A nursing decision-making support system based on clinical knowledge graph

Abstract

The present invention discloses a nursing decision-making support system based on clinical knowledge graph, which specifically relates to the technical field of nursing decision-making support. The system monitors the health of patients, obtains health-related data of patients, extracts the physiological index data of patients from the health-related data of patients; analyzes the physiological index data of patients, quantifies the health degree of patients based on the physiological index data, and outputs the health quality evaluation index of patients; obtains the actual patient health trend coefficient and the patient health trend prediction coefficient based on the health trend analysis module; jointly analyzes the patient health trend prediction coefficient and the actual patient health trend coefficient to obtain the patient trend abnormality index; judges the relationship between the patient trend abnormality index and the preset value to generate early warning information, so as to solve the problem that the existing nursing decision-making support system lacks in-depth analysis of the patient health trend and cannot provide sufficient forward-looking guidance.

Description

Technical Field

[0001] The present invention relates to the technical field of nursing auxiliary decision-making, and more specifically, to a nursing auxiliary decision-making system based on clinical knowledge graph. Background Art

[0002] Clinical nursing decision-making refers to the complex process of professional decisions made by nurses about patient care during clinical nursing practice. It can be for individual patients or for groups of patients. In the current medical care environment, nurses are faced with growing nursing needs and complex and changing clinical conditions. They need to process a large amount of patient information in a short period of time, including but not limited to medical images, diagnostic reports, instrument measurement results, etc., in order to make accurate and timely nursing decisions. However, the traditional nursing decision-making process often relies on the personal experience and intuition of nurses, which may not only lead to subjectivity and uncertainty in decision-making, but also miss critical treatment opportunities due to insufficient information processing.

[0003] A patient's health status is a dynamic process, and their physiological indicators fluctuate over time. Therefore, accurately predicting a patient's health trends is crucial to developing an effective care plan.

[0004] However, in actual use, it still has many shortcomings. For example, the existing nursing decision support system often lacks the ability to conduct in-depth analysis and prediction of patients' health trends and cannot provide sufficient forward-looking guidance. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a nursing decision support system based on a clinical knowledge graph to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: a nursing decision support system based on clinical knowledge graph, comprising:

[0007] The patient health monitoring module is used to monitor the patient's health and obtain the patient's physiological index data; obtain the patient's health-related data, which includes but is not limited to medical imaging data, diagnostic data and instrument measurement data; extract the patient's physiological index data from the patient's health-related data;

[0008] The patient health status assessment module is used to analyze the patient's physiological index data, quantify the patient's health status based on the physiological index data, and output the patient's health quality assessment index JP;

[0009] The health trend analysis module is used to obtain the actual patient health trend coefficient Qs and the patient health trend prediction coefficient Qy; input the updated actual physiological index data into the health quality assessment model, output the actual patient health quality assessment index JP_b, and record the initial value of the patient health quality assessment index as JP_a. Calculate the actual patient health trend coefficient Qs, T J Represents the time interval; based on the machine learning model, the predicted physiological indicators are output, the predicted physiological indicators are input into the health quality assessment model, and the predicted patient health quality assessment index JP_yb is output. The patient's health trend prediction coefficient Qy is calculated;

[0010] The trend anomaly analysis module jointly analyzes the patient health trend prediction coefficient Qy and the actual patient health trend coefficient Qs to obtain the patient trend anomaly index;

[0011] The abnormal trend judgment module is used to judge the relationship between the abnormal trend index of the patient and the preset value. If the abnormal trend index of the patient exceeds the preset value, it indicates that the patient's nursing process is abnormal, and an early warning is issued and an early warning message is generated. If the abnormal trend index of the patient does not exceed the preset value, it indicates that the patient's nursing process is normal and no measures are taken.

