Patient anesthesia risk assessment method based on multiple features

Abstract

The invention relates to a patient anesthesia risk assessment method based on multiple features, and the method comprises the steps: obtaining the physiological index information, medical history record information and drug use information of a patient as original data, screening the original data, obtaining feature data with high correlation with the anesthesia risk, constructing a composite feature vector, and carrying out the analysis of the composite feature vector; the method comprises the following steps: constructing a preoperative anesthesia risk assessment model based on a random forest algorithm, inputting physiological index information, medical history record information and drug use information of a patient into the preoperative anesthesia risk assessment model, outputting a preoperative anesthesia risk prediction result of the patient, and evaluating the accuracy of the preoperative anesthesia risk prediction result. By optimizing a data processing flow and adopting an advanced feature selection and model construction method, the accuracy of patient anesthesia risk prediction is remarkably improved, a perfect evaluation system is constructed based on specific physiological indexes, medical history and drug use conditions of the patient, the reliability of a model prediction result is ensured, and the medical work efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and in particular to a method for assessing patient anesthesia risk based on multiple features. Background Art

[0002] Anesthesia is an indispensable part of the surgical process. However, complications that may occur during anesthesia, such as hypotension and arrhythmia, will not only affect the effect of the operation, but may also threaten the patient's life safety. How to accurately predict and effectively prevent the occurrence of anesthesia complications has always been a research hotspot in the medical field. In order to ensure that anesthesiologists provide patients with safe and effective treatment, many regions have formulated corresponding anesthesia quality control standards to ensure the quality and safety of anesthesia, including examinations and preparations before anesthesia, basic anesthesia monitoring, and monitoring and treatment during the anesthesia recovery period. New standards are constantly being formulated to further increase patient safety.

[0003] After searching, according to Chinese invention CN202310303505.4, a method for generating and using a prediction model for nausea and vomiting after general anesthesia is proposed. The invention content records that "the model is generated through big data samples, integrating the clinical experience of medical experts and patient feedback, can optimize PONV prevention and treatment strategies, assist in individualized and multi-modal prevention and treatment of PONV, improve patient satisfaction rate, and can predict and give personalized treatment plans, predict the probability of postoperative nausea and vomiting, classify according to the probability, and further give targeted treatment plans. The generated model integrates the clinical experience of medical experts and patient feedback, and the patient information is input into the PONV prediction model to obtain the patient's PONV incidence, medication information, the PONV prediction model can also generate a report for doctors and technicians to view. The content of the report includes: the number of sample sets, the patient information used, the determination coefficient, the number and weight of decision trees, etc. A small program questionnaire or an intelligent voice question and answer can be designed according to the present invention to obtain the patient information required by the present invention, and then the model generated by the present invention is used to finally give the probability of nausea and vomiting of the patient and the recommended medication measures, so that PONV prevention and treatment is more simple, accurate and effective, which is helpful to optimize and homogenize the medical quality of PONV prevention and treatment work in medical institutions at all levels, and improve the overall PONV prevention and treatment level of the medical system.

[0004] Among them, the generation and use methods of the prediction model for postoperative nausea and vomiting after general anesthesia are diverse and may have quality problems, and no effective preprocessing is performed. Features that are highly correlated with PONV need to be accurately screened in high-dimensional data. When building the model, high prediction accuracy must be ensured and overfitting must be avoided. There is a lack of personalized prevention and treatment strategies based on individual differences among patients, and there is a lack of a comprehensive evaluation system to verify the model effect. Therefore, a patient anesthesia risk assessment method based on multiple features is proposed to solve the above problems. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a patient anesthesia risk assessment method based on multiple features, which has the advantages of accurate prediction and solves the problem of poor prediction accuracy.

