend data to obtain the response improvement temporal discontinuity difference data; Step S223: Calculate the median of the improvement differences between adjacent time series for the response improvement temporal discontinuity difference data to obtain the median of the improvement differences between adjacent time series; Step S224: Optimize the linear structure of the clinical response improvement trend data based on the median of the improvement differences between adjacent time series to obtain the response improvement trend linear data; Step S225: Perform global variance nearest neighbor interpolation on the response improvement trend linear data to obtain the improvement trend global interpolation data; Step S226: Perform tolerance reduction interpolation processing on the symptom treatment trend data based on the improvement trend global interpolation data to obtain the symptom treatment tolerance interpolation data.

[0011] Through the analysis of the improvement trend of clinical responses on symptomatic treatment trend data, various changes in clinical responses shown by patients after receiving treatment can be identified. This analysis usually adopts statistical methods and data visualization techniques to clearly display the improvement of patients' symptoms, such as pain level, functional recovery, etc. Through in-depth analysis of the improvement trend, the nursing team can effectively evaluate the effectiveness of treatment plans and provide a basis for subsequent nursing decisions. The results of this process can not only help identify effective nursing measures but also reveal potential patient needs and treatment bottlenecks, thereby providing data support for the formulation of personalized nursing plans and ensuring that patients obtain the best nursing experience. Conducting a time-series discontinuous difference analysis on the clinical response improvement trend data aims to identify and evaluate significant changes in patients' clinical responses at different time points. This process usually uses statistical test methods, such as t-tests or analysis of variance, to detect the degree of difference between time periods. The key to this analysis lies in determining which periods have significant improvements, thus providing strong support for clinical decisions. By revealing the discontinuous differences in response improvement, nursing staff can identify the peaks and troughs of patients' symptom improvement, further optimize nursing intervention measures, improve the pertinence and effectiveness of nursing, and help patients better go through the treatment process. Calculating the median between adjacent time series for the discontinuous difference data of response improvement is to obtain typical values of symptom improvement at different time points. This step can effectively eliminate the influence of extreme values on the results by calculating the median between adjacent time series, providing more stable and reliable improvement difference data. The calculation of the median helps identify the trend of treatment effects and enables clinicians to understand patients' responses during specific treatment periods. This data will provide a key reference basis for subsequent linear structure optimization, ensuring scientific and accurate data processing and laying a good foundation for formulating more personalized nursing plans for patients. Optimizing the linear structure of clinical response improvement trend data using the median of improvement differences between adjacent time series aims to improve the accuracy and interpretability of analysis results through data reorganization. This optimization process usually involves linear regression analysis to establish a model to capture the relationship between clinical responses and time, so as to more clearly present the dynamic changes in symptom improvement. This kind of optimization can not only improve the smoothness and readability of data but also provide a clear improvement trend for nursing staff, enabling them to make more effective decisions based on data when formulating treatment plans, thereby improving the overall nursing quality and promoting patients' recovery. Conducting global variance nearest neighbor interpolation on the linear data of response improvement trend aims to fill in missing values in the data and enhance data continuity. This interpolation method relies on global data characteristics to estimate missing values, ensuring smooth transition and consistency of data. By using nearest neighbor interpolation, nursing staff can obtain a complete symptom improvement trend chart to help identify the overall pattern and changes in treatment effects.The effective execution of this process will make subsequent analyses more accurate, ensuring the reliability of the data on which clinical decisions are based, thereby providing a solid foundation for optimizing treatment plans and ultimately improving the treatment effect and satisfaction of patients. Based on the global interpolation data of the improvement trend, the tolerance reduction interpolation process is performed on the symptom treatment trend data to obtain more accurate patient symptom treatment tolerance. This process interpolates the tolerance data during the symptom treatment process to make the data more complete, facilitating subsequent analysis and application. Through this method, nursing staff can identify the tolerance of patients at different treatment stages and understand their response to treatment. This will provide strong data support for clinical practice, enabling the nursing team to adjust nursing measures in a timely manner, optimize the treatment experience of patients, reduce discomfort and side effects, and ultimately improve the overall quality of nursing and the quality of life of patients.

[0012] Preferably, the global variance nearest neighbor interpolation of the response improvement trend linear data includes the following steps: Using a preset improvement value assignment model to evaluate the adjacent trend improvement value of the response improvement trend linear data to obtain adjacent trend improvement value data; Calculating the global numerical variance of the adjacent trend improvement value data to obtain the global trend numerical variance; Performing global variance nearest neighbor interpolation on the adjacent trend improvement value data according to the global trend numerical variance to obtain the global interpolation data of the improvement trend.

[0013] The present invention can systematically capture the dynamic changes in the improvement of patients' symptoms over different time periods by evaluating the adjacent trend improvement values of the reaction improvement trend linear data using a preset improvement value. The preset improvement value serves as an evaluation criterion, enabling the model to accurately analyze clinical data while taking into account individual patient differences. Through this evaluation, a set of adjacent trend improvement value data can be obtained, which helps to understand the actual impact of different treatment methods or time nodes on patient improvement. The effective implementation of this step will lay the foundation for subsequent analysis, ensuring that various factors can be comprehensively considered when processing clinical data, thereby improving the prediction accuracy of treatment effects and providing data support for the formulation of personalized care plans. Calculating the global numerical variance of the adjacent trend improvement value data aims to evaluate the degree of variation of these values in the entire dataset. By calculating the global variance, the stability and consistency of symptom improvement can be quantified, and the significance of the improvement trend at different time points or treatment regimens can be identified. This analysis not only helps nursing staff understand whether the patient's response is persistent but also reveals potential influencing factors such as changes in treatment regimens and individual patient differences. The calculation of the global variance provides the necessary statistical basis for subsequent interpolation processing, enabling the interpolation results to more accurately reflect the patient's changing trend throughout the treatment process, thereby optimizing nursing decisions and enhancing the overall treatment experience of the patient. Conducting global variance nearest neighbor interpolation on the adjacent trend improvement value data based on the global trend numerical variance aims to improve the integrity and reliability of the data. This interpolation method can fill in missing values or discontinuities in the data by utilizing known improvement values and global variances, making the description of the improvement trend smoother and more consistent. Through global variance nearest neighbor interpolation, the nursing team can obtain more accurate global interpolation data of the improvement trend, which can not only better reflect the symptom changes of the patient during the treatment process but also provide strong data support for future clinical decisions. Ultimately, the implementation of this process will greatly enhance the ability to evaluate the treatment effects of symptoms, ensure continuous monitoring of patients during treatment, and timely adjustment of care plans, thereby improving patient safety and treatment satisfaction.

[0014] Preferably, step S23 includes the following steps: Step S231: Based on case-based reasoning, simulate the intervention effect matching during the nursing process for the interpolation data of symptom treatment tolerance to obtain symptom nursing intervention effect simulation data; Step S232: Cluster the differences in the effects of similar cases for the symptom nursing intervention effect simulation data based on case-based reasoning and ICU nursing historical case data to obtain similar case effect difference clustering data; Step S233: Analyze the stability of the effect clustering for the similar case effect difference clustering data to obtain effect clustering stable data; Step S234: Analyze the nursing intervention efficacy based on the stable data clustered by effects to obtain the symptom nursing intervention efficacy data.

[0015] The present invention performs an intervention effect matching simulation on the interpolation data of symptom treatment tolerance through case reasoning technology during the nursing process, aiming to simulate the specific impacts of different nursing intervention measures on the improvement of patients' symptoms. Case reasoning allows the combination of experience and data from previous clinical cases. By establishing corresponding models, it predicts the responses of current patients under specific nursing measures. This process can not only provide nursing staff with a preliminary estimate of the potential effects of intervention measures but also help identify personalized nursing strategies that are most suitable for the current patients. Ultimately, the simulated data of symptom nursing intervention effects will provide a scientific basis for subsequent intervention decisions, making the nursing process more targeted and effective, thereby improving the overall treatment experience and satisfaction of patients. Using case reasoning and ICU nursing historical case data, a clustering of the differences in the effects of similar cases on the simulated data of symptom nursing intervention effects is carried out. The purpose is to identify cases similar to the current patient from a large amount of historical data and analyze the differences in their effects. Through this clustering analysis, the nursing team can discover potential patterns and trends, thereby evaluating the actual effects of different nursing measures in similar situations. This step will help nursing staff quickly obtain the effect comparison of relevant cases, provide a reference basis for the current treatment plan, and ensure that successful case experiences can be learned from when formulating nursing measures. At the same time, the clustering results can also reveal which factors have had a significant impact on the effects in different cases, thereby providing support for subsequent personalized nursing decisions. An analysis of the stability of the effect clustering of the data on the differences in the effects of similar cases is carried out, aiming to evaluate whether the effect data obtained through clustering is consistent and reliable. The analysis of the stability of the effect clustering can reveal the stability between different clusters and judge whether the clustering results are affected by extreme values or noise in individual cases. Through this analysis, nursing staff can confirm which clustering results are robust and can be applied in actual nursing, thereby improving the scientific nature of clinical decisions. The stability analysis not only enhances the credibility of the data but also provides a solid foundation for subsequent analysis of the effectiveness of nursing interventions, ensuring that the nursing team can rely on these data for effective decision-making and ultimately promoting the patient's recovery process. Based on the stable data of the effect clustering, an analysis of the effectiveness of nursing interventions is carried out, aiming to evaluate the actual effectiveness of different nursing intervention measures in improving patients' symptoms. This analysis usually uses statistical analysis methods such as regression analysis and analysis of variance to quantify the effectiveness of different nursing measures. Through in-depth analysis of the data on the effectiveness of nursing interventions, the nursing team can clearly understand the degree of impact of each intervention method on the improvement of patients' symptoms and can identify the best nursing plan. This process not only helps optimize nursing practice but also improves the quality of nursing and the treatment satisfaction of patients. Ultimately, the results of the analysis of the effectiveness of nursing interventions will provide a scientific basis for future nursing decisions, ensuring that patients can receive personalized and effective nursing services during treatment, thereby promoting their recovery and the improvement of their quality of life.