[0012] Preferably, the process of analyzing the patient's physiological index data from medical imaging data, diagnostic data and instrument measurement data based on the clinical knowledge graph includes the following steps:

[0013] Step S11, data preprocessing: cleaning the patient's health-related data and unifying the data format to facilitate subsequent analysis and processing;

[0014] Step S12, data annotation and mapping: using the clinical knowledge graph, annotate the key anatomical structures and pathological features in the medical imaging data, and annotate the key physiological indicators in the diagnostic data and instrument measurement data, and map the annotated key physiological indicators to the corresponding nodes and relationships in the clinical knowledge graph to establish associations between the data;

[0015] Step S13, feature extraction: used to extract and output patient health features from patient health-related data, wherein the patient health features include image features, diagnostic features, and measurement data features;

[0016] Step S14, fusion analysis: fusing the extracted image features, diagnostic features and measurement data features, and outputting the fused data;

[0017] Step S15, reasoning based on knowledge graph: using the rules and relationships in the clinical knowledge graph, reasoning and analyzing the fused data to obtain the patient's physiological index data;

[0018] Step S16, result verification and correction: Compare and verify the physiological index data obtained by the reasoning analysis with the clinical standards or expert opinions to ensure the reliability of the reasoning analysis.

[0019] Preferably, the patient health characteristics are integrated, including:

[0020] The patient's health characteristics are divided into complementary feature data and redundant feature data. Weights are assigned according to the reliability and accuracy of data from different sources. The weighted average method is used to complete feature fusion and eliminate redundant feature data. The complementary feature data are merged to output the patient's physiological indicator data.

[0021] Preferably, the process of outputting predicted physiological indicators based on the physiological indicator prediction model includes the following steps:

[0022] Step S21, data collection and preprocessing: collect the patient's historical health data, nursing plan and past medical history; clean and preprocess the data, remove outliers and missing values, and perform standardization or normalization;

[0023] Step S22, feature selection and extraction: extract features from the original data according to the prediction target, and determine the key features that have a significant impact on the prediction results through statistical methods, expert experience or machine learning algorithms;

[0024] Step S23, model selection and training: building a physiological index prediction model based on a machine learning algorithm, using the preprocessed data to train the physiological index prediction model, and adjusting the model parameters to minimize the prediction error;

[0025] Step S24, model evaluation and optimization: evaluate the performance of the physiological index prediction model through cross-validation and hold-out method, optimize the physiological index prediction model according to the evaluation results, and output the trained physiological index prediction model;

[0026] Step S25: Use the trained physiological index prediction model to output predicted physiological indexes.

[0027] Preferably, the patient health quality assessment index JP is obtained in the following manner:

[0028] Get the patient's physiological index data. Suppose there are n physiological index data. Use i, j to represent the order of the physiological index data, and i≠j. Use yz i Indicates the value of the ith physiological index deviating from the preset standard, yz j Indicates the value of the jth physiological index deviating from the preset standard; g i Represents the weight coefficient of the ith physiological index. The patient health quality assessment index JP is calculated by the following health quality assessment model: Among them, h ij represents the weight coefficient of the interaction between the ith and jth physiological indicators, and ε represents the basic value of the patient's health quality.

[0029] Preferably, the patient abnormal trend index YQ is obtained in the following manner:

[0030] According to the timestamp, the actual patient health trend coefficient Qs and the patient health trend prediction coefficient Qy are time-matched by interpolation or mean / median filling. Based on the actual patient health trend coefficient Qs and the patient health trend prediction coefficient Qy after time matching, the formula The patient's abnormal trend index YQ was calculated.

[0031] Preferably, the nursing decision support system further includes an abnormality analysis module.

[0032] In the abnormality analysis module, an abnormality analysis instruction is generated based on the early warning information, wherein the early warning information includes the patient identity information, the nursing environment information and the nursing operation information, and the nursing environment abnormality index and the nursing operation abnormality index are output;

[0033] When the nursing environment abnormality index and nursing operation abnormality index meet the corresponding preset values, it indicates that the nursing decision matching is abnormal;

[0034] When the nursing environment abnormality index and nursing operation abnormality index do not meet the corresponding preset values, an early warning is issued to indicate that the nursing environment and nursing operation are abnormal.