[0006] To achieve the above object, the present invention provides the following technical solution: a patient anesthesia risk assessment method based on multiple features, S1: Obtain the patient's physiological index information, medical history record information, and drug use information as raw data, and pre-process the raw data, and normalize the pre-processed numerical raw data; S1.1: Obtain the patient's physiological index information, medical history information and medication information as raw data. The physiological index information includes heart rate, blood pressure, blood oxygen saturation, body temperature, respiratory rate and blood sugar. The medical history information includes the type of surgery, the type of disease and the type of allergy. The medication information includes the type of medication, dosage, frequency and time of medication. S1.2: Preprocess the acquired raw data, add missing values ​​in the raw data based on the mean difference method, remove outliers in the raw data based on the IQR interquartile range method, obtain the preprocessed raw data, divide the preprocessed raw data into preprocessed numerical raw data and preprocessed non-numerical raw data, and convert the preprocessed non-numerical raw data into numerical data using one-hot encoding, with each category of non-numerical raw data corresponding to a binary column; S1.3: The preprocessed numerical raw data is normalized to obtain the calculation expression of the preprocessed and normalized numerical raw data: In the formula, is the preprocessed numerical raw data, is the preprocessed and normalized numerical raw data, is the minimum value of the original numerical data after preprocessing, It is the maximum value of the original numerical data after preprocessing; S2: Screen the preprocessed raw data based on correlation analysis to obtain feature data with high correlation with anesthesia risk and construct a composite feature vector; S2.1: Use the Pearson correlation coefficient to perform correlation analysis on the preprocessed raw data. The calculation expression is: In the formula, is the sample size, and are the observed values ​​of two feature data respectively, is the correlation coefficient, ; S2.2: Based on the correlation coefficient The size of the feature data with high correlation with anesthesia risk is screened out, and the threshold , remove the feature data with correlation lower than the threshold, and retain the feature data with correlation higher than the threshold; S2.3: Construct a composite feature vector A based on the retained feature data combination , ; S3: Construct a preoperative anesthesia risk assessment model based on the random forest algorithm, divide the composite feature vector into a training set and a validation set, initialize the preoperative anesthesia risk assessment model, input the training set into the preoperative anesthesia risk assessment model for training, and use the validation set to validate the trained preoperative anesthesia risk assessment model to obtain the optimal preoperative anesthesia risk assessment model; S3.1: Construct a preoperative anesthesia risk assessment model based on the random forest algorithm. Divide into K subsets of equal size, select K-1 subsets each time as training sets, and the remaining subset as validation set. Repeat this process K times, select a different subset as validation set each time, and finally calculate the average result of K validations as the performance indicator of the model; S3.2: Initialize the preoperative anesthesia risk assessment model, use the sampled training set to generate a decision tree, randomly select n features without duplication in the process of generating each node, use the n features to divide the training set respectively, find the best division feature, use the Gini coefficient to measure the purity of the decision tree, and the calculation expression is: in, is the training set, is the number of categories in the training set, is the proportion of the i-th decision tree in the training set; S3.3: Use the validation set to validate the trained preoperative anesthesia risk assessment model based on the accuracy , Accuracy , recall rate The accuracy of the prediction was evaluated by the F1 score, and the stability of the prediction was evaluated based on the standard deviation and variance. The optimal preoperative anesthesia risk assessment model was obtained, and the calculation expression was: in, is a true positive example, that is, the number of samples that are predicted as positive examples by the preoperative anesthesia risk assessment model and are actually positive examples. is a true negative example, that is, the number of samples that are predicted as negative examples by the preoperative anesthesia risk assessment model and are actually negative examples. are false positives, that is, the number of samples that are predicted as positive by the preoperative anesthesia risk assessment model but are actually negative. is a true negative example, that is, the number of samples predicted by the preoperative anesthesia risk assessment model as negative examples but actually positive examples. For each observation in the validation set, is the mean of the validation set, is the total number of observations in the validation set; S4: Input the patient's physiological index information, medical history record information, and drug use information into the optimal preoperative anesthesia risk assessment model, and output the patient's preoperative anesthesia risk prediction result; S5: Compare the patient's preoperative anesthesia risk prediction results with the patient's actual postoperative anesthesia risk results, build an evaluation system based on the patient's actual postoperative anesthesia risk results, and evaluate the accuracy of the preoperative anesthesia risk prediction results; S5.1: Obtain the postoperative predicted physiological index information and the actual postoperative physiological index information from the patient's preoperative anesthesia risk prediction results, and calculate the comprehensive value of the predicted physiological indexes And the comprehensive value of physiological indicators of actual results , the calculation expression is: In the formula, To predict the resulting heart rate, is the actual result heart rate, is the heart rate weight, To predict the outcome blood pressure, For actual blood pressure results, is the blood pressure weight, To predict the result blood oxygen saturation, is the actual result blood oxygen saturation, is the blood oxygen saturation weight, To predict the resulting body temperature, For the actual result body temperature, is the body temperature weight, To predict the resulting respiratory rate, is the actual result respiratory rate, is the respiratory rate weight, To predict the outcome blood glucose, For actual blood sugar results, is the blood sugar weight; S5.2: Determine the comprehensive value of physiological indicators of the predicted results Comprehensive value of physiological indicators with actual results Is the absolute value of the difference less than the threshold? If yes, go to S5.3; if no, go to S5.4. The calculation expression is: In the formula, Comprehensive value of physiological indexes Comprehensive value of physiological indicators with actual results The absolute value of the difference; S5.2.1: Get the absolute values ​​of all differences , and arrange them from largest to smallest; S5.2.2: Select all absolute values ​​of differences the median of ; S5.2.3: Based on the absolute value of all differences The median calculation threshold is: In the formula, is the threshold value, is the coefficient, , is the absolute value of all differences the median; S5.3: There is no need to optimize the preoperative anesthetic risk assessment model; S5.4: The patient's physiological indicator information, medical history record information, and drug use information are used as raw data for the preoperative anesthesia risk assessment model.