[0016] Preferably, the effect clustering stability analysis of the similar case effect difference clustering data includes the following steps: Perform within-cluster dispersion analysis on the similar case effect difference clustering data to obtain the within-cluster discrete data of the effect clustering; Perform conflict effect analysis on the similar case effect difference clustering data according to the within-cluster discrete data of the effect clustering to obtain the clustering effect conflict effect data; Perform conflict commonality identification on the similar case effect difference clustering data according to the clustering effect conflict effect data to obtain the case effect conflict commonality data; Perform case effect life cycle analysis on the case effect conflict commonality data to obtain the case effect conflict life cycle data; Perform effect clustering stability analysis on the similar case effect difference clustering data based on the case effect conflict life cycle data to obtain the effect clustering stable data.

[0017] The present invention conducts within-cluster dispersion analysis on the data of the effect differences of similar cases, aiming to quantify the variability among cases within each cluster. This analysis helps identify which cases have significant differences in effects by calculating the dispersion degree of each case within the cluster. Low dispersion indicates relatively consistent effects among cases within the cluster, while high dispersion means there may be potential influencing factors or individual differences. The implementation of this step can provide basic data for subsequent effect analysis, ensuring that when deeply analyzing the clustering effect, the internal connections among cases can be fully considered, thus providing a reliable basis for nursing decisions. Conduct conflict effect analysis on the data of the effect differences of similar cases using the within-cluster dispersion data of the effect clusters, with the aim of identifying the effect conflicts that occur under different nursing interventions. By analyzing the performance of different cases in the cluster, factors leading to inconsistent effects, such as nursing methods, patient individual differences, or external environments, can be explored. This analysis can reveal potential risk points and problems, providing important improvement suggestions for subsequent nursing interventions. Understanding the sources of conflict effects will help nursing staff consider these complex factors when formulating nursing plans, thereby reducing the occurrence of adverse effects and improving the quality of nursing and patient satisfaction. Conduct conflict commonality identification on the data of the effect differences of similar cases based on the conflict effect data of the clustering effect, aiming to find out the common conflict factors that appear in different cases. This process can reveal which factors frequently lead to inconsistent effects in different nursing cases, and then identify the problem areas that need special attention. By identifying these common conflicts, the nursing team can adjust nursing strategies and optimize nursing plans in a targeted manner to ensure better adaptation to the individual needs of patients during treatment. This analysis not only helps improve the effectiveness of current nursing interventions but also provides data-based decision support for subsequent nursing practice. Conduct case effect life-cycle analysis on the data of the commonality of case effect conflicts, aiming to understand the characteristics of these conflict factors changing over time during the patient care process. By analyzing the life cycle of the conflicts, the nursing team can identify long-term and short-term factors affecting the effect, and then evaluate the persistence and stability of different nursing interventions. This analysis will help clarify when and under what circumstances specific conflicts will have a greater impact on the effect, thus providing a dynamic perspective for nursing decisions. Through in-depth study of the life-cycle data, nursing staff can better manage patients' expectations and responses, improving the effectiveness and safety of treatment. Conduct effect clustering stability analysis on the data of the effect differences of similar cases based on the life-cycle data of case effect conflicts, aiming to evaluate the reliability and consistency of the clustering effect at different time periods. Through this analysis, the nursing team can determine which clustering results are robust and which clusters affect the effect due to conflicts at specific time points. The implementation of this process will provide important information for nursing decisions, ensuring that in practical applications, intervention measures that show stable effects in various situations can be selected.Finally, the obtained effect clustering stable data will provide a solid theoretical basis for optimizing the patient care plan and improve the scientificity and effectiveness of clinical nursing.

[0018] Preferably, step S3 includes the following steps: Step S31: Perform an analysis of reducing the probability of complication occurrence on the multi-dimensional dynamic data of physiological sign indicators based on the symptom nursing intervention efficacy data to obtain the data of reducing the probability of complication occurrence; Step S32: Perform a positive correlation analysis of nursing effectiveness on the data of reducing the probability of complication occurrence based on the symptom nursing intervention efficacy data to obtain the positive correlation data of nursing effectiveness; Step S33: Perform an abnormal decision deduction on the data of reducing the probability of complication occurrence based on the symptom nursing intervention efficacy data and the positive correlation data of nursing effectiveness to obtain the symptom nursing abnormal decision deduction data.

[0019] Based on the efficacy data of symptom-based nursing interventions, this invention conducts an analysis to reduce the probability of complications for the multi-dimensional dynamic data of physiological sign indicators, aiming to identify the impact of nursing interventions on the complication incidence rate. Through integrating the physiological data of patients and the results of previous nursing interventions, this analysis can predict how specific nursing measures reduce the probability of complications. Through this process, the nursing team can quantify the effectiveness of different intervention measures in reducing the risk of complications, thereby providing data support for clinical decision-making. Ultimately, the obtained data on the reduction of the probability of complications can provide a reference for nursing staff to formulate more effective nursing plans, ensuring that patients can avoid complications as much as possible during treatment, improving the overall nursing quality and patient safety. Based on the efficacy data of symptom-based nursing interventions, a positive correlation analysis of nursing effectiveness is conducted on the data of the reduction of the probability of complications, aiming to identify the positive correlation between nursing interventions and the reduction of the probability of complications. Through statistical methods, this analysis explores the relationship between the effect of nursing interventions and the risk of complications, thereby evaluating the effectiveness of specific intervention measures in reducing the complication incidence rate. Through this process, the nursing team can identify which nursing measures are more effective and can significantly reduce the probability of complications for patients, thus optimizing the nursing plan. In addition, the acquisition of positive correlation data on nursing effectiveness will contribute to decision-making in future nursing practice, improving the overall nursing effect and patient satisfaction. According to the efficacy data of symptom-based nursing interventions and the positive correlation data of nursing effectiveness, an abnormal decision deduction is conducted on the data of the reduction of the probability of complications, aiming to identify the abnormality of the effect of nursing interventions under specific circumstances. Through establishing a decision-making model, this process simulates and deduces the nursing effect under different circumstances, thereby revealing potential risk points and decision-making mistakes. Through abnormal decision deduction, the nursing team can identify which nursing measures cannot effectively reduce the risk of complications under specific conditions, and then provide a basis for timely adjustment of nursing strategies. This analysis not only improves the scientific nature of nursing decision-making but also effectively prevents and controls the occurrence of complications, thereby enhancing patient safety and treatment effect. Ultimately, the obtained data on abnormal decision deduction of symptom-based nursing will provide an important reference for nursing staff to respond in complex situations, ensuring that patients receive the best nursing services.

[0020] Preferably, step S33 includes the following steps: Step S331: Conduct a dynamic risk simulation of complications on the data of the reduction of the probability of complications based on the efficacy data of symptom-based nursing interventions and the positive correlation data of nursing effectiveness, to obtain dynamic risk simulation data of complications; Step S332: Identify abnormal triggering factors for the dynamic risk simulation data of complications to obtain risk abnormal triggering factors; Step S333: Conduct an abnormal decision deduction on the data of the reduction of the probability of complications based on the risk abnormal triggering factors to obtain data on abnormal decision deduction of symptom-based nursing.

[0021] Based on the data of the efficacy of symptom-based nursing intervention and the positive correlation data of nursing effectiveness, the present invention conducts dynamic risk simulation of complications on the data of the reduction in the probability of complication occurrence. This process constructs mathematical models and algorithms, and uses historical data to simulate the risk changes of patients developing complications under different nursing intervention measures within a certain time range. Through dynamic risk simulation, the nursing team can evaluate in real time the probability of complications occurring in different situations, and then identify nursing measures or patient conditions with higher risks. This analysis provides a forward-looking reference for clinical nursing, ensuring that nursing interventions can adapt to the changes of patients and taking timely and effective preventive measures in high-risk situations, thereby reducing the incidence of complications and improving patient safety. Identifying abnormal triggering factors for the dynamic risk simulation data of complications aims to find specific factors that significantly increase the risk of complications during the simulation process. This analysis uses data mining and statistical analysis techniques to identify abnormal factors closely related to the occurrence of complications, such as specific changes in physiological indicators, delays in nursing interventions, or other clinical events. By identifying these risk abnormal triggering factors, the nursing team can gain a deeper understanding of the potential mechanisms of complication occurrence and provide a scientific basis for subsequent intervention measures. In addition, identifying triggering factors can help nurses better monitor the patient's condition in actual operation, take preventive measures earlier, and reduce the occurrence of adverse events. Based on the identification results of risk abnormal triggering factors, abnormal decision deduction is carried out on the data of the reduction in the probability of complication occurrence, aiming to simulate and evaluate the effectiveness and potential risks of nursing interventions in the presence of abnormal triggering factors. This process establishes a decision-making model to analyze how the adjustment of nursing measures affects the incidence of complications under the influence of different triggering factors. Through this deduction, the nursing team can identify nursing decisions that are ineffective under specific risk conditions and adjust intervention measures in a timely manner to optimize the nursing plan. The results of this analysis can not only enhance the scientific nature of nursing decisions, but also effectively reduce the risk of complications caused by nursing mistakes, thereby improving the treatment effect and overall nursing quality of patients. Finally, the abnormal decision deduction data of symptom-based nursing provides an important basis for nurses to make decisions in a complex clinical environment, ensuring that patients receive the most suitable nursing services.