[0035] Preferably, the nursing environment abnormality index HY is obtained in the following manner:

[0036] Obtain the deviation Dev of the environmental parameters. Suppose there are Q environmental parameters, p represents the sequential number of the environmental parameters, the deviation of the pth environmental parameter is recorded as Dev_p, and the weight coefficient of the pth environmental parameter is recorded as Wd_p; the actual measured value of the nursing environmental parameter is recorded as M, and the upper and lower limits of the appropriate range of the nursing environmental parameter are recorded as L and U respectively; through the formula Deviation Dev is calculated; mid(L, U) represents the midpoint of the appropriate range of nursing environment parameters, and range(L, U) represents the size of the appropriate range of nursing environment parameters; The abnormal nursing environment index EAI was calculated.

[0037] Preferably, the nursing operation abnormality index is obtained in the following manner:

[0038] Suppose there are K nursing operations in the nursing plan, the standard score of each operation is S_s, and the actual score is A _ s, through the formula The abnormal nursing operation index EHI was calculated.

[0039] Preferably, the nursing decision support system further includes an abnormality analysis module.

[0040] The nursing plan management module is used to obtain nursing alternative plans and screen out recommended nursing plans from the nursing alternative plans based on the plan recommendation index.

[0041] Technical effects and advantages of the present invention:

[0042] The nursing decision-making support system provided by the present invention obtains a patient health trend prediction coefficient Qy and an actual patient health trend coefficient Qs, jointly analyzes the patient health trend prediction coefficient Qy and the actual patient health trend coefficient Qs, and obtains a patient trend abnormality index; determines the relationship between the patient trend abnormality index and a preset value, and if the patient trend abnormality index exceeds the preset value, it indicates that the patient's nursing process is abnormal, and an early warning is issued to the outside, and early warning information is generated, and the early warning information includes patient identity information, nursing environment information and nursing operation information; if the patient trend abnormality index does not exceed the preset value, it indicates that the patient's nursing process is normal, and no measures are taken, thereby solving the problem of lack of in-depth analysis of patient health trends and inability to provide sufficient forward-looking guidance in existing nursing decision-making support systems.

[0043] The nursing decision support system provided by the present invention generates abnormal analysis instructions based on early warning information, further analyzes abnormal situations in the nursing environment and nursing operations, and can identify potential problems more deeply; calculates the nursing environment abnormality index and the nursing operation abnormality index, which helps to locate the problem more accurately; when the nursing environment abnormality index or the nursing operation abnormality index meets the corresponding preset value, it prompts that the nursing decision matching is abnormal, which helps medical staff to adjust the nursing plan in time; when these indices do not meet the preset values, it points out whether there is an abnormality in the nursing environment or the nursing operation, which improves the pertinence and accuracy of the early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a structural block diagram of the nursing decision support system based on clinical knowledge graph of the present invention.

[0045] Figure 2 This is a flowchart of the nursing auxiliary decision-making system of the present invention. DETAILED DESCRIPTION

[0046] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0047] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the present application, its application, or uses.

[0048] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered as part of the specification.

[0049] Example 1, see Figure 1 The present invention provides a structural block diagram of a nursing decision support system based on a clinical knowledge graph. Figure 1 A nursing decision support system based on clinical knowledge graph is shown, comprising:

[0050] Patient health monitoring module, used to monitor patient health and obtain patient physiological index data;

[0051] The patient health monitoring module includes a patient information collection unit and a patient information analysis unit. The patient information collection unit is used to obtain patient health-related data, which includes but is not limited to medical imaging data, diagnostic data and instrument measurement data; the patient information analysis unit is used to extract the patient's physiological index data from the patient's health-related data;

[0052] The patient health status assessment module is used to analyze the patient's physiological index data, quantify the patient's health status based on the physiological index data, and output the patient's health quality assessment index JP;

[0053] A health trend analysis module is used to obtain the actual patient health trend coefficient Qs and the patient health trend prediction coefficient Qy;

[0054] The health trend analysis module includes an actual health trend analysis unit and a health trend prediction analysis unit. The actual health trend analysis unit inputs the updated actual physiological index data into the health quality assessment model, outputs the actual patient health quality assessment index JP_b, and records the initial value of the patient health quality assessment index as JP_a. Calculate the actual patient health trend coefficient Qs, T J Indicates a time interval;

[0055] The health trend prediction and analysis unit outputs predicted physiological indicators based on the machine learning model, inputs the predicted physiological indicators into the health quality assessment model, and outputs a predicted patient health quality assessment index JP_yb. The patient's health trend prediction coefficient Qy is calculated;

