Intelligent diagnosis system for postoperation extubation delay analysis of general anesthesia

Abstract

The invention provides an intelligent diagnosis system capable of autonomously judging whether extubation delay occurs to a patient after general anesthesia, and relates to the field of clinical anesthesia research. The system is mainly composed of an embedded logistic regression model used for postoperation extubation delay analysis. The system takes past general anesthesia cases meeting conditions as reference data, trains, optimizes and evaluates a machine learning model based on logistic regression, and can be used for predicting whether a patient to be subjected to a general anesthesia operation has extubation delay or not after being packaged and embedded into a computer system. Compared with a traditional mode of making judgment only according to subjective experience of a physician, the system has the absolute advantages of high efficiency, low cost, high capability and the like, and has a larger application prospect.

Description

Technical Field:

[0001] The present invention relates to the field of clinical anesthesia research, and specifically to an intelligent diagnosis system for analyzing delayed extubation after general anesthesia. Background Art:

[0002] Previous studies have shown that globally, tens of millions of patients undergo general anesthesia surgery every year. Delayed extubation after surgery not only leads to respiratory complications, such as adverse events like vocal cord paralysis and pneumonia, but also increases the patient's treatment cycle and cost. Therefore, early extubation after surgery is beneficial for the patient's early recovery and reducing treatment costs. Predicting whether a patient will have delayed extubation based on preoperative and intraoperative data is crucial for early intervention to avoid delayed extubation. Currently, the judgment and analysis of delayed extubation in general anesthesia patients rely solely on the subjective experience of physicians and manual data analysis, and the differences obtained by different subjects or analysis methods are relatively large, which is not conducive to the early prevention and treatment of patients. The present invention constructs an intelligent diagnosis system that autonomously determines whether a patient after general anesthesia will have delayed extubation based on the surgical data of previous patients of the same type, and it can efficiently and accurately give a judgment result, providing a reference basis for the subsequent treatment of patients. Summary of the Invention:

[0003] In response to the need for more scientific and accurate judgment of whether delayed extubation will occur in general anesthesia patients after surgery, the present invention provides an intelligent diagnosis system for analyzing delayed extubation after general anesthesia. This system is based on computer hardware and consists of an embedded logistic regression machine learning model for analyzing delayed extubation after surgery.

[0004] The present invention conducts a retrospective case-control study on patients undergoing general anesthesia surgery of the same type, queries systems such as the surgical anesthesia system and the electronic medical record system to collect the patients' clinical data. This includes preoperative data such as gender, surgical history, and comorbidities, intraoperative data such as blood loss, hypotension, and anesthetic drugs, and postoperative data such as extubation time. The present invention takes the extubation time as the target value y and other patient data as the feature value x i , and establishes a regression equation:

[0005] y(θ) = θ1x1 + θ2x2 + L + θ n x n + ε (1)

[0006] In formula (1), x1, x2,...x n are feature values, y is the extubation time, θ1, θ2,..., θ n are regression coefficients, and ε is the bias. Using the clinical data of previous patients, a total loss function can be constructed to solve a set of appropriate regression coefficients and biases, so as to more accurately describe the relationship between patient characteristic information and extubation time. The formula of the total loss function is as follows:

[0007]

[0008] In formula (2), m is the number of patient samples, n is the number of features of each sample, θ is the regression coefficient, and y θ (xi) is the extubation time predicted based on the i-th patient's characteristic data, and y i is the actual extubation time of the i-th patient. is the penalty term, and λ is the penalty coefficient. The penalty term can effectively prevent overfitting of patient characteristic data. Combining formula (1) and formula (2), the Stochastic Average Gradient (SAG) algorithm can quickly and accurately determine a set of regression coefficients and biases to establish a linear model between patient characteristic information and extubation time. In addition, the present invention uses 1 hour after surgery as the threshold for judging whether extubation is delayed, and establishes a judgment function as follows:

[0009]

[0010] In formula (3), T = 1 indicates normal extubation, and T = 0 indicates delayed extubation. Inputting the patient data that needs to analyze extubation delay into the linear model established in this study can obtain the predicted extubation time of the patient. Using it as the input of the judgment function can predict whether the patient will have delayed extubation. Description of the Drawings:

[0011] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention and do not constitute a limitation to the present invention.