[0022] Preferably, step S4 includes the following steps: Step S41: Conduct a logistic regression analysis on the abnormal decision deduction data of symptom-based nursing to obtain abnormal decision logistic regression data; Step S42: Conduct a sensitivity analysis of abnormal nursing decisions based on the abnormal decision logistic regression data to obtain abnormal nursing decision sensitivity data; Step S43: Based on the random forest algorithm, construct an ICU nursing decision support plan for the abnormal nursing decision sensitivity data to obtain an ICU nursing decision support plan.

[0023] Through logical regression analysis of the data on abnormal decision-making deduction for symptom care in the present invention, it aims to explore the relationship between abnormal decision-making and the occurrence of complications. This process involves establishing a logical regression model to quantify the impact of different nursing decisions on the health outcomes of patients under specific conditions. Through this analysis, the nursing team can identify the key factors affecting the effectiveness of nursing decisions and assign corresponding risk scores to each decision. This quantitative analysis method can provide an important theoretical basis for clinical nursing, helping nursing staff understand under what circumstances specific decisions will lead to adverse outcomes, thus providing a direction for future nursing practice. Ultimately, the generated data on abnormal decision-making logical regression provides a basis for subsequent sensitivity analysis, enhancing the scientific nature and rationality of nursing decisions. Based on the data on abnormal decision-making logical regression, sensitivity analysis of abnormal nursing decisions is carried out, aiming to evaluate the sensitivity of different nursing decisions to the health outcomes of patients. This analysis determines through statistical methods the volatility of patient outcomes and their degree of response to different factors under various nursing decisions. Through this process, the nursing team can identify which nursing decisions have a greater impact on changes in patient health outcomes under specific circumstances, that is, the decisions with higher sensitivity. This can not only help nursing staff better understand the consequences of decisions but also enable more flexible coping strategies in clinical practice to focus on monitoring and adjusting high-sensitivity decisions. The ultimately obtained data on sensitivity of abnormal nursing decisions provides a basis for further constructing a scientific and effective nursing decision support plan, thus enhancing patient safety and nursing quality. The random forest algorithm is used to analyze the data on sensitivity of abnormal nursing decisions to construct an ICU nursing decision support plan. As a powerful machine learning method, the random forest algorithm can effectively process complex clinical data, identify the relationships between multiple variables, and generate a series of decision rules. By combining the results of sensitivity analysis with other clinical data, the nursing team can construct a comprehensive nursing decision support system. This system can not only evaluate the potential impact of nursing decisions in real time but also provide personalized nursing suggestions to formulate the optimal nursing plan for each patient in the ICU. The ultimately obtained ICU nursing decision support plan will help nursing staff make more scientific and accurate decisions in a complex clinical environment, improve nursing efficiency and patient safety, and reduce the incidence of complications.

[0024] The beneficial effects of the present invention are as follows. By obtaining the historical case data of ICU care, clinical information of patients in intensive care can be systematically collected. These data usually include patients' physiological indicators, medical histories, treatment plans and their responses, etc. Through the analysis of these data, the extraction of structured time-series sign changes can transform these complex time-series data into an easily processable structured format. This process not only helps to identify the changing trends of patients' states, but also provides a basis for subsequent data analysis and processing, enabling nursing staff to more intuitively understand the changes in patients' conditions and make timely nursing decisions accordingly. For the structured time-series sign change data, tolerance reduction interpolation processing is carried out to fill in the blanks caused by data missing or interruption, ensuring the continuity and integrity of the data. Through this interpolation method, the tolerance of patients to symptom treatment can be more accurately evaluated. In addition, by analyzing the efficacy of nursing interventions on the interpolated data, the actual effects of different nursing measures in improving patients' symptoms can be revealed. This process not only helps nursing staff understand the effectiveness of the nursing measures adopted, but also provides data support for further optimizing the nursing plan, thereby improving the overall nursing quality and safety of patients. Through the analysis of the efficacy data of symptom nursing interventions, a multi-dimensional dynamic assessment of physiological sign indicators is carried out. This step focuses on the comprehensive analysis of patients' physiological indicators, such as the comprehensive changes of multiple indicators such as heart rate, blood pressure, and respiratory rate. This assessment can be carried out by constructing a multi-dimensional model and using data analysis tools to deeply analyze the correlation between various physiological indicators and their dynamic changing trends. This not only helps to identify the short-term and long-term effects of nursing intervention measures, but also provides a scientific basis for formulating personalized nursing plans clinically. Finally, this series of analyses and assessments will greatly improve the accuracy and personalization of intensive care, thereby improving the treatment effect and quality of life of patients. First, a reduction analysis of the probability of complications is carried out on the multi-dimensional dynamic data of physiological sign indicators. The key to this process lies in analyzing the relationship between the changes of different physiological indicators of patients and the occurrence of complications in the ICU environment through statistical and machine learning methods, identifying potential risk factors, and calculating the probability of complications occurring in different situations. This kind of analysis can help the clinical team identify high-risk patients in advance, so as to take preventive measures and reduce the incidence of complications. Next, an abnormal decision deduction is carried out on the data of the reduction of the probability of complications, and the results based on probability analysis are deeply mined to identify the key factors leading to nursing abnormalities. This deduction can not only reveal the potential risk points in the nursing process, but also provide empirical evidence for subsequent clinical decisions, ensuring that patients receive timely and effective interventions, and ultimately improving the safety and effect of nursing. The random forest algorithm is used to analyze the data of the abnormal decision deduction of symptom nursing to construct an ICU nursing decision support plan. The random forest algorithm is a powerful ensemble learning method that can handle high-dimensional data and effectively identify the complex relationship between features and results.In this step, the algorithm comprehensively considers various variables in historical cases and identifies key factors related to the patient's symptoms, complication risks, and nursing measures. Through this analysis, scientific decision-making support can be provided for ICU nurses, including recommended nursing measures, strategies for preventing complications, and personalized nursing plans. This not only improves the accuracy and effectiveness of nursing but also helps clinical staff make wise decisions quickly in complex situations, thereby improving the overall nursing experience and prognosis of patients. This data-driven decision-making support solution will significantly improve the nursing quality and efficiency in the ICU. Therefore, the present invention makes an improvement to the construction method of a traditional case-based reasoning ICU nursing decision support system, solves the problem that the construction method of a traditional case-based reasoning ICU nursing decision support system has inaccurate analysis of patient signs, resulting in large nursing decision errors, improves the accuracy of patient sign analysis, and reduces the errors of nursing decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a schematic flowchart of the steps of a construction method of a case-based reasoning ICU nursing decision support system; Figure 2 is Figure 1 a detailed implementation step flowchart of step S2 in

[0026] The realization, functional features, and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those skilled in the art within the scope of the present invention without creative efforts belong to the protection scope of the present invention.

[0028] In addition, the accompanying drawings are only schematic diagrams of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and thus their repeated description will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0029] It should be understood that although terms such as "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly, the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0030] To achieve the above object, please refer to Figures 1 to 2 , a method for constructing an ICU nursing decision support system based on case-based reasoning, the method comprising the following steps: Step S1: Obtain ICU nursing historical case data; perform structured time-series sign change extraction based on the ICU nursing historical case data to obtain structured time-series sign change data; Step S2: Perform tolerance reduction interpolation processing on the structured time-series sign change data to obtain symptom treatment tolerance interpolation data; perform nursing intervention efficacy analysis on the symptom treatment tolerance interpolation data to obtain symptom nursing intervention efficacy data; perform multi-dimensional dynamic evaluation of physiological sign indicators based on the symptom nursing intervention efficacy data to obtain multi-dimensional dynamic data of physiological sign indicators; Step S3: Perform complication occurrence probability reduction analysis on the multi-dimensional dynamic data of physiological sign indicators to obtain complication occurrence probability reduction data; perform abnormal decision deduction on the complication occurrence probability reduction data to obtain symptom nursing abnormal decision deduction data; Step S4: Construct an ICU nursing decision support plan based on the symptom nursing abnormal decision deduction data by using the random forest algorithm to obtain an ICU nursing decision support plan.

