Method and system for early warning delirium of patient after abdominal operation

Abstract

The invention provides an early warning method and system for delirium of a patient after an abdominal operation, and relates to the technical field of medical treatment.The early warning method comprises the steps that historical clinical data corresponding to the abdominal operation and the historical postoperative delirium occurrence probability are obtained; receiving the historical clinical data and the historical postoperative delirium occurrence probability, and extracting historical clinical features in the historical clinical data based on a feature selection algorithm; based on a machine learning algorithm, constructing a corresponding postoperative delirium risk prediction model according to the historical clinical features and the historical postoperative delirium occurrence probability; obtaining target clinical data corresponding to the patient end, extracting corresponding target clinical features, inputting the target clinical features into the postoperative delirium risk prediction model, and calculating to obtain a target postoperative delirium occurrence probability corresponding to the patient end; according to the target postoperative delirium occurrence probability, whether intervention treatment needs to be carried out on the patient side or not is judged, so that the postoperative delirium occurrence rate of the patient can be reduced, and the treatment effect and life quality of the patient are improved.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and in particular to an early warning method and system for delirium in patients after abdominal surgery. Background Art

[0002] Postoperative delirium is an acute, temporary brain dysfunction state that usually occurs within a week after surgery and is characterized by inattention, changes in consciousness level, and acute changes in cognitive function. The incidence of postoperative delirium is particularly high in elderly patients, which increases medical costs, prolongs hospital stays, and may affect patients' long-term cognitive function and even increase the risk of death.

[0003] In the prior art, before abdominal surgery, traditional postoperative delirium warning methods often fail to accurately predict the risk of postoperative delirium in patients, and thus take early intervention measures. At the same time, this method is difficult to formulate personalized intervention methods according to the actual situation of the patient, which often leads to poor treatment effects, high medical costs, and may prolong the patient's postoperative recovery time.

[0004] Therefore, it is necessary to provide an early warning method and system for delirium in patients after abdominal surgery to solve the above technical problems. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides an early warning method for delirium in patients after abdominal surgery and a system thereof, which are used to solve the problem that traditional early warning methods for postoperative delirium cannot accurately predict the risk of postoperative delirium in patients before abdominal surgery, and then take early intervention measures. At the same time, it is difficult to formulate personalized intervention methods according to the actual situation of the patients, resulting in poor treatment effects, high medical costs, and may prolong the patient's postoperative recovery time.

[0006] The present invention provides a method for early warning of delirium in patients after abdominal surgery, the method comprising:

[0007] Obtain historical clinical data corresponding to abdominal surgery and historical postoperative delirium incidence;

[0008] Receiving the historical clinical data and the historical probability of postoperative delirium occurrence, and extracting historical clinical features from the historical clinical data based on a feature selection algorithm;

[0009] Based on the machine learning algorithm, a corresponding postoperative delirium risk prediction model is constructed according to the historical clinical characteristics and the historical probability of postoperative delirium occurrence;

[0010] Obtaining target clinical data corresponding to the patient side, extracting corresponding target clinical features and inputting them into the postoperative delirium risk prediction model, and calculating the target postoperative delirium occurrence probability corresponding to the patient side;

[0011] Whether intervention treatment is required for the patient is determined based on the target probability of postoperative delirium occurrence.

[0012] Preferably, the historical clinical data includes the patient's age, gender, past medical history, medication history, preoperative cognitive function assessment, frailty index, abdominal surgery time, anesthesia time, anesthetic drugs, anesthesia methods, analgesia and sedation history and infection status.

[0013] Preferably, the receiving of the historical clinical data and the historical probability of postoperative delirium occurrence, and extracting the historical clinical features in the historical clinical data based on a feature selection algorithm, specifically includes:

[0014] The historical clinical data and the historical probability of postoperative delirium were preprocessed;

[0015] Based on the feature selection algorithm, the historical clinical data is used as the variable X, and the historical probability of postoperative delirium is used as the variable Y. The mutual information value I (X, Y) between the historical clinical data and the historical probability of postoperative delirium is calculated. The calculation formula of the mutual information value is:

[0016]

[0017] Where, I(X, Y) represents the mutual information value between historical clinical data and historical postoperative delirium occurrence probability; X represents historical clinical data; Y represents historical postoperative delirium occurrence probability; x represents the value of historical clinical data; y represents the value of historical postoperative delirium occurrence probability; p(x, y) represents the joint probability distribution of X and Y taking x and y at the same time; p(x) represents the marginal probability distribution of X taking x; p(y) represents the marginal probability distribution of Y taking y; log() represents the logarithmic function; Σ represents the summation symbol;

[0018] A preset mutual information value threshold is obtained, and historical clinical data corresponding to a mutual information value greater than or equal to the preset mutual information value threshold is selected as the historical clinical feature.