[0056] The trend anomaly analysis module jointly analyzes the patient health trend prediction coefficient Qy and the actual patient health trend coefficient Qs to obtain the patient trend anomaly index;

[0057] The abnormal trend judgment module is used to judge the relationship between the abnormal trend index of the patient and the preset value. If the abnormal trend index of the patient exceeds the preset value, it indicates that the patient's nursing process is abnormal, and an early warning is issued and an early warning message is generated. If the abnormal trend index of the patient does not exceed the preset value, it indicates that the patient's nursing process is normal and no measures are taken.

[0058] It needs to be further explained in the embodiment of the present invention that the process of analyzing the patient's physiological index data from the medical imaging data, the diagnostic data and the instrument measurement data based on the clinical knowledge graph includes the following steps:

[0059] Step S11, data preprocessing: cleaning patient health-related data (including medical imaging data, diagnostic data and instrument measurement data) to remove duplicate, erroneous or irrelevant information; unifying the data format to facilitate subsequent analysis and processing;

[0060] Step S12, data annotation and mapping: using the clinical knowledge graph, annotate the key anatomical structures and pathological features in the medical imaging data, and annotate the key physiological indicators in the diagnostic data and instrument measurement data, and map the annotated key physiological indicators to the corresponding nodes and relationships in the clinical knowledge graph to establish associations between the data;

[0061] Step S13, feature extraction: used to extract and output patient health features from patient health-related data, wherein the patient health features include image features, diagnostic features, and measurement data features;

[0062] The image feature extraction method is: using image processing technology to extract image features related to physiological indicators from medical image data, such as morphological features and texture features;

[0063] The diagnostic feature extraction method is: performing natural language processing on the diagnostic data, extracting keywords, phrases or sentences related to the physiological indicators, and outputting diagnostic features;

[0064] The measurement data feature extraction method is: directly extracting physiological index data from the instrument measurement data and outputting the measurement data features, such as heart rate, blood pressure, and blood sugar;

[0065] Step S14, fusion analysis: fusing the extracted image features, diagnostic features and measurement data features, and outputting the fused data;

[0066] Step S15, reasoning based on knowledge graph: using the rules and relationships in the clinical knowledge graph, reasoning and analyzing the fused data to obtain the patient's physiological index data;

[0067] Step S16, result verification and correction: Compare and verify the physiological index data obtained by the reasoning analysis with the clinical standards or expert opinions to ensure the reliability of the reasoning analysis.

[0068] It is necessary to further explain in the embodiments of the present invention that, in the fusion process, the complementarity and redundancy between different data sources are taken into consideration, and appropriate data fusion techniques such as weighted averaging, Bayesian networks, and deep learning are used to fuse the patient's health characteristics, including:

[0069] Divide the patient's health characteristics into complementary feature data and redundant feature data, and transform the redundant feature data into complementary feature data; explain that complementary feature data means that the features of different data sources each provide unique and non-overlapping information, which is crucial for a comprehensive understanding of the patient's physiological state. For example, imaging features provide detailed information about organ morphology and structure, while diagnostic features contain the doctor's detailed description of the patient's condition, and measurement data directly reflects the patient's physiological indicators; redundant feature data means that the features of different data sources provide the same or similar information. For example, two different instruments may measure the same heart rate data, or both medical imaging data and diagnostic data contain descriptions of a certain pathological feature.

[0070] What needs to be further explained in the embodiments of the present invention is that the method of converting redundant feature data into complementary feature data is: assigning weights according to the reliability and accuracy of data from different sources, using weighted average method to complete feature fusion, and eliminating redundant feature data. For example, if the measurement accuracy of an instrument is higher, give it a greater weight, or consider using more complex fusion methods, such as Kalman filtering and Bayesian fusion, to process the differences between redundant data; merge complementary feature data and output the patient's physiological indicator data.