[0031] In an embodiment of the present invention, referring to Figure 1 as described, it is a schematic diagram of the step flow of a method for constructing an ICU nursing decision support system based on case-based reasoning of the present invention. In this example, the method for constructing an ICU nursing decision support system based on case-based reasoning comprises the following steps: Step S1: Obtain ICU nursing historical case data; perform structured time-series sign change extraction based on the ICU nursing historical case data to obtain structured time-series sign change data; In the embodiments of the present invention, first, for the nursing historical case data in the ICU, core indicators related to the changes in the patient's vital signs are extracted, including physiological signals such as heart rate, blood pressure, blood oxygen saturation, body temperature, etc. By directly obtaining data from the ICU historical case database, data cleaning and preprocessing are carried out to ensure the accuracy and consistency of the data. Using the time series analysis method in statistical analysis, the multi-dimensional vital sign data of the patient is classified and arranged in chronological order. In order to avoid the influence of data noise on the analysis results, wavelet transform is applied to smooth the time series fluctuations in the data to retain the key vital sign change characteristics. The extracted vital sign data is standardized, and correlation analysis is applied to identify the dependence relationships between different vital sign data. Finally, an algorithm based on structured time series pattern mining is used to extract the time series features of the data and generate structured time series vital sign change data. This data reflects the dynamic evolution trend of each patient's vital signs during the treatment process.

[0032] Step S2: Perform tolerance reduction interpolation processing on the structured time series vital sign change data to obtain symptom treatment tolerance interpolation data; perform nursing intervention efficacy analysis on the symptom treatment tolerance interpolation data to obtain symptom nursing intervention efficacy data; perform multi-dimensional dynamic evaluation of physiological vital sign indicators based on the symptom nursing intervention efficacy data to obtain multi-dimensional dynamic data of physiological vital sign indicators; In the embodiments of the present invention, after obtaining the structured time-series physical sign change data, interpolation analysis is performed on the reduction of the physical sign tolerance during the patient's treatment process using the B-spline interpolation method. This method identifies the fluctuation range of the physical sign tolerance during the treatment process by establishing a functional relationship between the patient's physical sign changes and their treatment tolerance. For the moments when the tolerance decreases, data interpolation processing is carried out to accurately predict the tolerance changes of the patient at different time points. Then, the results of the interpolation processing are integrated into the tolerance reduction assessment model, and by calculating the fitting curve of the physical sign tolerance and time, the symptom treatment tolerance of the patient in each time period is analyzed. In this process, a numerical optimization method based on gradient descent is used to ensure the accuracy and continuity of the interpolation results, generating the interpolated data of the symptom treatment tolerance. Subsequently, the nursing intervention efficacy analysis is carried out based on the interpolated data of the symptom treatment tolerance. The multiple linear regression algorithm in regression analysis is used to calculate the intervention effects of various nursing measures by analyzing the efficacy relationship between the patient's different symptoms and the nursing intervention measures. Specifically, by constructing a mathematical model of the patient's symptoms and the nursing intervention effects, the influence degree of each nursing measure on different physical sign changes is evaluated, and then the symptom nursing intervention efficacy data is obtained. This data reflects the effectiveness of the nursing measures and their specific improvement effects on the patient's symptoms. After obtaining the symptom nursing intervention efficacy data, a comprehensive assessment of the patient's physiological signs is carried out based on the multi-dimensional dynamic analysis theory. This step uses the principal component analysis method (PCA) to reduce the dimension of the patient's various physical sign indicators and evaluate the overall health status of the patient from multiple dimensions. The most representative physical sign change characteristics are extracted through the PCA method to construct a multi-dimensional dynamic assessment model, and the trend of the patient's physical sign changes and their correlation with the nursing measures are analyzed. At the same time, combined with the time-series regression model, the change trends of various physical sign indicators in the future period are predicted, generating the multi-dimensional dynamic data of the physiological sign indicators. During the operation process, a dynamic system modeling method based on the Markov chain is used to analyze the mutual influence and feedback mechanism between various physical sign indicators. By performing a time-series dependence analysis on the physical sign changes in different time periods, the stability of the patient's physical signs and the possibility of future changes are evaluated. Finally, the generated multi-dimensional dynamic data of the physiological sign indicators can provide a comprehensive view of the patient's physical sign changes for the nursing staff, assisting them in formulating more accurate nursing plans.

[0033] Step S3: Perform an analysis on the reduction of the complication occurrence probability for the multi-dimensional dynamic data of the physiological sign indicators to obtain the data on the reduction of the complication occurrence probability; perform an abnormal decision deduction on the data on the reduction of the complication occurrence probability to obtain the data on the abnormal decision deduction of the symptom nursing. In the embodiments of the present invention, first, for the multi-dimensional dynamic data of physiological sign indicators, a Bayesian statistical method is used to perform a loss analysis on the probability of complication occurrence. The specific operation process is to construct a conditional probability model of the patient's sign changes and the probability of complication occurrence, use the known sign data as input, and calculate the initial probability of each complication occurrence. Then, a probability update technique based on maximum likelihood estimation is applied to gradually update the probability of complication occurrence of the patient at different time nodes. During this process, by analyzing the abnormal fluctuations and abnormal characteristics of each sign data, the key indicators that have a greater impact on the occurrence of complications are screened out. Subsequently, the Markov decision process is used to dynamically adjust the complication risk at each time step, identify the high-risk intervals leading to complications, and then perform a loss process on the initial probability to generate the loss data of the complication occurrence probability. This process ensures a more accurate assessment of the patient's complication risk. After generating the loss data of the complication occurrence probability, the dynamic programming method is used to perform an abnormal decision deduction on this data. The dynamic programming method constructs a state transition matrix to describe the possible abnormal decision paths under different sign states. Through a recursive algorithm, the historical changes of the patient's signs are traced back to deduce the key time nodes and decision paths that lead to abnormal nursing decisions. During the operation process, multivariate covariance analysis is used to identify the high degree of correlation between the patient's sign changes and the potential causes of decision abnormalities. On this basis, a deduction model is constructed to identify the potential abnormal decision behaviors in each nursing measure. This model gradually simulates all the abnormal decision paths in ICU nursing and corrects the deduction results during the process to ensure the accuracy and reliability of the deduction results. Finally, the abnormal decision deduction data for symptom nursing is generated, which can be used to further optimize the nursing process.

[0034] Step S4: Based on the random forest algorithm, construct an ICU nursing decision support plan for the abnormal decision deduction data of symptom nursing to obtain the ICU nursing decision support plan.

[0035] In the embodiments of the present invention, based on the abnormal decision deduction data of symptom care, a random forest algorithm is used to construct an ICU nursing decision support scheme. The random forest algorithm constructs multiple decision tree models, trains each tree separately, and each tree generates a set of decision suggestions according to the different physical signs changes and deduction data of the patient. The random forest, through the way of ensemble learning, weights and votes the results of multiple decision trees to obtain the optimal nursing decision scheme. In specific implementation, first, a training set and a test set are extracted from the abnormal decision deduction data through random sampling technology to ensure the diversity of the decision tree models. Each decision tree is independently trained on different random samples, and the Gini coefficient is used to measure the importance of each feature, thereby optimizing the accuracy of the nursing decision. During the training process, combined with the multi-dimensional physical sign dynamic data of the patient, a recommendation scheme for nursing measures is generated. Finally, the random forest algorithm aggregates the prediction results of all decision trees to generate an ICU nursing decision support scheme, ensuring that the optimal nursing suggestions can be provided in different nursing scenarios. This scheme corrects the abnormal behaviors in the nursing process and optimizes the accuracy of nursing intervention.

[0036] Preferably, step S1 includes the following steps: Step S11: Obtain the historical case data of ICU nursing; Step S12: Classify the historical case data of ICU nursing by subject to obtain the subject data of ICU nursing cases; Step S13: Fill in the missing values of the nursing content for the subject data of ICU nursing cases to obtain the filled subject nursing content data; Step S14: Extract the structured time-series physical sign changes according to the filled subject nursing content data to obtain the structured time-series physical sign change data.

[0037] In the embodiments of the present invention, historical ICU nursing case data is obtained by directly connecting to the ICU database through a data interface. The obtained nursing case data includes detailed records during the period from the patient's admission to discharge, covering multiple data dimensions, such as the patient's basic information, diagnosis and treatment process, drug usage, vital sign monitoring records, nursing operation records, etc. During the data acquisition process, specialized data screening and cleaning tools are used to first eliminate duplicate or abnormal nursing records to ensure the accuracy and validity of the data. By directly accessing the ICU database, the integrity of data collection is guaranteed, and it is ensured that the obtained data has time stamps and specific records of the nursing process. All data is processed in accordance with data privacy and security requirements, and is transmitted and stored in an encrypted manner to ensure the confidentiality of sensitive information. For the obtained historical ICU nursing case data, subject classification is first performed based on the patient's diagnosis information, ward allocation, and main disease characteristics. Text mining techniques and natural language processing (NLP) techniques are used to analyze the text data in the medical records and nursing records, and key feature words related to the subject are extracted. By constructing a predefined subject classification dictionary and using a rule matching algorithm, each nursing record is classified according to the ICU subject it belongs to, generating structured data on the subjects to which the ICU nursing cases belong. To ensure the accuracy of classification, a consistency check mechanism is used. By marking and processing outliers and classification conflicts in the data, the integrity and reliability of the classification results are ensured. During this process, the subject classification results are arranged according to the time axis to ensure that each case corresponds to the correct subject classification during its nursing cycle. In the data on the subjects to which the ICU nursing cases belong, there may be missing data in some nursing records, especially in cases where the patient transfers to another department, there are sudden situations, or the nursing records are incomplete. To fill in these missing values, multiple interpolation methods are used to fill the missing data. First, for different subjects, cases with similar medical conditions and similar nursing processes in the historical data are used for data comparison to find the reference record that best matches the current case. Then, a covariance-based interpolation algorithm is applied to fill in the missing nursing content data according to the time series. During the specific operation process, the specific requirements of different subject nursings are combined to ensure the rationality of the filled data. For example, for the missing vital sign data, linear interpolation is used to calculate the continuous time series, and for the missing nursing measure data, rule interpolation based on clinical standards is used for supplementation. After data filling, it is ensured that each record has complete nursing content, generating filled data on the nursing content of the subjects to which it belongs. Based on the filled data on the nursing content of the subjects to which it belongs, time series analysis techniques are used to structurally extract the patient's physiological sign data. By analyzing key sign indicators such as the patient's heart rate, blood pressure, and blood oxygen saturation according to the time series, their dynamic changes during the entire nursing process are extracted.In specific operations, first, the sliding window method is adopted to segment the vital sign data at a set time interval, and the average value, standard deviation, and fluctuation trend of the vital sign changes within each time window are calculated. Then, the autoregressive integrated moving average (ARIMA) model is used to predict and model the long-term trend and short-term fluctuations of the vital sign changes. Through this model, abnormal fluctuations and key change nodes in the vital sign changes are identified, and this data is structured and stored to form structured time-series vital sign change data with time stamps. The finally generated time-series vital sign change data will serve as an important basis for subsequent complication prediction and nursing decision support.