[0019] Preferably, the method based on the machine learning algorithm and the historical clinical characteristics and the historical probability of postoperative delirium occurrence constructs a corresponding postoperative delirium risk prediction model, which specifically includes:

[0020] Dividing the historical clinical characteristics into training clinical characteristics and testing clinical characteristics, and dividing the historical postoperative delirium occurrence probability into training postoperative delirium occurrence probability and testing postoperative delirium occurrence probability;

[0021] Based on the support vector machine, the postoperative delirium risk prediction model is constructed according to the training clinical characteristics and the training postoperative delirium occurrence probability, and the corresponding objective function is:

[0022]

[0023] In the formula, max represents the maximum value operation; w represents the weight vector of the objective function corresponding to the postoperative delirium risk prediction model; ||w|| 2 represents the square of the modulus of the weight vector; C represents the regularization parameter of the objective function; m represents the total number of training data sets corresponding to the training clinical features and the training probability of postoperative delirium; ξ i represents the slack variable corresponding to the i-th training data set; Σ represents the summation symbol;

[0024] The constraints corresponding to the objective function are:

[0025]

[0026] In the formula, st represents the constraint condition; y′ i represents the probability of postoperative delirium during training; w represents the weight vector of the objective function corresponding to the postoperative delirium risk prediction model; x′ i represents the training clinical characteristics; b represents the intercept term of the objective function corresponding to the postoperative delirium risk prediction model; ξ i represents the slack variable corresponding to the i-th training data group;

[0027] Based on the constraints, the objective function is maximized and the postoperative delirium risk prediction model is trained.

[0028] Preferably, after obtaining the postoperative delirium risk prediction model, the prediction effect of the postoperative delirium risk prediction model is evaluated based on the tested clinical characteristics and the tested probability of postoperative delirium occurrence, and the postoperative delirium risk prediction model is optimized according to the predicted effect, specifically including:

[0029] Obtaining a confusion matrix corresponding to the tested clinical characteristics and the tested probability of postoperative delirium occurrence, and evaluating the prediction effect of the postoperative delirium risk prediction model according to the confusion matrix;

[0030] The elements of the confusion matrix include true positive examples, false positive examples, true negative examples and false negative examples, wherein the true positive example is the correct identification of postoperative delirium occurrence; the false positive example is the wrong identification of postoperative delirium occurrence; the true negative example is the correct identification of postoperative delirium non-occurrence; the false negative example is the wrong identification of postoperative delirium non-occurrence;

[0031] Based on the confusion matrix, the prediction indicators corresponding to the postoperative delirium risk prediction model are calculated, wherein the prediction indicators include accuracy, precision, recall, F1 score and false positive rate;

[0032] The calculation formula of the accuracy is:

[0033]

[0034] In the formula, ZQL represents the accuracy rate corresponding to the postoperative delirium risk prediction model; TP represents true positive examples, that is, correctly identifying the occurrence of postoperative delirium; TN represents true negative examples, that is, correctly identifying that postoperative delirium did not occur; FP represents false positive examples, that is, incorrectly identifying the occurrence of postoperative delirium; FN represents false negative examples, that is, incorrectly identifying that postoperative delirium did not occur;

[0035] The calculation formula of the accuracy is:

[0036]

[0037] In the formula, JQL represents the accuracy corresponding to the postoperative delirium risk prediction model; TP represents true positives, that is, correctly identifying the occurrence of postoperative delirium; FP represents false positives, that is, incorrectly identifying the occurrence of postoperative delirium;

[0038] The calculation formula of the recall rate is:

[0039]

[0040] In the formula, ZHL represents the recall rate corresponding to the postoperative delirium risk prediction model; TP represents a true positive example, that is, correctly identifying the occurrence of postoperative delirium; FN represents a false negative example, that is, incorrectly identifying that postoperative delirium did not occur;

[0041] The calculation formula of the F1 score is:

[0042]

[0043] In the formula, F1 represents the F1 score corresponding to the postoperative delirium risk prediction model; JQL represents the precision corresponding to the postoperative delirium risk prediction model; ZHL represents the recall rate corresponding to the postoperative delirium risk prediction model;

[0044] The calculation formula of the false positive rate is:

[0045]

[0046] In the formula, JZL represents the false positive rate corresponding to the postoperative delirium risk prediction model; FP represents the false positive example, that is, the wrong identification of postoperative delirium; TN represents the true negative example, that is, the correct identification of postoperative delirium not occurring;

[0047] The prediction indicators are analyzed, and the precision and recall are weighed to evaluate the prediction effect of the postoperative delirium risk prediction model, and the postoperative delirium risk prediction model is optimized according to the prediction effect.