[0007] Compared with the prior art, the technical solution of this application has the following beneficial effects: This multi-feature-based patient anesthesia risk assessment method significantly improves the accuracy of patient anesthesia risk prediction by optimizing the data processing process and adopting advanced feature selection and model building methods. It builds a complete evaluation system based on the patient's specific physiological indicators, medical history and drug usage, ensures the reliability of the model's prediction results, and improves medical work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 It is a flow chart of the present invention; Figure 2 This is a flow chart of step 1 of the present invention; Figure 3 This is a flow chart of step 2 of the present invention; Figure 4 This is a flow chart of step 3 of the present invention; Figure 5 This is a flow chart of step 5 of the present invention. DETAILED DESCRIPTION

[0009] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0010] See also Figure 1-5 , a patient anesthesia risk assessment method based on multiple features in this embodiment, S1: Obtain the patient's physiological index information, medical history record information, and drug use information as raw data, and pre-process the raw data, and normalize the pre-processed numerical raw data; S1.1: Obtain the patient's physiological index information, medical history information and medication information as raw data. The physiological index information includes heart rate, blood pressure, blood oxygen saturation, body temperature, respiratory rate and blood sugar. The medical history information includes the type of surgery, the type of disease and the type of allergy. The medication information includes the type of medication, dosage, frequency and time of medication. S1.2: Preprocess the acquired raw data. For missing raw data, first check whether it can be filled by interpolating data from adjacent time points. If interpolation is not possible, use the mean difference method to fill it. Use the IQR interquartile range method to detect and remove outliers in the raw data. For outliers, further analyze their causes, such as equipment failure, special patient conditions, etc., and record them. For non-numerical data, such as types of surgery, types of diseases, and types of allergies, use one-hot encoding to convert them into numerical data to obtain preprocessed raw data. The preprocessed raw data is divided into preprocessed numerical raw data and preprocessed non-numerical raw data, and the preprocessed non-numerical raw data is converted into numerical data using one-hot encoding. Each category of non-numerical raw data corresponds to a binary column. S1.3: The preprocessed numerical raw data is normalized to obtain the calculation expression of the preprocessed and normalized numerical raw data: In the formula, is the preprocessed numerical raw data, is the preprocessed and normalized numerical raw data, is the minimum value of the original numerical data after preprocessing, It is the maximum value of the original numerical data after preprocessing; S2: Screen the preprocessed raw data based on correlation analysis to obtain feature data with high correlation with anesthesia risk and construct a composite feature vector; S2.1: Use the Pearson correlation coefficient to perform correlation analysis on the preprocessed raw data. The Pearson correlation coefficient can quantify the strength and direction of the linear relationship between two continuous variables. Its value ranges from -1 to 1, where -1 indicates a complete negative correlation, 1 indicates a complete positive correlation, and 0 indicates no correlation. This quantitative method allows researchers to intuitively understand the degree of association between variables. The calculation of the Pearson correlation coefficient is relatively simple. It is based on the covariance of two variables and their respective standard deviations and does not require complex mathematical models or algorithms, which makes it easy to apply and understand in data analysis. Correlation analysis can reveal the degree of association between variables and help researchers understand the interaction between variables, which is of great significance for subsequent data mining, model building, and predictive analysis. Through correlation analysis, redundant features or noise in the data can be discovered, thereby guiding data preprocessing. For example, for highly correlated features, one of them can be selected as a representative to reduce the number of features and reduce model complexity. When building a predictive model, selecting features that are highly correlated with the target variable as input variables can improve the predictive performance of the model. Correlation