[0071] It needs to be further explained in the embodiment of the present invention that the process of outputting predicted physiological indicators based on the physiological indicator prediction model includes the following steps:

[0072] Step S21, data collection and preprocessing: collect the patient's historical health data, including but not limited to physiological indicators (such as heart rate, blood pressure, blood sugar, etc.), nursing plans, and past medical history; clean and preprocess the data, remove outliers and missing values, and perform standardization or normalization to ensure the accuracy and consistency of the data;

[0073] Step S22, feature selection and extraction: extract features from the original data according to the prediction target (such as the trend of physiological indicators), and determine the key features that have a significant impact on the prediction results through statistical methods, expert experience or machine learning algorithms (such as feature selection algorithms);

[0074] Step S23, model selection and training: building a physiological index prediction model based on a machine learning algorithm, such as time series analysis (such as ARIMA, LSTM, etc.), a neural network (such as RNN, CNN, etc.) or other supervised learning algorithms, using the preprocessed data to train the physiological index prediction model, and adjusting the model parameters to minimize the prediction error;

[0075] Step S24, model evaluation and optimization: evaluate the performance of the physiological index prediction model through cross-validation and hold-out method, optimize the physiological index prediction model according to the evaluation results, such as adjusting the model structure and increasing the number of features; output the trained physiological index prediction model;

[0076] Step S25: Use the trained physiological index prediction model to output predicted physiological indexes.

[0077] Explanation: The patient health trend prediction coefficient Qy reflects the expected changes in the patient's health trend in the future.

[0078] It needs to be further explained in the embodiment of the present invention that the patient health quality assessment index JP is obtained in the following manner:

[0079] Get the patient's physiological index data. Suppose there are n physiological index data. Use i, j to represent the order of the physiological index data, and i≠j. Use yz i Indicates the value of the ith physiological index deviating from the preset standard, yz j Indicates the value of the jth physiological index deviating from the preset standard; g i Represents the weight coefficient of the ith physiological index. The patient health quality assessment index JP is calculated by the following health quality assessment model: Among them, h ij represents the weight coefficient of the interaction between the ith and jth physiological indicators, and ε represents the basic value of the patient's health quality.

[0080] It needs to be further explained in the embodiment of the present invention that the patient abnormal trend index YQ is obtained in the following manner:

[0081] According to the timestamp, the actual patient health trend coefficient Qs and the patient health trend prediction coefficient Qy are time-matched by interpolation method (such as linear interpolation, nearest neighbor interpolation) or mean / median filling. Based on the actual patient health trend coefficient Qs and the patient health trend prediction coefficient Qy after time matching, the formula The patient's abnormal trend index YQ was calculated.

[0082] Summary: The embodiment of the present invention integrates patient health monitoring, health status assessment, health trend analysis, and trend abnormality analysis and judgment modules to form a comprehensive and integrated nursing decision-making support system, which can monitor and analyze the patient's health status from multiple angles and levels, thereby improving the accuracy and comprehensiveness of decision-making; real-time monitoring of the patient's physiological indicator data, assessment of the patient's health status based on the physiological indicator data, and output of the health quality assessment index, thereby achieving real-time understanding and quantitative assessment of the patient's health status; predicting future physiological indicators through machine learning models to provide forward-looking guidance for nursing decisions; the trend abnormality analysis module and the trend abnormality judgment module work together. Once the abnormal index exceeds the preset value, the system immediately issues an early warning and generates early warning information. This timely early warning mechanism helps medical staff to take measures quickly to prevent the disease from worsening.

[0083] Example 2, see Figure 2 The nursing decision-making support system flow chart of the present invention is different from the embodiment 1 in that the nursing decision-making support system further includes an abnormality analysis module.

[0084] In the abnormality analysis module, an abnormality analysis instruction is generated based on the early warning information, wherein the early warning information includes the patient identity information, the nursing environment information and the nursing operation information, and the nursing environment abnormality index and the nursing operation abnormality index are output;

[0085] When the nursing environment abnormality index and nursing operation abnormality index meet the corresponding preset values, it indicates that the nursing decision matching is abnormal;

[0086] When the nursing environment abnormality index and nursing operation abnormality index do not meet the corresponding preset values, an early warning is issued to indicate that the nursing environment and nursing operation are abnormal.