[0048] Preferably, judging whether intervention treatment is needed for the patient according to the target probability of postoperative delirium specifically includes:

[0049] If intervention treatment is required for the patient end, the patient end is divided into a low-risk patient end, a medium-risk patient end and a high-risk patient end according to the target probability of postoperative delirium;

[0050] For the patients at different risk levels, formulate corresponding personalized intervention treatment plans, and perform intervention treatment on the patients according to the personalized intervention treatment plans;

[0051] The patient's intervention treatment process is monitored in real time, and the corresponding intervention treatment effect is evaluated, and the personalized intervention treatment plan is adjusted accordingly according to the evaluation results.

[0052] The present invention provides an early warning system for postoperative delirium in abdominal surgery patients, the system comprising:

[0053] An acquisition module is used to obtain historical clinical data corresponding to abdominal surgery and the historical probability of postoperative delirium;

[0054] An extraction module, configured to receive the historical clinical data and the historical probability of postoperative delirium occurrence, and extract historical clinical features from the historical clinical data based on a feature selection algorithm;

[0055] A construction module is used to construct a corresponding postoperative delirium risk prediction model based on a machine learning algorithm according to the historical clinical characteristics and the historical probability of postoperative delirium occurrence;

[0056] An input module, used to obtain target clinical data corresponding to the patient side, extract corresponding target clinical features and input them into the postoperative delirium risk prediction model, and calculate the target postoperative delirium occurrence probability corresponding to the patient side;

[0057] The judgment module is used to judge whether intervention treatment is needed for the patient according to the target probability of postoperative delirium.

[0058] Compared with the related art, the early warning method and system for delirium in patients after abdominal surgery provided by the present invention have the following beneficial effects:

[0059] The present invention obtains historical clinical data and historical postoperative delirium occurrence probability corresponding to abdominal surgery; receives historical clinical data and historical postoperative delirium occurrence probability, extracts historical clinical features in historical clinical data based on feature selection algorithm; constructs corresponding postoperative delirium risk prediction model based on historical clinical features and historical postoperative delirium occurrence probability based on machine learning algorithm; obtains target clinical data corresponding to the patient end, extracts corresponding target clinical features and inputs them into postoperative delirium risk prediction model, calculates target postoperative delirium occurrence probability corresponding to the patient end; judges whether intervention treatment is needed on the patient end according to target postoperative delirium occurrence probability. The present invention can predict the risk of postoperative delirium of patients before and during surgery, so as to take intervention measures at an early stage; at the same time, according to the actual condition of the patient, formulates corresponding personalized intervention treatment plan to improve the treatment effect; and the present invention can reduce the occurrence of postoperative delirium of patients through early warning and intervention, reduce medical costs and hospitalization time, and thus help patients recover after surgery and improve their quality of life. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a flow chart of a method for early warning of delirium in patients after abdominal surgery according to the present invention;

[0061] Figure 2 is a schematic diagram of confusion matrix elements of the present invention;

[0062] Figure 3 A schematic diagram for classifying risk levels on the patient side;

[0063] Figure 4 The present invention is a system block diagram of an early warning system for delirium in patients after abdominal surgery. DETAILED DESCRIPTION

[0064] The present invention will be further described below in conjunction with the accompanying drawings and implementation modes.

[0065] Embodiment 1

[0066] like Figure 1 As shown, a method for early warning of delirium in patients after abdominal surgery, the method comprising:

[0067] S1, obtain historical clinical data corresponding to abdominal surgery and the historical probability of postoperative delirium;

[0068] Among them, in order to accurately assess the risk of delirium in patients after abdominal surgery, we can first systematically obtain historical clinical data corresponding to abdominal surgery and the historical probability of postoperative delirium. Specifically, detailed clinical data can be extracted from the medical record system, including age, gender, medical history, medication history, preoperative cognitive function assessment, frailty index, abdominal surgery time, anesthesia time, anesthetic drugs, anesthesia methods, analgesia and sedation history, and infection.