[0038] Preferably, step S2 includes the following steps: Step S21: Analyze the symptom treatment trend of the structured time-series vital sign change data to obtain symptom treatment trend data; Step S22: Perform tolerance reduction interpolation processing on the symptom treatment trend data to obtain symptom treatment tolerance interpolation data; Step S23: Analyze the nursing intervention efficacy of the symptom treatment tolerance interpolation data based on case-based reasoning and ICU nursing historical case data to obtain symptom nursing intervention efficacy data; Step S24: Perform multi-dimensional dynamic assessment of the physiological sign indicators on the structured time-series vital sign change data according to the symptom nursing intervention efficacy data to obtain multi-dimensional dynamic data of the physiological sign indicators.

[0039] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes: Step S21: Analyze the symptom treatment trend of the structured time-series vital sign change data to obtain symptom treatment trend data; In the embodiment of the present invention, the structured time-series vital sign change data is analyzed by time series analysis technology to extract the symptom change trend of the patient during the treatment. The specific operations include using the time series decomposition method to decompose the patient's physiological sign data into three parts: long-term trend, periodic fluctuation, and random fluctuation. First, the long-term change trend of the data is identified through the weighted moving average method, especially the trend of symptom improvement or deterioration of the patient during the treatment process. Subsequently, the Fourier transform method is used to process the periodic fluctuation part to extract the potential periodic change factors in the vital sign data. For the remaining random fluctuation part, an autoregressive model is used for noise filtering. Through the above steps, a detailed symptom treatment trend data is generated, including the vital sign change trend, periodic fluctuation characteristics, and annotation of random abnormal points of the patient during the treatment, laying a foundation for subsequent tolerance analysis.

[0040] Step S22: Perform tolerance reduction interpolation processing on the symptom treatment trend data to obtain symptom treatment tolerance interpolation data; In the embodiments of the present invention, the symptom treatment trend data is processed to evaluate the symptom treatment tolerance of patients. In the specific process, the missing points in the symptom treatment trend data are first filled by the interpolation method. The spline interpolation method is used for data interpolation at different treatment time points to ensure the continuity of the time series. Subsequently, in combination with the patient's clinical background and nursing records, the tolerance reduction value of the patient at different treatment stages is calculated. Here, a tolerance reduction function based on linear interpolation is used to combine the symptom treatment trend of the patient with the trend of the decline in their tolerance, and the specific value of the tolerance reduction is obtained. The tolerance reduction interpolation process not only takes into account the overall trend during the treatment period but also includes the correction process for local symptom fluctuations, and finally generates a set of symptom treatment tolerance interpolation data reflecting the changes in the patient's treatment tolerance.

[0041] Step S23: Perform a nursing intervention efficacy analysis on the symptom treatment tolerance interpolation data based on case reasoning and ICU nursing historical case data to obtain symptom nursing intervention efficacy data; In the embodiments of the present invention, through the combination of case reasoning technology and ICU nursing historical case data, an analysis of the nursing intervention efficacy of the symptom treatment tolerance interpolation data is carried out. First, the case-based reasoning (CBR) method is adopted, and historical cases similar to the current patient's symptoms are selected as references. In the specific operation, through the similarity calculation of a large number of records in the ICU nursing historical case database, the K-nearest neighbor algorithm (KNN) is used to find the case set closest to the current situation. Then, based on the effectiveness of the nursing intervention measures in these reference cases, an efficacy analysis model is applied to evaluate the expected effects of implementing the same or similar nursing measures in the current situation. This efficacy analysis model uses a weighted function to quantitatively evaluate different intervention measures according to the changes in symptom treatment tolerance, and generates the expected efficacy scores for each intervention measure. After comprehensive evaluation, a set of symptom nursing intervention efficacy data is obtained, which specifically reflects the impact degree of each intervention measure on the current patient's symptoms.

[0042] Step S24: Perform a multi-dimensional dynamic assessment of the physiological sign index on the structured time-series sign change data according to the symptom nursing intervention efficacy data to obtain multi-dimensional dynamic data of the physiological sign index.

[0043] In the embodiments of the present invention, a multi-dimensional dynamic analysis method is used to further evaluate the structured time-series physical sign change data. First, by correlating the symptom care intervention efficacy data with the structured time-series physical sign change data, the specific impact of the care intervention measures on the patient's physiological signs is identified. In the operation, the principal component analysis (PCA) technique is adopted to reduce the dimension of the multi-dimensional sign data, and the most important several variables are extracted as the main evaluation objects. Then, using the multiple regression analysis method, a fitting analysis is performed on each main sign index and the symptom care intervention efficacy data to establish a quantitative relationship between the sign change and the intervention efficacy. Through these analyses, a set of multi-dimensional dynamic sign index data is generated, reflecting the sign change trend and fluctuation characteristics of the patient under the influence of the care intervention measures. This process ensures a comprehensive evaluation of the physiological sign changes and provides support for subsequent complication prediction and care decision-making.

[0044] Preferably, step S22 includes the following steps: Step S221: Perform a clinical response improvement trend analysis on the symptom treatment trend data to obtain clinical response improvement trend data; Step S222: Perform a time-series discontinuity difference analysis on the clinical response improvement trend data to obtain response improvement time-series discontinuity difference data; Step S223: Calculate the median of the improvement differences between adjacent time series for the response improvement time-series discontinuity difference data to obtain the median of the improvement differences between adjacent time series; Step S224: Perform a linear structure optimization on the clinical response improvement trend data according to the median of the improvement differences between adjacent time series to obtain response improvement trend linear data; Step S225: Perform a global variance nearest neighbor interpolation on the response improvement trend linear data to obtain improvement trend global interpolation data; Step S226: Perform a tolerance reduction interpolation process on the symptom treatment trend data according to the improvement trend global interpolation data to obtain symptom treatment tolerance interpolation data.

[0045] In the embodiments of the present invention, through time series analysis and trend extraction techniques, clinical response improvement trends are analyzed for symptom treatment trend data. The specific operations include, first, smoothing the symptom treatment data to reduce noise interference. The commonly used smoothing method is the moving average method. Subsequently, a linear regression model is used to analyze the change trends of clinical responses at different time points, and by fitting the change relationship between the treatment duration and the degree of symptom response, a trend line representing the improvement of the patient's clinical response is obtained. This trend line reflects whether the patient's symptoms show continuous improvement during the treatment process. The generated clinical response improvement trend data will be used as the basic data input for subsequent analysis. To identify whether there are obvious time discontinuity differences during the clinical response improvement process. First, the break point analysis technique (Break Point Analysis) is used to segment the clinical response improvement trend data, and the break points existing in the trend line are mainly identified. The break points can be sudden fluctuations in the patient's condition or changes in treatment effects at certain moments. After identifying the break points, the difference method is used to calculate the improvement differences in adjacent time periods, and by comparing the difference amplitudes between each segment, the response improvement characteristics in different time periods are judged. Finally, response improvement time series discontinuity difference data is generated, which details the change rate of improvement in each time period and provides a basis for subsequent data processing. By using statistical methods to calculate the median of adjacent time periods for the time series difference data, the typical values of symptom response improvement in adjacent time periods are obtained. During the operation process, first select a group of adjacent time periods and sort the improvement differences between each time period. Subsequently, calculate the median of the difference values of each group of adjacent time periods to avoid the influence of extreme values on the data results. The median can better reflect the typical improvement amplitude between adjacent time periods and can accurately capture the continuous and stable changes during the treatment process. The finally obtained median of the improvement differences between adjacent time series provides an accurate benchmark for subsequent structural optimization. Perform linear structure optimization on the previously generated clinical response improvement trend data. The specific operations include, first, using the median of the improvement differences between adjacent time series as the reference standard for optimization, and performing linear fitting on the trend data through the least squares method to eliminate local fluctuations and non-linear noise. Secondly, the interpolation method is used to fill in the missing time series data points to ensure the continuity of the linear structure of the clinical response trend. The optimized trend line can more clearly present the overall trend of the patient's clinical response, and the generated response improvement trend linear data provides a solid foundation for further global interpolation. Perform interpolation processing on the response improvement trend linear data through the global variance nearest neighbor interpolation technique. First, calculate the variance values between each time point based on the analysis of variance and identify the adjacent time series point pairs with the smallest variance. Then, apply the nearest neighbor interpolation method between these adjacent points to calculate the improvement trend values of the unobserved time series points through the linear interpolation method. This method ensures the smooth transition of the data within the global range and avoids data deviation caused by local abnormal fluctuations.The finally generated global interpolation data of the improvement trend can provide an accurate basis for the overall assessment of clinical responses. According to the previously generated global interpolation data of the improvement trend, tolerance reduction interpolation processing is performed on the symptom treatment trend data. First, by analyzing the change trend of treatment tolerance reflected in the global interpolation data of the improvement trend, the tolerance reduction value of the patient is calculated. In specific operations, a tolerance reduction function is used to combine the tolerance change amount at each time point with the treatment trend for differential compensation processing. Then, the piecewise linear interpolation method is used to interpolate the tolerance in the symptom treatment trend data to ensure the continuity and accuracy of the tolerance data in each time period. The finally obtained interpolation data of the symptom treatment tolerance provides high-precision basic data for the analysis of nursing intervention efficacy and decision support.