[0087] It needs to be further explained in the embodiment of the present invention that the nursing environment abnormality index HY is obtained in the following manner:

[0088] Obtain the deviation Dev of the environmental parameters. Suppose there are Q environmental parameters, p represents the sequential number of the environmental parameters, the deviation of the pth environmental parameter is recorded as Dev_p, and the weight coefficient of the pth environmental parameter is recorded as Wd_p; the actual measured value of the nursing environmental parameter is recorded as M, and the upper and lower limits of the appropriate range of the nursing environmental parameter are recorded as L and U respectively; through the formula Deviation Dev is calculated; mid(L, U) represents the midpoint of the appropriate range of nursing environment parameters, and range(L, U) represents the size of the appropriate range of nursing environment parameters; The abnormal nursing environment index EAI was calculated.

[0089] It needs to be further explained in the embodiment of the present invention that the nursing operation abnormality index is obtained in the following manner:

[0090] Suppose there are K nursing operations in the nursing plan, the standard score of each operation is S_s, and the actual score is A _ s, through the formula The abnormal nursing operation index EHI was calculated.

[0091] Summary: Explanation: The main reasons for abnormal health trends of patients are: improper matching of nursing environment, nursing quality and nursing plan with patients; based on abnormal analysis instructions, the nursing environment abnormality index, nursing operation abnormality index and nursing decision matching abnormality coefficient are obtained, and targeted measures are taken based on the analysis results; the embodiment of the present invention introduces an abnormal analysis module to generate abnormal analysis instructions based on early warning information, and further analyzes abnormal conditions in the nursing environment and nursing operations, which enhances the diagnostic ability of the system and enables it to identify potential problems more deeply; calculates the nursing environment abnormality index (HY) and the nursing operation abnormality index (EHI), which reflect the abnormal conditions in the patient care process from the two perspectives of environment and operation, respectively, and help to locate the problem more accurately; when the nursing environment abnormality index or the nursing operation abnormality index meets the corresponding preset value, it prompts that the nursing decision matching is abnormal, which helps medical staff to adjust the nursing plan in time; when these indexes do not meet the preset values, it is pointed out whether there is an abnormality in the nursing environment or the nursing operation, which improves the pertinence and accuracy of the early warning.

[0092] Embodiment 3: The difference between the embodiment of the present invention and the embodiment 1 and embodiment 2 is that the nursing decision support system further includes an abnormality analysis module.

[0093] A nursing plan management module is used to obtain nursing alternative plans and select a recommended nursing plan from the nursing alternative plans based on the plan recommendation index. The nursing alternative plan refers to a collection of nursing operations in chronological order and under rule constraints.

[0094] The process of obtaining a recommended care plan includes the following steps:

[0095] Step 101: Based on the knowledge graph and the patient's physiological characteristics, obtain patient cases similar to the patient from the database as several nursing alternatives;

[0096] Explanation: Based on the patient's physiological characteristics (such as age, gender, weight, medical history, current symptoms, etc.), the patient's feature vector is constructed using the relevant information in the knowledge graph; the patient case data in the database is used to calculate the similarity between each case and the patient's feature vector, and this is achieved through algorithms such as cosine similarity and Euclidean distance to find the patient case that is most similar to the patient; similar cases are screened: based on the similarity calculation results, patient cases with a higher similarity to the patient are screened to form a similar population data set.

[0097] Step 102: Evaluate the applicability and cost of the alternative nursing plan, and output the recommendation index of the alternative nursing plan based on the plan recommendation index evaluation model, including:

[0098] 201. Obtain patient cases similar to the patient from the database to form a similar population data set;

[0099] 202. Analyze similar population data sets and output the suitability index of nursing alternatives;

[0100] 203. Analyze the cost impact index of nursing alternatives, and obtain the cost impact index of each nursing alternative based on economic cost, time cost and labor cost;

[0101] 204. Combined analysis of the applicability index and the cost impact index, calculated the recommended index of the nursing alternatives, obtained the patient's cost sensitivity factor, and obtained the time cost sensitivity factor and the labor cost sensitivity factor;