[0069] S2, receiving the historical clinical data and the historical probability of postoperative delirium occurrence, and extracting historical clinical features from the historical clinical data based on a feature selection algorithm;

[0070] It is understandable that based on data processing technology, these historical clinical data and historical postoperative delirium occurrence probabilities can be received and integrated. Using feature selection algorithms, historical clinical features that have a greater impact on the incidence of postoperative delirium can be screened out from a large number of candidate features, thereby reducing the complexity of the subsequent postoperative delirium risk prediction model and improving the accuracy of postoperative delirium risk prediction.

[0071] S3, based on a machine learning algorithm, constructing a corresponding postoperative delirium risk prediction model according to the historical clinical characteristics and the historical probability of postoperative delirium occurrence;

[0072] It should be noted that based on the machine learning algorithm, a corresponding postoperative delirium risk prediction model can be constructed according to historical clinical characteristics and historical postoperative delirium probability. Then, by learning the patterns and associations in the historical data, the probability of patients developing delirium after undergoing the same surgery can be predicted.

[0073] S4, obtaining target clinical data corresponding to the patient side, extracting corresponding target clinical features and inputting them into the postoperative delirium risk prediction model, and calculating the target postoperative delirium occurrence probability corresponding to the patient side;

[0074] In practical applications, the target clinical data corresponding to the patient can be obtained, and the same feature extraction method can be used to extract the target clinical features of the patient. These features are then input into a trained postoperative delirium risk prediction model, and the target probability of postoperative delirium occurring in the patient after abdominal surgery is calculated and output.

[0075] S5: judging whether intervention treatment is needed for the patient according to the target probability of postoperative delirium.

[0076] Finally, based on the target probability of postoperative delirium output by the postoperative delirium risk prediction model, combined with clinical guidelines and expert opinions, it can be judged whether the patient needs early intervention treatment, such as strengthening preoperative psychological support, optimizing anesthesia regimen, close postoperative monitoring and early identification of delirium symptoms, etc., thereby reducing the incidence of postoperative delirium in patients and improving the quality of postoperative recovery and overall medical outcomes.

[0077] During the specific implementation process, the historical clinical data includes the patient's age, gender, medical history, medication history, preoperative cognitive function assessment, frailty index, abdominal surgery time, anesthesia time, anesthetic drugs, anesthesia methods, analgesia and sedation history and infection status.

[0078] The receiving of the historical clinical data and the historical probability of postoperative delirium occurrence, and extracting the historical clinical features in the historical clinical data based on a feature selection algorithm, specifically includes:

[0079] The historical clinical data and the historical probability of postoperative delirium were preprocessed;

[0080] Based on the feature selection algorithm, the historical clinical data is used as the variable X, and the historical probability of postoperative delirium is used as the variable Y. The mutual information value I (X, Y) between the historical clinical data and the historical probability of postoperative delirium is calculated. The calculation formula of the mutual information value is:

[0081]

[0082] Where, I(X, Y) represents the mutual information value between historical clinical data and historical postoperative delirium occurrence probability; X represents historical clinical data; Y represents historical postoperative delirium occurrence probability; x represents the value of historical clinical data; y represents the value of historical postoperative delirium occurrence probability; p(x, y) represents the joint probability distribution of X and Y taking x and y at the same time; p(x) represents the marginal probability distribution of X taking x; p(y) represents the marginal probability distribution of Y taking y; log() represents the logarithmic function; Σ represents the summation symbol;

[0083] A preset mutual information value threshold is obtained, and historical clinical data corresponding to a mutual information value greater than or equal to the preset mutual information value threshold is selected as the historical clinical feature.

[0084] It is understandable that the historical clinical data and the historical probability of postoperative delirium can be preprocessed first to ensure the quality and consistency of the data. The corresponding preprocessing steps include data cleaning, data conversion, and data coding, which can improve the effectiveness and accuracy of subsequent analysis. Specifically, data cleaning includes removing missing values ​​and correcting outliers; data conversion includes standardization and normalization; data coding includes converting categorical variables into numerical form.

[0085] Based on the feature selection algorithm, the mutual information value between the historical clinical data and the historical probability of postoperative delirium was calculated. Mutual information is a statistical measure that measures the amount of shared information between two variables.

[0086] In order to screen out historical clinical features that contribute significantly to the prediction of postoperative delirium, a preset mutual information value threshold can be set. By comparing the mutual information value of each feature with the threshold, features with mutual information values ​​greater than or equal to the threshold can be selected as the final historical clinical features.