analysis provides a basis for this selection. The calculation expression is: In the formula, is the sample size, and are the observed values ​​of two feature data respectively, is the correlation coefficient, ∈ ; S2.2: Based on the correlation coefficient The size of the feature data with high correlation with anesthesia risk is screened out, and the threshold , remove the feature data with correlation lower than the threshold, and retain the feature data with correlation higher than the threshold; S2.3: Construct a composite feature vector A based on the retained feature data combination , ; S3: A preoperative anesthesia risk assessment model is constructed based on the random forest algorithm. The composite feature vector is divided into a training set and a validation set. The preoperative anesthesia risk assessment model is initialized, and the training set is input into the preoperative anesthesia risk assessment model for training. The trained preoperative anesthesia risk assessment model is validated using the validation set to obtain the optimal preoperative anesthesia risk assessment model. The random forest constructs multiple decision trees and combines their outputs for classification or regression, thereby reducing the risk of overfitting and improving the generalization ability and accuracy of the model. In preoperative anesthesia risk assessment, high accuracy means that the patient's anesthesia risk can be predicted more accurately, providing doctors with more reliable decision support. The patient's physiological indicators, medical history records and other information often contain multiple dimensions. The random forest algorithm can directly process these high-dimensional data without the need for additional feature selection steps. The patient's physiological indicator data may be affected by many factors, such as equipment errors, patient mood swings, etc. The random forest algorithm can resist the interference of these noises to a certain extent and improve the stability of the model. S3.1: K-fold cross validation is widely used in various tasks in data science and machine learning. A preoperative anesthesia risk assessment model is constructed based on the random forest algorithm. The composite feature vector A Divide into K subsets of equal size, select K-1 subsets as training sets each time, and the remaining subset as validation set. Repeat this process K times, select different subsets as validation sets each time, and finally calculate the average result of K validations as the performance indicator of the model. K-fold cross validation uses all samples of the data set for training and testing, avoiding data waste and obtaining a more stable model performance evaluation result. S3.2: Initialize the preoperative anesthesia risk assessment model, use the sampled training set to generate a decision tree, randomly select n features without duplication in the process of generating each node, use the n features to divide the training set respectively, find the best division feature, use the Gini coefficient to measure the purity of the decision tree, and the calculation expression is: in, is the training set, is the number of categories in the training set, is the proportion of the i-th decision tree in the training set; S3.3: Use the validation set to validate the trained preoperative anesthesia risk assessment model based on the accuracy , Accuracy , recall rate The accuracy of the prediction was evaluated by the F1 score, and the stability of the prediction was evaluated based on the standard deviation and variance. The optimal preoperative anesthesia risk assessment model was obtained, and the calculation expression was: in, is a true positive example, that is, the number of samples that are predicted as positive examples by the preoperative anesthesia risk assessment model and are actually positive examples. is a true negative example, that is, the number of samples that are predicted as negative examples by the preoperative anesthesia risk assessment model and are actually negative examples. are false positives, that is, the number of samples that are predicted as positive by the preoperative anesthesia risk assessment model but are actually negative. is a true negative example, that is, the number of samples predicted by the preoperative anesthesia risk assessment model as negative examples but actually positive examples. For each observation in the validation set, is the mean of the validation set, is the total number of observations in the validation set; S4: Input the patient's physiological index information, medical history record information, and drug use information into the optimal preoperative anesthesia risk assessment model, and output the patient's preoperative anesthesia risk prediction result. The patient's preoperative