[0046] Preferably, the global variance nearest neighbor interpolation for the linear data of the response improvement trend includes the following steps: Using a preset improvement value assignment model to perform a neighboring trend improvement value assessment on the linear data of the response improvement trend to obtain neighboring trend improvement value data; Performing a global numerical variance calculation on the neighboring trend improvement value data to obtain a global trend numerical variance; Performing global variance nearest neighbor interpolation processing on the neighboring trend improvement value data according to the global trend numerical variance to obtain global interpolation data of the improvement trend.

[0047] In the embodiments of the present invention, first, an evaluation framework for linear data of reaction improvement trends needs to be established. This framework contains a set of preset improvement values. These improvement values are based on clinical experience data and represent the standard improvement amplitudes of patients at different treatment stages. Subsequently, numerical matching technology is used to compare these preset improvement values with the actual linear data of reaction improvement trends to identify improvement changes in adjacent time periods. By comparing the deviation between the reaction improvement changes in each time period and the standard value, the trend improvement characteristics in adjacent time periods can be evaluated. The result of this operation generates adjacent trend improvement numerical data, which reflects the specific improvement trends of patients in each time period. This process ensures the reasonable alignment of the preset improvement values and the actual data through the minimum deviation matching algorithm. The dispersion degree of the adjacent trend improvement numerical data is evaluated through variance calculation in statistics. First, the adjacent trend improvement numerical data is segmented to ensure that the data in each time period can reflect the overall trend of the patient's treatment. Next, the variance calculation formula is used to process these data segments. The specific steps are as follows: first, calculate the mean value of each time period, then add the squared differences between each data point and its mean value, and finally obtain the variance of this time period. By accumulating the variances of each time period, the overall global trend numerical variance is obtained. This variance value represents the fluctuation amplitude of the patient's improvement trend in the time series during the entire treatment process. The global trend numerical variance provides a quantitative basis for the nearest neighbor selection in subsequent interpolation processing. Based on the global trend numerical variance generated in the previous step, nearest neighbor interpolation processing is performed on the adjacent trend improvement numerical data. The specific operations include, first, identifying adjacent time periods with the smallest variance values in the global trend numerical variance. These time periods have a smaller fluctuation amplitude and stronger continuity. Subsequently, a global interpolation algorithm (such as Lagrange interpolation method or Newton interpolation method) is used to perform numerical interpolation between these adjacent time periods to ensure that reasonable estimated values can be obtained for the data points not observed in these time periods through interpolation. The core of the interpolation processing is to ensure the smooth transition of the data and ensure that the interpolated values are highly consistent with the original data through the variance nearest neighbor principle, and finally generate global interpolation data of the improvement trend. This data provides a stable basis for the accurate analysis of symptom treatment tolerance.

[0048] Preferably, step S23 includes the following steps: Step S231: Based on case-based reasoning, perform a matching simulation of the intervention effect during the nursing process on the interpolation data of symptom treatment tolerance to obtain symptom nursing intervention effect simulation data; Step S232: Based on case-based reasoning and ICU nursing historical case data, perform a clustering of the differences in the effects of similar cases on the symptom nursing intervention effect simulation data to obtain similar case effect difference clustering data; Step S233: Perform a stability analysis of the effect clustering on the similar case effect difference clustering data to obtain effect clustering stable data; Step S234: Analyze the nursing intervention efficacy based on the effect-clustered stable data to obtain the symptom nursing intervention efficacy data.

[0049] In the embodiments of the present invention, first, based on the theoretical framework of case-based reasoning, similar cases in the ICU nursing historical case data are associated with the interpolated data of symptom treatment tolerance. Case-based reasoning simulates the intervention effect in the current symptom nursing by analyzing the intervention measures and treatment results in previous similar nursing cases. The specific operations include extracting patient data similar to the current symptoms and evaluating the improvement of symptoms under specific nursing interventions. By establishing the mapping relationship between nursing interventions and patient symptom improvement, the possible range of the current nursing intervention effect can be deduced and matched with the interpolated symptom treatment tolerance data to generate the simulated data of symptom nursing intervention effect. The effect simulation in the matching process adopts numerical fitting and case comparison analysis methods to ensure the rationality and accuracy of the simulated data. After obtaining the simulated data of symptom nursing intervention effect, combined with the ICU nursing historical case data, the case-based reasoning method is used to conduct cluster analysis on the differences in the effects of similar cases. First, similar cases in the nursing historical case are extracted, and their nursing effects are compared with the current intervention effect simulation data. Subsequently, the differences in nursing effects between different cases are quantified through difference calculation methods (such as Euclidean distance or cosine similarity). After comparison, the K-means clustering or DBSCAN clustering algorithm is used to divide these cases into different clustering groups according to the similarity of nursing effects, and the cases within each group have similar nursing effect differences. Through this clustering process, the group of cases closest to the current symptom nursing intervention can be revealed, generating the cluster data of the differences in the effects of similar cases. By conducting a stability analysis on the cluster data of the differences in the effects of similar cases, the reliability of the clustering results is ensured. First, the clustering stability evaluation indicators (such as silhouette coefficient, CH index, or DB index) are used to evaluate the clustering results, and the compactness and separation degree of each clustering group are analyzed to judge the rationality of the cluster of the differences in the effects of similar cases. Then, cross-validation is performed on the nursing effect difference data within each clustering group. Some cases are randomly selected for re-clustering, and the deviation degree between the results and the original clustering is observed. If the deviation is small, it indicates that the clustering results have high stability; if the deviation is large, the clustering algorithm and the characteristic parameters of the input data need to be further optimized. Finally, through multiple iterations, the stable data of effect clustering is obtained to ensure the stability and consistency of the cluster analysis. Based on the stable data of effect clustering, a quantitative analysis of the nursing intervention efficacy is carried out. The specific operations include first extracting the nursing intervention effect data of the similar case group from the stable clustering data, and identifying the key factors affecting the nursing intervention effect through multivariate regression analysis or factor analysis. According to these factors, the efficacy of each intervention measure in the current symptom nursing is evaluated, that is, the contribution degree to the improvement of patient symptoms. In the analysis process, specific numerical evaluation criteria are adopted. For example, through the cumulative efficacy value of the nursing intervention (such as efficacy score or efficacy index), the effect of each intervention measure is quantified, thereby generating the final symptom nursing intervention efficacy data.This data provides an objective reference basis for clinical decision-making, enabling nursing staff to optimize nursing plans based on the results of efficacy evaluations.

[0050] Preferably, the effect clustering stability analysis of the clustering data of the effect differences of similar cases includes the following steps: Perform within-cluster dispersion analysis on the clustering data of the effect differences of similar cases to obtain within-cluster discrete data of effect clustering; Perform conflict effect analysis on the clustering data of the effect differences of similar cases according to the within-cluster discrete data of effect clustering to obtain clustering effect conflict effect data; Perform conflict commonality identification on the clustering data of the effect differences of similar cases according to the clustering effect conflict effect data to obtain case effect conflict commonality data; Perform case effect lifecycle analysis on the case effect conflict commonality data to obtain case effect conflict lifecycle data; Perform effect clustering stability analysis on the clustering data of the effect differences of similar cases based on the case effect conflict lifecycle data to obtain effect clustering stable data.