[0087] Through the above method, historical clinical characteristics can be effectively extracted from historical clinical data, thereby reducing the complexity of the subsequently generated postoperative delirium risk prediction model and improving the accuracy, generalization ability and interpretability of the prediction model.

[0088] The method based on the machine learning algorithm and the historical clinical characteristics and the historical probability of postoperative delirium occurrence constructs a corresponding postoperative delirium risk prediction model, which specifically includes:

[0089] Dividing the historical clinical characteristics into training clinical characteristics and testing clinical characteristics, and dividing the historical postoperative delirium occurrence probability into training postoperative delirium occurrence probability and testing postoperative delirium occurrence probability;

[0090] Based on the support vector machine, the postoperative delirium risk prediction model is constructed according to the training clinical characteristics and the training postoperative delirium occurrence probability, and the corresponding objective function is:

[0091]

[0092] In the formula, max represents the maximum value operation; w represents the weight vector of the objective function corresponding to the postoperative delirium risk prediction model; ||w|| 2 represents the square of the modulus of the weight vector; C represents the regularization parameter of the objective function; m represents the total number of training data sets corresponding to the training clinical features and the training probability of postoperative delirium; ξ i represents the slack variable corresponding to the i-th training data set; Σ represents the summation symbol;

[0093] The constraints corresponding to the objective function are:

[0094]

[0095] In the formula, st represents the constraint condition; y′ i represents the probability of postoperative delirium during training; w represents the weight vector of the objective function corresponding to the postoperative delirium risk prediction model; x′ i represents the training clinical characteristics; b represents the intercept term of the objective function corresponding to the postoperative delirium risk prediction model; ξ irepresents the slack variable corresponding to the I-th training data group;

[0096] Based on the constraints, the objective function is maximized and the postoperative delirium risk prediction model is trained.

[0097] After obtaining the postoperative delirium risk prediction model, the prediction effect of the postoperative delirium risk prediction model is evaluated based on the tested clinical characteristics and the tested probability of postoperative delirium occurrence, and the postoperative delirium risk prediction model is optimized according to the predicted effect, specifically including:

[0098] Obtaining a confusion matrix corresponding to the tested clinical characteristics and the tested probability of postoperative delirium occurrence, and evaluating the prediction effect of the postoperative delirium risk prediction model according to the confusion matrix;

[0099] like Figure 2 As shown, the elements of the confusion matrix include true positive examples, false positive examples, true negative examples and false negative examples, wherein the true positive example is the correct identification of postoperative delirium occurrence; the false positive example is the wrong identification of postoperative delirium occurrence; the true negative example is the correct identification of postoperative delirium non-occurrence; the false negative example is the wrong identification of postoperative delirium non-occurrence;

[0100] Based on the confusion matrix, the prediction indicators corresponding to the postoperative delirium risk prediction model are calculated, wherein the prediction indicators include accuracy, precision, recall, F1 score and false positive rate;

[0101] The calculation formula of the accuracy is:

[0102]

[0103] In the formula, ZQL represents the accuracy rate corresponding to the postoperative delirium risk prediction model; TP represents true positive examples, that is, correctly identifying the occurrence of postoperative delirium; TN represents true negative examples, that is, correctly identifying that postoperative delirium did not occur; FP represents false positive examples, that is, incorrectly identifying the occurrence of postoperative delirium; FN represents false negative examples, that is, incorrectly identifying that postoperative delirium did not occur;

[0104] The calculation formula of the accuracy is:

[0105]

[0106] In the formula, JQL represents the accuracy corresponding to the postoperative delirium risk prediction model; TP represents true positive, that is, correctly identifying the occurrence of postoperative delirium; FP represents false positive, that is, incorrectly identifying the occurrence of postoperative delirium;

[0107] The calculation formula of the recall rate is:

[0108]

[0109] In the formula, ZHL represents the recall rate corresponding to the postoperative delirium risk prediction model; TP represents a true positive example, that is, correctly identifying the occurrence of postoperative delirium; FN represents a false negative example, that is, incorrectly identifying that postoperative delirium did not occur;

[0110] The calculation formula of the F1 score is:

[0111]

[0112] In the formula, F1 represents the F1 score corresponding to the postoperative delirium risk prediction model; JQL represents the precision corresponding to the postoperative delirium risk prediction model; ZHL represents the recall rate corresponding to the postoperative delirium risk prediction model;

[0113] The calculation formula of the false positive rate is:

[0114]

[0115] In the formula, JZL represents the false positive rate corresponding to the postoperative delirium risk prediction model; FP represents the false positive example, that is, the wrong identification of postoperative delirium; TN represents the true negative example, that is, the correct identification of postoperative delirium not occurring;

[0116] The prediction indicators are analyzed, and the precision and recall are weighed to evaluate the prediction effect of the postoperative delirium risk prediction model, and the postoperative delirium risk prediction model is optimized according to the prediction effect.