anesthesia risk prediction result can serve as an important reference for anesthesiologists to formulate anesthesia plans and make decisions, which helps doctors to better plan the anesthesia process before surgery, reduce surgical risks, and improve the success rate of surgery. Doctors can use medical resources more effectively. For high-risk patients, preventive measures can be taken in advance to reduce intraoperative and postoperative complications. For low-risk patients, a simpler and more economical anesthesia plan can be used; S5: Compare the patient's preoperative anesthesia risk prediction results with the patient's actual postoperative anesthesia risk results, build an evaluation system based on the patient's actual postoperative anesthesia risk results, and evaluate the accuracy of the preoperative anesthesia risk prediction results. By comparing the preoperative prediction results with the actual postoperative results, the accuracy of the prediction model can be intuitively evaluated. Accurate preoperative anesthesia risk prediction helps anesthesiologists develop safer and more effective anesthesia plans for patients. Based on the prediction results, doctors can take preventive measures in advance to reduce the anesthesia risk during surgery and ensure the safety of patients. The accuracy of preoperative anesthesia risk prediction results is directly related to the quality of medical care. By building an evaluation system, hospitals can monitor the quality of anesthesia risk prediction, promptly discover and correct deviations in the prediction, thereby improving the overall quality of medical care. Accurate preoperative anesthesia risk prediction and comparison of actual postoperative results can help enhance patients' trust in the medical team, which helps to establish a good doctor-patient relationship and improve patient satisfaction and compliance, thereby ensuring the smooth progress of the surgery and anesthesia process. S5.1: Obtain the postoperative predicted physiological index information and the actual postoperative physiological index information from the patient's preoperative anesthesia risk prediction results, and calculate the comprehensive value of the predicted physiological indexes And the comprehensive value of physiological indicators of actual results , the calculation expression is: In the formula, To predict the resulting heart rate, is the actual result heart rate, is the heart rate weight, To predict the outcome blood pressure, For actual blood pressure results, is the blood pressure weight, To predict the result blood oxygen saturation, is the actual result blood oxygen saturation, is the blood oxygen saturation weight, To predict the resulting body temperature, For the actual result body temperature, is the body temperature weight, To predict the resulting respiratory rate, is the actual result respiratory rate, is the respiratory rate weight, To predict the outcome blood glucose, For actual blood sugar results, is the blood sugar weight; Table 1: Weights of different types of physiological indicators corresponding to different types of diseases S5.2: Determine the comprehensive value of physiological indicators of the predicted results Comprehensive value of physiological indicators with actual results Is the absolute value of the difference less than the threshold? If yes, go to S5.3; if no, go to S5.4. The calculation expression is: In the formula, Comprehensive value of physiological indexes Comprehensive value of physiological indicators with actual results The absolute value of the difference; S5.2.1: Get the absolute values ​​of all differences , and arrange them from largest to smallest; S5.2.2: Select all absolute values ​​of differences the median of ; S5.2.2.1: Determine the absolute value of all differences Is the number of is an odd number? If yes, proceed to S5.2.2.2; if no, proceed to S5.2.2.3; S5.2.2.2: Select all absolute values ​​of differences The absolute value of the middle difference As the median ; S5.2.2.3: Select all absolute values ​​of differences The absolute value of the difference between the two middle values The median is the average value of ; S5.2.3: Based on the absolute value of all differences the median of Calculate the threshold value. The calculation expression is: In the formula, is the threshold value, is the coefficient, , is the absolute value of all differences the median; Table 2: Coefficients of different types of diseases corresponding to different types of physiological indicator information S5.3: There is no need to optimize the preoperative anesthetic risk assessment model; S5.4: The patient's physiological indicator information, medical history record information, and drug use information are used as raw data for the preoperative anesthesia risk assessment model.