[0051] In the embodiments of the present invention, a dispersion analysis method in statistics is used to analyze each clustering cluster of the similar case effect difference clustering data. The specific operations include calculating the dispersion coefficients (such as standard deviation, variance) within each cluster to evaluate the similarity and consistency among the data points within the cluster. The lower the dispersion, the smaller the effect difference within the cluster, indicating a higher clustering tightness; the higher the dispersion, the greater the effect difference within the cluster, indicating inconsistent effects. Through the within-cluster dispersion analysis, the dispersion of the data within different clustering clusters can be quantified, thereby obtaining the within-cluster discrete data of the effect clustering clusters. This data provides a reference basis for identifying potential conflict effects in subsequent analysis. Based on the within-cluster discrete data of the effect clustering clusters, conflict effect analysis is carried out. The conflict effect analysis aims to identify the effect conflicts existing within and between different clustering clusters. The specific operation is to compare the differences in the case effects within each cluster, focusing on the clusters with larger dispersion, and analyze whether there are obvious conflicts in the nursing intervention effects within them. For example, some cases show opposite effects under the same nursing conditions. The conflict effect analysis method, such as the difference comparison matrix method or the conflict detection algorithm, is used to conduct paired analysis on the cases within each cluster to identify the cases with opposite or significantly different effects, thereby generating the clustering effect conflict effect data. This step provides the basic data support for the subsequent identification of conflict commonalities. Using the clustering effect conflict effect data, conflict commonalities are identified. The operation methods include first extracting all the case data with conflict effects and analyzing the common characteristics of these cases in terms of nursing intervention, patient symptom characteristics, treatment history, etc. The common characteristics among these conflict cases are identified through statistical methods (such as frequency analysis, association rule mining, etc.), such as the correlation between specific nursing plans or patient traits and conflict effects. Then, the cases with common characteristics are divided into conflict groups, and their common characteristics and the conflict effects shown are recorded to generate the case effect conflict commonality data. This data provides a basis for the subsequent case life cycle analysis, helping to better understand the sources and development laws of conflict effects. The case effect life cycle analysis aims to study the change trend and life cycle of the conflict case effects. First, based on the case effect conflict commonality data, track the change in the intervention effects of these cases during the nursing process and observe the fluctuation of their effects over time. By plotting the case effect life cycle curve, evaluate whether the conflict effect is short-term fluctuation or long-term existence, and analyze each stage of its life cycle (such as the initial effect fluctuation period, stable period, decline period). The analysis method uses time series analysis and life cycle models to segmentally model the effect changes of each conflict case to reveal its life cycle characteristics, thereby generating the case effect conflict life cycle data. This data can provide a reference for the subsequent stability analysis, helping to evaluate the long-term impact of the conflict effect. Using the case effect conflict life cycle data, stability analysis is carried out on the similar case effect difference clustering data.By comparing the life cycle characteristics of each clustering cluster with the changing trend of the case effect, analyze whether the clustering effect remains stable over time. The specific operations include calculating the stability indicators of each clustering cluster, such as the life cycle length, the range of effect fluctuations, and the effect persistence, etc. For clustering clusters with a short life cycle and large fluctuations, their stability is poor, and the clustering quality needs to be re-evaluated; for clusters with a long life cycle and stable effects, they have a high clustering stability. Through this process, finally, the effect clustering stable data is obtained to ensure that the analyzed clustering results have high reliability and persistence in practical applications.

[0052] Preferably, step S3 includes the following steps: Step S31: Perform an analysis of reducing the probability of complications for the multi-dimensional dynamic data of physiological sign indicators based on the symptom care intervention efficacy data to obtain the data of reducing the probability of complications; Step S32: Perform a positive correlation analysis of nursing effectiveness on the data of reducing the probability of complications based on the symptom care intervention efficacy data to obtain the positive correlation data of nursing effectiveness; Step S33: Perform an abnormal decision deduction on the data of reducing the probability of complications based on the symptom care intervention efficacy data and the positive correlation data of nursing effectiveness to obtain the symptom care abnormal decision deduction data.

[0053] In the embodiments of the present invention, based on the symptom care intervention efficacy data, the dynamic changes of the multi-dimensional physiological sign indicators of the patient are analyzed. First, by monitoring the correlation between physiological signs (such as heart rate, blood pressure, blood oxygen saturation, etc.) and nursing intervention measures, and combining statistical analysis methods, an impact model of nursing intervention on the probability of complication occurrence is established. Specific methods can adopt statistical means such as regression analysis to conduct a loss analysis of the intervention efficacy data. Through the time series analysis of multi-dimensional physiological sign data, the reduction effect of the intervention measures on the complication incidence rate within a specific time period is calculated, and the reduction data of the complication occurrence probability is obtained. This data reflects the actual effect of nursing intervention on reducing the risk of complications and lays a foundation for the subsequent analysis of nursing effectiveness. By analyzing the positive correlation between the symptom care intervention efficacy data and the reduction data of the complication occurrence probability, the effectiveness of the nursing intervention measures is determined. Specific operations include adopting correlation analysis methods (such as Pearson correlation coefficient, Spearman rank correlation analysis, etc.) to evaluate the linear or non-linear relationship between the nursing intervention efficacy and the reduction of complication probability. During the analysis process, the multi-dimensional data is normalized to eliminate biases. By drawing a positive correlation trend chart, it is clear which nursing intervention measures have the best preventive effect on complications and which measures show higher efficacy under different physiological states of patients, thereby obtaining the positive correlation data of nursing effectiveness. This data provides a basis for judging the actual effectiveness of nursing measures and can guide the optimization and adjustment of subsequent nursing plans. Through abnormal decision deduction, the nursing anomalies that occur are identified and processed. Based on the positive correlation data of nursing effectiveness obtained in the previous step and combined with the reduction data of the complication occurrence probability, a deduction analysis is carried out. Specific operations include adopting abnormal detection algorithms (such as Z-score method or box plot outlier detection method) to identify the points in the data that represent nursing anomalies or non-compliant decisions. During the deduction process, by simulating nursing interventions in different scenarios, the patient responses under abnormal decision conditions and their impacts on the complication risk are predicted. Through the statistical evaluation of the deduction results, potential abnormal nursing decisions are identified and their impacts are recorded, generating the symptom care abnormal decision deduction data. This data can be used to prevent potential mistakes in nursing decisions and ensure more accurate decisions in the ICU nursing process.

[0054] Preferably, step S33 includes the following steps: Step S331: Perform dynamic risk simulation of complications on the reduction data of the complication occurrence probability according to the symptom care intervention efficacy data and the positive correlation data of nursing effectiveness, and obtain dynamic risk simulation data of complications; Step S332: Identify abnormal triggering factors for the dynamic risk simulation data of complications to obtain risk abnormal triggering factors; Step S333: Perform abnormal decision deduction on the reduction data of the complication occurrence probability according to the risk abnormal triggering factors to obtain symptom care abnormal decision deduction data.

[0055] In the embodiments of the present invention, the simulation analysis of the dynamic risk of complications is carried out by using the symptom care intervention efficacy data and the data positively correlated with nursing effectiveness. In the specific implementation process, first, the main indicators of the occurrence of complications need to be defined, such as physiological parameters such as heart rate, blood pressure, and respiratory rate. Then, a dynamic simulation method based on time series (such as the ARIMA model or the state space model) is used to establish a dynamic risk model that can reflect the change of the complication risk. By inputting the symptom care intervention efficacy and the data positively correlated with nursing effectiveness, the change trend of the probability of complication occurrence under different nursing measures is simulated, so as to obtain the dynamic risk simulation data of complications. This data provides a basis for subsequent abnormal decision deduction and can clearly show the impact of nursing intervention on the complication risk. Identification of potential abnormal trigger factors in the dynamic risk simulation data of complications. First, the simulation data is preprocessed by using statistical analysis methods (such as principal component analysis PCA or clustering analysis) to identify the abnormal patterns in the data. Next, an anomaly detection algorithm (such as the isolation forest or the threshold-based method) is applied to analyze the dynamic risk of complications at each time point to determine which factors cause the abnormal change of the risk. The identification process needs to combine clinical knowledge and pay attention to the variables related to the patient's condition and nursing measures, so as to effectively identify the risk abnormal trigger factors. The identification of these factors will provide a key basis for subsequent abnormal decision deduction and help optimize the nursing intervention strategy. Based on the identified risk abnormal trigger factors, the deduction analysis of abnormal decisions is carried out on the data of the reduction of the probability of complication occurrence. First, an abnormal decision scenario is set, and the corresponding decision changes are formulated according to the identified trigger factors. For example, for a known risk factor (such as a certain specific physiological characteristic change of the patient), the change of the complication risk caused by the nursing decision under this condition is deduced. The simulated annealing algorithm or the game theory method is used for decision deduction to simulate how the probability of complication occurrence of the patient will change under different decision choices. Finally, the symptom care abnormal decision deduction data is generated, which can provide decision support for clinical nurses for specific abnormal situations, help avoid the resulting complication risk, and improve the quality and safety of nursing.

[0056] Preferably, step S4 includes the following steps: Step S41: Perform logistic regression analysis on the symptom care abnormal decision deduction data to obtain abnormal decision logistic regression data; Step S42: Perform sensitivity analysis of abnormal nursing decisions according to the abnormal decision logistic regression data to obtain sensitivity data of abnormal nursing decisions; Step S43: Based on the random forest algorithm, construct an ICU nursing decision support plan for the sensitivity data of abnormal nursing decisions to obtain an ICU nursing decision support plan.