[0117] The determining whether intervention treatment is required for the patient end according to the target postoperative delirium occurrence probability specifically includes:

[0118] If intervention treatment is required for the patient end, the patient end is divided into a low-risk patient end, a medium-risk patient end and a high-risk patient end according to the target probability of postoperative delirium;

[0119] For the patients at different risk levels, formulate corresponding personalized intervention treatment plans, and perform intervention treatment on the patients according to the personalized intervention treatment plans;

[0120] The patient's intervention treatment process is monitored in real time, and the corresponding intervention treatment effect is evaluated, and the personalized intervention treatment plan is adjusted accordingly according to the evaluation results.

[0121] In practical applications, the probability of postoperative delirium in patients can be accurately quantified, and then it can be determined whether appropriate intervention treatment is needed for the patient.

[0122] Specifically, if Figure 3As shown in the figure, firstly, the patient side can be divided into low-risk patient side, medium-risk patient side and high-risk patient side according to the probability of postoperative delirium. It should be noted that the classification process is not only based on the probability of postoperative delirium, but also takes into account multiple factors such as the patient's age, underlying disease, type of surgery, preoperative psychological state and previous history of delirium, so as to ensure the accuracy and rationality of the classification results.

[0123] For patients with different risk levels, a highly personalized intervention treatment strategy can be adopted to customize a comprehensive treatment plan for each patient, including drug therapy, non-drug therapy, nutritional support, and psychological intervention. By customizing a personalized intervention treatment plan, the incidence of postoperative delirium can be minimized.

[0124] During the implementation of intervention treatment, wearable devices, remote monitoring systems, etc. can be used to monitor the patient's physiological indicators, cognitive functions, and emotional states in real time, so that potential problems can be discovered and dealt with in a timely manner, helping to ensure the effectiveness and safety of intervention measures.

[0125] Furthermore, the effect of intervention therapy can be comprehensively evaluated, including the recovery of patients' cognitive functions, improvement of delirium symptoms, and improvement of quality of life. Based on the evaluation results, personalized intervention therapy plans can be adjusted in a timely manner, including adjusting drug dosages, adding or replacing non-drug therapies, etc., to ensure that the treatment plan always meets the actual needs of patients and accelerates the patient's recovery process.

[0126] Through the above methods, by formulating personalized intervention treatment plans, we can effectively improve the prevention and treatment level of postoperative delirium in patients while ensuring patient safety.

[0127] Embodiment 2

[0128] like Figure 4 As shown, an early warning system for delirium in patients after abdominal surgery, the system comprising:

[0129] An acquisition module is used to obtain historical clinical data corresponding to abdominal surgery and the historical probability of postoperative delirium;

[0130] An extraction module, configured to receive the historical clinical data and the historical probability of postoperative delirium occurrence, and extract historical clinical features from the historical clinical data based on a feature selection algorithm;

[0131] A construction module is used to construct a corresponding postoperative delirium risk prediction model based on a machine learning algorithm according to the historical clinical characteristics and the historical probability of postoperative delirium occurrence;

[0132] An input module, used to obtain target clinical data corresponding to the patient side, extract corresponding target clinical features and input them into the postoperative delirium risk prediction model, and calculate the target postoperative delirium occurrence probability corresponding to the patient side;

[0133] The judgment module is used to judge whether intervention treatment is needed for the patient according to the target probability of postoperative delirium.

[0134] Through the introduction of the above embodiments, the present invention uses an early warning method and system for postoperative delirium in abdominal surgery patients, by obtaining historical clinical data and historical postoperative delirium occurrence probability corresponding to abdominal surgery; receiving historical clinical data and historical postoperative delirium occurrence probability, extracting historical clinical features in historical clinical data based on feature selection algorithm; constructing a corresponding postoperative delirium risk prediction model based on historical clinical features and historical postoperative delirium occurrence probability based on machine learning algorithm; obtaining target clinical data corresponding to the patient end, extracting corresponding target clinical features and inputting them into the postoperative delirium risk prediction model, and calculating the target postoperative delirium occurrence probability corresponding to the patient end; judging whether intervention treatment is needed on the patient end according to the target postoperative delirium occurrence probability. The present invention can predict the risk of postoperative delirium in patients before and during surgery, so as to take intervention measures early; at the same time, according to the actual condition of the patient, a corresponding personalized intervention treatment plan can be formulated to improve the treatment effect; and the present invention can reduce the occurrence of postoperative delirium in patients through early warning and intervention, reduce medical costs and hospitalization time, and thus help the patient's postoperative recovery and improve the quality of life.