[0102] It needs to be further explained in the embodiment of the present invention that the scheme recommendation index evaluation model is: Among them, FV represents the recommendation index of the nursing alternative plan. The larger the FV value, the more recommended the corresponding nursing alternative plan is; Cz represents the cost index of the nursing alternative plan, C1, C2, and C3 represent the economic cost, time cost, and labor cost of the nursing alternative plan respectively; α, β, and γ represent the sensitivity coefficients of economic cost, time cost, and labor cost respectively; SV represents the applicability index of the nursing alternative plan. Suppose there are K patient cases in the similar population data set, and s represents the sequential number of the patient case. The similarity weight between the sth patient case and the current patient is obtained. Su_s and W_s represent the success rate and similarity weight corresponding to the sth patient case respectively;

[0103] Step 103: Make a selection based on the recommendation index of the alternative nursing plans and output a recommended nursing plan.

[0104] Summary: Through the advantages of accurate recommendation of personalized nursing plans, scientific evaluation system, improving the accuracy and efficiency of nursing services, promoting the rational use of medical resources, and supporting scientific and intelligent medical decision-making, it provides strong support and guarantee for modern medical services and helps to improve patients' treatment effects and satisfaction.

[0105] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A nursing decision support system based on clinical knowledge graph, characterized in that: include: The patient health monitoring module is used to monitor the patient's health and obtain the patient's physiological index data; obtain the patient's health-related data, which includes but is not limited to medical imaging data, diagnostic data and instrument measurement data; extract the patient's physiological index data from the patient's health-related data; The process of analyzing the patient's physiological index data from medical imaging data, diagnostic data, and instrument measurement data based on the clinical knowledge graph includes the following steps: Step S11, data preprocessing: cleaning the patient's health-related data and unifying the data format to facilitate subsequent analysis and processing; Step S12, data annotation and mapping: using the clinical knowledge graph, annotate the key anatomical structures and pathological features in the medical imaging data, and annotate the key physiological indicators in the diagnostic data and instrument measurement data, and map the annotated key physiological indicators to the corresponding nodes and relationships in the clinical knowledge graph to establish associations between the data; Step S13, feature extraction: used to extract and output patient health features from patient health-related data, wherein the patient health features include image features, diagnostic features, and measurement data features; Step S14, fusion analysis: fusing the extracted image features, diagnostic features and measurement data features, and outputting the fused data; Step S15, reasoning based on knowledge graph: using the rules and relationships in the clinical knowledge graph, reasoning and analyzing the fused data to obtain the patient's physiological index data; Step S16, result verification and correction: Compare and verify the physiological index data obtained by the reasoning analysis with the clinical standards or expert opinions to ensure the reliability of the reasoning analysis; The patient health status assessment module is used to analyze the patient's physiological index data, quantify the patient's health status based on the physiological index data, and output the patient's health quality assessment index JP; Get the patient's physiological index data. Suppose there are n physiological index data. Use i, j to represent the order of the physiological index data, and i≠j. Use yz i Indicates the value of the ith physiological index deviating from the preset standard, yz j Indicates the value of the jth physiological index deviating from the preset standard; g i Represents the weight coefficient of the ith physiological index. The patient health quality assessment index JP is calculated by the following health quality assessment model: Among them, h ij represents the weight coefficient of the interaction between the ith and jth physiological indicators, and ε represents the basic value of the patient's health quality; The health trend analysis module is used to obtain the actual patient health trend coefficient Qs and the patient health trend prediction coefficient Qy; input the updated actual physiological index data into the health quality assessment model, output the actual patient health quality assessment index JP_b, and record the initial value of the patient health quality assessment index as JP_a. Calculate the actual patient health trend coefficient Qs, T J Represents the time interval; based on the machine learning model output predicted physiological indicators, the predicted physiological indicators are input into the health quality assessment model, and the predicted patient health quality assessment index JP_yb is output. The patient's health trend prediction coefficient Qy is calculated; The trend anomaly analysis module jointly analyzes the patient health trend prediction coefficient Qy and the actual patient health trend coefficient Qs to obtain the patient trend anomaly index; The abnormal trend judgment module is used to judge the relationship between the abnormal trend index of the patient and the preset value. If the abnormal trend index of the patient exceeds the preset value, it indicates that the patient's nursing process is abnormal, and an early warning is issued and an early warning message is generated. If the abnormal trend index of the patient does not exceed the preset value, it indicates that the patient's nursing process is normal and no measures are taken.