[0011] The working principle of the above embodiment is: This multi-feature-based patient anesthesia risk assessment method significantly improves the accuracy of patient anesthesia risk prediction by optimizing the data processing process and adopting advanced feature selection and model building methods. It builds a complete evaluation system based on the patient's specific physiological indicators, medical history and drug usage, ensures the reliability of the model's prediction results, and improves medical work efficiency.

[0012] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0013] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-feature based patient anesthesia risk assessment method, characterized in that: The specific steps include: S1: Acquire the patient's physiological index information, medical history record information, and drug use information as raw data, preprocess the raw data, and normalize the preprocessed raw data; S2: Screen the preprocessed raw data based on correlation analysis to obtain feature data with high correlation with anesthesia risk and construct a composite feature vector; S3: Construct a preoperative anesthesia risk assessment model based on the random forest algorithm, divide the composite feature vector into a training set and a validation set, initialize the preoperative anesthesia risk assessment model, input the training set into the preoperative anesthesia risk assessment model for training, and use the validation set to validate the trained preoperative anesthesia risk assessment model to obtain the optimal preoperative anesthesia risk assessment model; S4: Input the patient's physiological index information, medical history record information, and drug use information into the optimal preoperative anesthesia risk assessment model, and output the patient's preoperative anesthesia risk prediction result; S5: Compare the patient's preoperative anesthesia risk prediction results with the patient's actual postoperative anesthesia risk results, build an evaluation system based on the patient's actual postoperative anesthesia risk results, and evaluate the accuracy of the preoperative anesthesia risk prediction results.

2. A multi-feature based patient anesthesia risk assessment method as claimed in claim 1, characterized in that: Step 1, in which the patient's physiological index information, medical history record information, and drug use information are obtained as raw data, and the raw data are preprocessed, and the preprocessed raw data are normalized, includes the following sub-steps: S1.1: Obtain the patient's physiological index information, medical history information and medication information as raw data. The physiological index information includes heart rate, blood pressure, blood oxygen saturation, body temperature, respiratory rate and blood sugar. The medical history information includes the type of surgery, the type of disease and the type of allergy. The medication information includes the type of medication, dosage, frequency and time of medication. S1.2: Preprocess the acquired raw data, add missing values ​​in the raw data based on the mean difference method, remove outliers in the raw data based on the IQR interquartile range method, obtain the preprocessed raw data, divide the preprocessed raw data into preprocessed numerical raw data and preprocessed non-numerical raw data, and convert the preprocessed non-numerical raw data into numerical data using one-hot encoding, with each category of non-numerical raw data corresponding to a binary column; S1.3: Perform normalization processing on the preprocessed numerical raw data to obtain preprocessed and normalized numerical raw data.

3. A multi-feature based patient anesthesia risk assessment method as claimed in claim 1, characterized in that: Step 2, based on correlation analysis, screens the preprocessed raw data to obtain feature data with high correlation with anesthesia risk, and constructs a composite feature vector, including the following sub-steps: S2.1: Use the Pearson correlation coefficient to perform correlation analysis on the preprocessed raw data. The calculation expression is: In the formula, is the sample size, and are the observed values ​​of two feature data respectively, is the correlation coefficient, ; S2.2: Based on the correlation coefficient The size of the feature data with high correlation with anesthesia risk is screened out, and the threshold , remove the feature data with correlation lower than the threshold, and retain the feature data with correlation higher than the threshold; S2.3: Construct a composite feature vector A based on the retained feature data combination , .