[0012] Figure 1 is a schematic diagram of the intelligent diagnosis system for analyzing extubation delay after general anesthesia in the present invention. Detailed Embodiments:

[0013] The present invention will be described in detail below with reference to the drawings and specific embodiments. This embodiment is implemented on the premise of the technical solution of the present invention, and gives detailed implementation manners and specific operation processes. However, the protection scope of the present invention is not limited to the following embodiments.

[0014] The specific implementation scheme of the intelligent diagnosis system for analyzing extubation delay after general anesthesia is as follows:

[0015] (1) Within the allowable range of conditions, collect as much clinical data of previous general anesthesia patients as possible, such as information on gender, age, comorbidities, anesthetic drugs, blood loss, and extubation time, etc. After data screening and processing (such as deleting samples with missing feature values, etc.), include them in the data set;

[0016] (2) After shuffling the dataset, randomly select 80% of the samples as training data and 20% of the samples as test data to improve the generalization ability and robustness of the intelligent diagnosis model based on logistic regression;

[0017] (3) Use feature extraction to convert the clinical data of patients (including training data and test data) into digital features that can be read by machine learning. Subsequently, perform standardization processing on the converted training data and test data, and transform the relevant features into the range with a mean of 0 and a standard deviation of 1. Standardized data can not only avoid individual features with large values from dominating the prediction results, but also weaken the influence of outliers in the sample data;

[0018] (4) Establish a machine learning model based on logistic regression, input the above-processed training data into the model for training, optimize the hyperparameters of the logistic regression model through cross-validation and grid search methods, select the optimal parameter combination to establish a diagnosis model for the analysis of delayed extubation after general anesthesia. At the same time, use the test data to evaluate the model. After passing the evaluation, save and package the model, which can be used to predict whether other general anesthesia patients will experience delayed extubation after surgery.

[0019] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent diagnosis system for the analysis of delayed extubation after general anesthesia, characterized in that: a. The system is based on computer hardware and consists of an embedded logistic regression machine learning model for the analysis of delayed extubation after surgery, which can be used to predict whether delayed extubation will occur in patients after general anesthesia; b. The system needs to collect as much clinical data of general anesthesia patients as possible, such as information on gender, age, comorbidities, anesthetic drugs, blood loss, and extubation time, etc. After data screening and processing (such as deleting samples with missing feature values, etc.), they are included in the data set to lay a foundation for subsequent modeling; c. The system takes the extubation time as the target value y and other patient data as the eigenvalue x i , and establishes a regression equation: y(θ) = θ1x1 + θ2x2 + L + θ n x n + ε (1) In formula (1), x1, x2, ... x n are eigenvalues, y is the extubation time, θ1, θ2, …, θ n are regression coefficients, and ε is the bias. Using the patient's clinical data, a total loss function can be constructed to solve a set of appropriate regression coefficients and biases, so as to more accurately describe the relationship between the patient's characteristic information and the extubation time. The formula for the total loss function is as follows: In formula (2), m is the number of patient samples, n is the number of features of each sample, θ is the regression coefficient, and y θ (x i ) is the extubation time predicted based on the feature data of the i-th patient, and y i is the actual extubation time of the i-th patient. is the penalty term, and λ is the penalty coefficient. The penalty term can effectively prevent overfitting of the patient feature data. Combining formula (1) and formula (2), the Stochastic Average Gradient (SAG) algorithm can be used to quickly and accurately determine a set of regression coefficients and biases to establish a linear model between the patient feature information and the extubation time. In addition, the present invention uses one hour after surgery as the threshold for judging whether extubation is delayed, and establishes a judgment function as shown in the following formula: In formula (3), T = 1 indicates normal extubation, and T = 0 indicates delayed extubation. Inputting the data of patients who need to analyze delayed extubation into the linear model established in this study, the predicted extubation time of the patients can be obtained. Taking it as the input of the judgment function, it is possible to predict whether the patients will have delayed extubation.

Images