[0135] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0136] Those skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0137] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity 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, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

Claims

1. A method for early warning of delirium in patients after abdominal surgery, characterized in that: The method comprises: Obtain historical clinical data corresponding to abdominal surgery and historical postoperative delirium incidence; Receiving the historical clinical data and the historical probability of postoperative delirium occurrence, and extracting historical clinical features from the historical clinical data based on a feature selection algorithm; Based on the machine learning algorithm, a corresponding postoperative delirium risk prediction model is constructed according to the historical clinical characteristics and the historical probability of postoperative delirium occurrence; Obtaining target clinical data corresponding to the patient side, extracting corresponding target clinical features and inputting them into the postoperative delirium risk prediction model, and calculating the target postoperative delirium occurrence probability corresponding to the patient side; Whether intervention treatment is required for the patient is determined based on the target probability of postoperative delirium occurrence.

2. The method for early warning of delirium in patients after abdominal surgery according to claim 1, characterized in that: The historical clinical data include the patient's age, gender, past medical history, medication history, preoperative cognitive function assessment, frailty index, abdominal surgery time, anesthesia time, anesthetic drugs, anesthesia methods, analgesia and sedation history and infection status.

3. The method for early warning of delirium in patients after abdominal surgery according to claim 1, characterized in that: The receiving of the historical clinical data and the historical probability of postoperative delirium occurrence, and extracting historical clinical features from the historical clinical data based on a feature selection algorithm, specifically includes: The historical clinical data and the historical probability of postoperative delirium were preprocessed; Based on the feature selection algorithm, the historical clinical data is used as the variable X, and the historical probability of postoperative delirium is used as the variable Y. The mutual information value I (X, Y) between the historical clinical data and the historical probability of postoperative delirium is calculated. The calculation formula of the mutual information value is: Where, I(X, Y) represents the mutual information value between historical clinical data and historical postoperative delirium occurrence probability; X represents historical clinical data; Y represents historical postoperative delirium occurrence probability; x represents the value of historical clinical data; y represents the value of historical postoperative delirium occurrence probability; p(x, y) represents the joint probability distribution of X and Y taking x and y at the same time; p(x) represents the marginal probability distribution of X taking x; p(y) represents the marginal probability distribution of Y taking y; log() represents the logarithmic function; Σ represents the summation symbol; A preset mutual information value threshold is obtained, and historical clinical data corresponding to a mutual information value greater than or equal to the preset mutual information value threshold is selected as the historical clinical feature.

4. The method for early warning of delirium in patients after abdominal surgery according to claim 1, characterized in that: The method based on the machine learning algorithm and the historical clinical characteristics and the historical probability of postoperative delirium occurrence constructs a corresponding postoperative delirium risk prediction model, which specifically includes: Dividing the historical clinical characteristics into training clinical characteristics and testing clinical characteristics, and dividing the historical postoperative delirium occurrence probability into training postoperative delirium occurrence probability and testing postoperative delirium occurrence probability; Based on the support vector machine, the postoperative delirium risk prediction model is constructed according to the training clinical characteristics and the training postoperative delirium occurrence probability, and the corresponding objective function is: In the formula, max represents the maximum value operation; w represents the weight vector of the objective function corresponding to the postoperative delirium risk prediction model; ||w|| 2 represents the square of the modulus of the weight vector; C represents the regularization parameter of the objective function; m represents the total number of training data sets corresponding to the training clinical features and the training probability of postoperative delirium; ξ i represents the slack variable corresponding to the i-th training data set; Σ represents the summation symbol; The constraints corresponding to the objective function are: In the formula, st represents the constraint condition; y′ i represents the probability of postoperative delirium during training; w represents the weight vector of the objective function corresponding to the postoperative delirium risk prediction model; x′ i represents the training clinical characteristics; b represents the intercept term of the objective function corresponding to the postoperative delirium risk prediction model; ξ i represents the slack variable corresponding to the i-th training data group; Based on the constraints, the objective function is maximized and the postoperative delirium risk prediction model is trained.