2. The nursing decision support system based on clinical knowledge graph according to claim 1 is characterized in that: Fusion of patient health characteristics, including: Divide the patient's health characteristics into complementary characteristic data and redundant characteristic data, assign weights according to the reliability and accuracy of data from different sources, use the weighted average method to complete feature fusion, eliminate redundant characteristic data; merge complementary characteristic data, and output the patient's physiological indicator data; Complementary feature data refers to the fact that features from different data sources each provide non-overlapping information; redundant feature data refers to the fact that features from different data sources provide the same or similar information.

3. The nursing decision support system based on clinical knowledge graph according to claim 1 is characterized in that: The process of outputting predicted physiological indicators based on the physiological indicator prediction model includes the following steps: Step S21, data collection and preprocessing: collect the patient's historical health data, nursing plan and past medical history; clean and preprocess the data, remove outliers and missing values, and perform standardization or normalization; Step S22, feature selection and extraction: extract features from the original data according to the prediction target, and determine the key features that have a significant impact on the prediction results through statistical methods, expert experience or machine learning algorithms; Step S23, model selection and training: building a physiological index prediction model based on a machine learning algorithm, using the preprocessed data to train the physiological index prediction model, and adjusting the model parameters to minimize the prediction error; Step S24, model evaluation and optimization: evaluate the performance of the physiological index prediction model through cross-validation and hold-out method, optimize the physiological index prediction model according to the evaluation results, and output the trained physiological index prediction model; Step S25: Use the trained physiological index prediction model to output predicted physiological indexes.

4. The nursing decision support system based on clinical knowledge graph according to claim 1 is characterized in that: The patient abnormal trend index YQ is obtained as follows: According to the timestamp, the actual patient health trend coefficient Qs and the patient health trend prediction coefficient Qy are time-matched by interpolation or mean / median filling. Based on the actual patient health trend coefficient Qs and the patient health trend prediction coefficient Qy after time matching, the formula The patient's abnormal trend index YQ was calculated.

5. The nursing decision support system based on clinical knowledge graph according to claim 1 is characterized in that: The nursing decision support system also includes an abnormality analysis module. In the abnormality analysis module, an abnormality analysis instruction is generated based on the early warning information, wherein the early warning information includes the patient identity information, the nursing environment information and the nursing operation information, and the nursing environment abnormality index and the nursing operation abnormality index are output; When the nursing environment abnormality index and nursing operation abnormality index meet the corresponding preset values, it indicates that the nursing decision matching is abnormal; When the nursing environment abnormality index and nursing operation abnormality index do not meet the corresponding preset values, an early warning is issued to indicate that the nursing environment and nursing operation are abnormal.

6. The nursing decision support system based on clinical knowledge graph according to claim 5 is characterized in that: The method for obtaining the abnormal nursing environment index HY is as follows: Obtain the deviation Dev of the environmental parameters. Suppose there are Q environmental parameters, p represents the sequential number of the environmental parameters, the deviation of the pth environmental parameter is recorded as Dev_p, and the weight coefficient of the pth environmental parameter is recorded as Wd_p; the actual measured value of the nursing environmental parameter is recorded as M, and the upper and lower limits of the appropriate range of the nursing environmental parameter are recorded as L and U respectively; through the formula Deviation Dev is calculated; mid(L, U) represents the midpoint of the appropriate range of nursing environment parameters, and range(L, U) represents the size of the appropriate range of nursing environment parameters; The abnormal nursing environment index EAI was calculated.

7. The nursing decision support system based on clinical knowledge graph according to claim 5 is characterized in that: The method for obtaining the nursing operation abnormality index is as follows: Suppose there are K nursing operations in the nursing plan, the standard score of each operation is S_s, and the actual score is A _ s, through the formula The abnormal nursing operation index EHI was calculated.

8. The nursing decision support system based on clinical knowledge graph according to claim 5 is characterized in that: The nursing decision support system also includes a nursing plan management module, which is used to obtain nursing alternative plans and screen out recommended nursing plans from the nursing alternative plans based on the plan recommendation index.

Images