4. A multi-feature based patient anesthesia risk assessment method as claimed in claim 1, characterized in that: In step 3, a preoperative anesthesia risk assessment model is constructed based on a random forest algorithm, the composite feature vector is divided into a training set and a validation set, the preoperative anesthesia risk assessment model is initialized, the training set is input into the preoperative anesthesia risk assessment model for training, and the trained preoperative anesthesia risk assessment model is validated using the validation set to obtain the optimal preoperative anesthesia risk assessment model, including the following sub-steps: S3.1: Construct a preoperative anesthesia risk assessment model based on the random forest algorithm. Divide into K subsets of equal size, select K-1 subsets each time as training sets, and the remaining subset as validation set. Repeat this process K times, select a different subset as validation set each time, and finally calculate the average result of K validations as the performance indicator of the model; S3.2: Initialize the preoperative anesthesia risk assessment model, use the sampled training set to generate a decision tree, randomly select n features without duplication in the process of generating each node, use the n features to divide the training set respectively, find the best division feature, and use the Gini coefficient to measure the purity of the decision tree; S3.3: Use the validation set to validate the trained preoperative anesthesia risk assessment model based on the accuracy , Accuracy , recall rate The accuracy of the prediction is evaluated by the F1 score, and the stability of the prediction is evaluated based on the standard deviation StandardDeviation and variance Variance. The optimal preoperative anesthesia risk assessment model is obtained, and the calculation expression is: in, is a true positive example, that is, the number of samples that are predicted as positive examples by the preoperative anesthesia risk assessment model and are actually positive examples. is a true negative example, that is, the number of samples that are predicted as negative examples by the preoperative anesthesia risk assessment model and are actually negative examples. are false positives, that is, the number of samples that are predicted as positive by the preoperative anesthesia risk assessment model but are actually negative. is a true negative example, that is, the number of samples predicted by the preoperative anesthesia risk assessment model as negative examples but actually positive examples. For each observation in the validation set, is the mean of the validation set, is the total number of observations in the validation set.

5. The multi-feature based patient anesthesia risk assessment method according to claim 1, characterized in that: Step 5 compares the patient's preoperative anesthesia risk prediction result with the patient's actual postoperative anesthesia risk result, builds an evaluation system based on the patient's actual postoperative anesthesia risk result, and evaluates the accuracy of the preoperative anesthesia risk prediction result, including the following sub-steps: S5.1: Obtain the postoperative predicted physiological index information and the actual postoperative physiological index information from the patient's preoperative anesthesia risk prediction results, and calculate the comprehensive value of the predicted physiological indexes And the comprehensive value of physiological indicators of actual results , the calculation expression is: In the formula, To predict the resulting heart rate, is the actual result heart rate, is the heart rate weight, To predict the outcome blood pressure, For actual blood pressure results, is the blood pressure weight, To predict the result blood oxygen saturation, is the actual result blood oxygen saturation, is the blood oxygen saturation weight, To predict the resulting body temperature, For the actual result body temperature, is the body temperature weight, To predict the resulting respiratory rate, is the actual result respiratory rate, is the respiratory rate weight, To predict the outcome blood glucose, For actual blood sugar results, is the blood sugar weight; S5.2: Determine the comprehensive value of physiological indicators of the predicted results Comprehensive value of physiological indicators with actual results Is the absolute value of the difference less than the threshold? If yes, go to S5.3; if no, go to S5.

4. The calculation expression is: In the formula, Comprehensive value of physiological indexes Comprehensive value of physiological indicators with actual results The absolute value of the difference; S5.3: There is no need to optimize the preoperative anesthetic risk assessment model; S5.4: The patient's physiological indicator information, medical history record information, and drug use information are used as raw data for the preoperative anesthesia risk assessment model.

6. A multi-feature based patient anesthesia risk assessment method as claimed in claim 5, characterized in that: The method for obtaining the threshold value in step 5.2 comprises the following steps: S5.2.1: Get the absolute values ​​of all differences , and arrange them from largest to smallest; S5.2.2: Select all absolute values ​​of differences the median of ; S5.2.3: Based on the absolute value of all differences The median calculation threshold is: In the formula, is the threshold value, is the coefficient, , is the absolute value of all differences of the median.

Images