[0057] In the embodiments of the present invention, first, using the symptom care abnormal decision deduction data, a logistic regression analysis is performed on the relationship between different nursing decision variables and the risk of complications. The logistic regression method is applicable to binary classification problems, especially for analyzing the influence of decision variables on binary outcomes (such as whether complications occur). In the specific implementation process, key nursing decision variables need to be selected first. These variables usually come from the specific operation steps and patient status change data in clinical nursing records. Then, by constructing a logistic regression equation, the contribution degree of each variable to the risk of complications is evaluated, and the regression coefficient of the variable is obtained, which represents the probability of causing an abnormality under specific circumstances for each nursing measure. The final output is the abnormal decision logistic regression data, which can show the strength and direction relationship between each nursing decision variable and the abnormal result, providing a basis for further decision-making support. The sensitivity analysis is carried out using the abnormal decision logistic regression data to identify which nursing decision variables are the most sensitive to abnormal situations. The purpose of the sensitivity analysis is to evaluate the impact of small changes in nursing decision variables on the probability of complications occurring under different nursing scenarios. Through gradient analysis or differential analysis methods, a perturbation test is performed on each nursing variable one by one to observe the degree of its influence on the abnormal nursing result. The specific process includes respectively performing an up or down adjustment on each variable, calculating the degree of risk change, and finally determining which variables are the most important influencing factors in nursing decisions. The results of the sensitivity analysis will generate abnormal nursing decision sensitivity data, which reflects the sensitivity degree of each nursing decision variable, facilitating nursing staff to prioritize and adjust these key variables to reduce the occurrence of abnormal events. The random forest algorithm is used to analyze the abnormal nursing decision sensitivity data, and based on this, an ICU nursing decision support plan is constructed. The random forest algorithm is an ensemble learning algorithm. By constructing a large number of decision tree models, a comprehensive evaluation of different nursing variables is carried out to form a more robust decision support plan. In the specific implementation, first, the abnormal nursing decision sensitivity data is input into the random forest model. The model analyzes the influence of combinations of different nursing decision variables on the risk of complications through multiple samplings and the construction of tree structures. Through iterative calculations and cross-validation, the optimal decision combination plan is determined. Finally, the ICU nursing decision support plan is output, which includes a series of nursing intervention measures for abnormal situations, can provide clear operation suggestions for ICU nursing staff, help optimize the nursing process, and improve the safety and treatment effect of patients.

[0058] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the application documents are intended to be included in the present invention.

[0059] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing an ICU nursing decision support system based on case reasoning, characterized in that: The following steps are involved: Step S1: Acquire ICU nursing history case data; perform structured time-series vital sign change extraction based on the ICU nursing history case data to obtain structured time-series vital sign change data; Step S2: performing tolerance impairment interpolation processing on the structured time series physical sign change data to obtain symptom treatment tolerance interpolation data; performing nursing intervention effectiveness analysis on the symptom treatment tolerance interpolation data to obtain symptom nursing intervention effectiveness data; performing multidimensional dynamic evaluation of physiological sign indicators based on the symptom nursing intervention effectiveness data to obtain multidimensional dynamic data of physiological sign indicators; Step S3: performing complication probability reduction analysis on the multidimensional dynamic data of physiological sign indicators to obtain complication probability reduction data; performing abnormal decision deduction on the complication probability reduction data to obtain symptom nursing abnormal decision deduction data; Step S4: Based on the random forest algorithm, an ICU nursing decision support plan is constructed for the symptom nursing abnormal decision deduction data to obtain an ICU nursing decision support plan.

2. The method for constructing an ICU nursing decision support system based on case reasoning according to claim 1 is characterized in that: Step S1 includes the following steps: Step S11: Acquire ICU nursing history case data; Step S12: classify the subject of the ICU nursing history case data to obtain the subject data of the ICU nursing case; Step S13: Filling the missing values ​​of nursing content in the subject data of the ICU nursing case to obtain the nursing content filled data of the subject; Step S14: Fill in the data according to the nursing content of the subject to perform structured time-series vital sign change extraction to obtain structured time-series vital sign change data.

3. The method for constructing an ICU nursing decision support system based on case reasoning according to claim 1 is characterized in that: Step S2 includes the following steps: Step S21: performing symptom treatment trend analysis on the structured time-series physical sign change data to obtain symptom treatment trend data; Step S22: performing tolerance impairment interpolation processing on the symptom treatment trend data to obtain symptom treatment tolerance interpolation data; Step S23: performing nursing intervention effectiveness analysis on symptom treatment tolerance interpolation data based on case reasoning and ICU nursing history case data to obtain symptom nursing intervention effectiveness data; Step S24: Perform multidimensional dynamic evaluation of physiological sign indicators on the structured time-series physical sign change data according to the symptom nursing intervention effectiveness data to obtain multidimensional dynamic data of physiological sign indicators.

4. The method for constructing an ICU nursing decision support system based on case reasoning according to claim 3 is characterized in that: Step S22 includes the following steps: Step S221: performing clinical response improvement trend analysis on the symptom treatment trend data to obtain clinical response improvement trend data; Step S222: performing temporal discontinuity difference analysis on the clinical response improvement trend data to obtain response improvement temporal discontinuity difference data; Step S223: Calculate the median of the reaction improvement time series discontinuity difference data between adjacent time series to obtain the median of the improvement difference between adjacent time series; Step S224: performing linear structure optimization on the clinical response improvement trend data according to the median of the improvement difference between adjacent time series to obtain the response improvement trend linear data; Step S225: performing global variance nearest neighbor interpolation on the linear data reflecting the improvement trend to obtain global interpolation data of the improvement trend; Step S226: Perform tolerance reduction interpolation processing on the symptom treatment trend data according to the improvement trend global interpolation data to obtain symptom treatment tolerance interpolation data.

5. The method for constructing an ICU nursing decision support system based on case reasoning according to claim 4 is characterized in that: The global variance nearest neighbor interpolation for linear data reflecting the improvement trend includes the following steps: Using a preset improvement value assignment model, the linear data reflecting the improvement trend are evaluated for the adjacent trend improvement value, thereby obtaining adjacent trend improvement value data; Perform global numerical variance calculation on the neighboring trend improvement numerical data to obtain the global trend numerical variance; According to the global trend numerical variance, the neighboring trend improvement numerical data is subjected to global variance nearest neighbor interpolation processing to obtain the improved trend global interpolation data.

6. The method for constructing an ICU nursing decision support system based on case reasoning according to claim 3 is characterized in that: Step S23 includes the following steps: Step S231: performing a matching simulation of the intervention effect in the nursing process on the symptom treatment tolerance interpolation data based on case reasoning to obtain symptom nursing intervention effect simulation data; Step S232: clustering the symptom nursing intervention effect simulation data based on case reasoning and ICU nursing history case data to obtain similar case effect difference clustering data; Step S233: performing effect clustering stability analysis on the effect difference clustering data of similar cases to obtain effect clustering stability data; Step S234: Perform nursing intervention effectiveness analysis based on the effect clustering stability data to obtain symptom nursing intervention effectiveness data.

7. The method for constructing an ICU nursing decision support system based on case reasoning according to claim 6 is characterized in that: The effect clustering stability analysis of similar case effect difference clustering data includes the following steps: Conduct intra-cluster dispersion analysis on the effect difference clustering data of similar cases to obtain the intra-cluster discrete data of effect clustering; According to the discrete data within the effect clustering cluster, the conflict effect analysis is performed on the clustering data of the effect difference of similar cases to obtain the clustering effect conflict effect data; According to the clustering effect conflict effect data, the conflict commonality of the clustering data of similar case effects is identified to obtain the case effect conflict commonality data; Conduct case effect life cycle analysis on the common data of case effect conflicts to obtain case effect conflict life cycle data; Based on the case effect conflict life cycle data, the effect clustering stability analysis is performed on the effect difference clustering data of similar cases to obtain the effect clustering stability data.

8. The method for constructing an ICU nursing decision support system based on case reasoning according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: performing complication probability reduction analysis on the multidimensional dynamic data of physiological sign indicators according to the symptom nursing intervention effectiveness data to obtain complication probability reduction data; Step S32: performing a nursing effectiveness positive correlation analysis on the complication probability reduction data based on the symptom nursing intervention efficacy data to obtain nursing effectiveness positive correlation data; Step S33: Perform abnormal decision deduction on the complication probability reduction data according to the symptom nursing intervention efficacy data and the nursing effectiveness positive correlation data to obtain symptom nursing abnormal decision deduction data.

9. The method for constructing an ICU nursing decision support system based on case reasoning according to claim 8, characterized in that: Step S33 includes the following steps: Step S331: performing complication dynamic risk simulation on complication occurrence probability reduction data according to symptom nursing intervention efficacy data and nursing effectiveness positive correlation data to obtain complication dynamic risk simulation data; Step S332: identifying abnormal trigger factors of complication dynamic risk simulation data to obtain abnormal risk trigger factors; Step S333: Perform abnormal decision deduction on the complication probability reduction data according to the risk abnormality triggering factor to obtain symptom care abnormal decision deduction data.

10. The method for constructing an ICU nursing decision support system based on case reasoning according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: performing a logistic regression analysis on the symptom nursing abnormal decision deduction data to obtain abnormal decision logistic regression data; Step S42: performing abnormal nursing decision sensitivity analysis based on abnormal decision logistic regression data to obtain abnormal nursing decision sensitivity data; Step S43: construct an ICU nursing decision support plan for abnormal nursing decision sensitivity data based on the random forest algorithm to obtain an ICU nursing decision support plan.

Images