5. The method for early warning of delirium in patients after abdominal surgery according to claim 4, characterized in that: After obtaining the postoperative delirium risk prediction model, the prediction effect of the postoperative delirium risk prediction model is evaluated based on the tested clinical characteristics and the tested probability of postoperative delirium occurrence, and the postoperative delirium risk prediction model is optimized according to the predicted effect, specifically including: Obtaining a confusion matrix corresponding to the tested clinical characteristics and the tested probability of postoperative delirium occurrence, and evaluating the prediction effect of the postoperative delirium risk prediction model according to the confusion matrix; The elements of the confusion matrix include true positive examples, false positive examples, true negative examples and false negative examples, wherein the true positive example is the correct identification of postoperative delirium occurrence; the false positive example is the wrong identification of postoperative delirium occurrence; the true negative example is the correct identification of postoperative delirium non-occurrence; the false negative example is the wrong identification of postoperative delirium non-occurrence; Based on the confusion matrix, the prediction indicators corresponding to the postoperative delirium risk prediction model are calculated, wherein the prediction indicators include accuracy, precision, recall, F1 score and false positive rate; The calculation formula of the accuracy is: In the formula, ZQL represents the accuracy rate corresponding to the postoperative delirium risk prediction model; TP represents true positive examples, that is, correctly identifying the occurrence of postoperative delirium; TN represents true negative examples, that is, correctly identifying that postoperative delirium did not occur; FP represents false positive examples, that is, incorrectly identifying the occurrence of postoperative delirium; FN represents false negative examples, that is, incorrectly identifying that postoperative delirium did not occur; The calculation formula of the accuracy is: In the formula, JQL represents the accuracy corresponding to the postoperative delirium risk prediction model; TP represents true positive, that is, correctly identifying the occurrence of postoperative delirium; FP represents false positive, that is, incorrectly identifying the occurrence of postoperative delirium; The calculation formula of the recall rate is: In the formula, ZHL represents the recall rate corresponding to the postoperative delirium risk prediction model; TP represents a true positive example, that is, correctly identifying the occurrence of postoperative delirium; FN represents a false negative example, that is, incorrectly identifying that postoperative delirium did not occur; The calculation formula of the F1 score is: In the formula, F1 represents the F1 score corresponding to the postoperative delirium risk prediction model; JQL represents the precision corresponding to the postoperative delirium risk prediction model; ZHL represents the recall rate corresponding to the postoperative delirium risk prediction model; The calculation formula of the false positive rate is: In the formula, JZL represents the false positive rate corresponding to the postoperative delirium risk prediction model; FP represents the false positive example, that is, the wrong identification of postoperative delirium; TN represents the true negative example, that is, the correct identification of postoperative delirium not occurring; The prediction indicators are analyzed, and the precision and recall are weighed to evaluate the prediction effect of the postoperative delirium risk prediction model, and the postoperative delirium risk prediction model is optimized according to the prediction effect.

6. The method for early warning of delirium in patients after abdominal surgery according to claim 1, characterized in that: The determining whether intervention treatment is required for the patient end according to the target postoperative delirium occurrence probability specifically includes: If intervention treatment is required for the patient end, the patient end is divided into a low-risk patient end, a medium-risk patient end and a high-risk patient end according to the target probability of postoperative delirium; For the patients at different risk levels, formulate corresponding personalized intervention treatment plans, and perform intervention treatment on the patients according to the personalized intervention treatment plans; The patient's intervention treatment process is monitored in real time, and the corresponding intervention treatment effect is evaluated, and the personalized intervention treatment plan is adjusted accordingly according to the evaluation results.

7. An early warning system for delirium in patients after abdominal surgery, applied to an early warning method for delirium in patients after abdominal surgery as claimed in any one of claims 1 to 6, the system comprising: An acquisition module is used to obtain historical clinical data corresponding to abdominal surgery and the historical probability of postoperative delirium; An extraction module, configured to receive the historical clinical data and the historical probability of postoperative delirium occurrence, and extract historical clinical features from the historical clinical data based on a feature selection algorithm; A construction module is used to construct a corresponding postoperative delirium risk prediction model based on a machine learning algorithm according to the historical clinical characteristics and the historical probability of postoperative delirium occurrence; An input module, used to obtain target clinical data corresponding to the patient side, extract corresponding target clinical features and input them into the postoperative delirium risk prediction model, and calculate the target postoperative delirium occurrence probability corresponding to the patient side; The judgment module is used to judge whether intervention treatment is needed for the patient according to the target probability of postoperative delirium.

Images