Postoperative pulmonary complication risk prediction and auxiliary decision-making system

Abstract

The invention relates to the technical field of medical care informatics, in particular to a postoperative pulmonary complication risk prediction and auxiliary decision making system. The system comprises a data acquisition module used for acquiring admission information, preoperative information and postoperative information of a patient; the model training module is used for creating and training an admission prediction model, a preoperative prediction model and a postoperative prediction model; the feature processing module is used for extracting a first feature, a second feature and a third feature from the patient information, the preoperative information and the postoperative information and preprocessing the features; the risk prediction module is used for acquiring output results and corresponding risk factors of the admission prediction model, the preoperative prediction model and the postoperative prediction model; and the intervention means recommendation module is used for acquiring preoperative intervention measures, intraoperative intervention measures and postoperative intervention measures of each predicted pulmonary complication. According to the method, the problems of inaccurate prediction of postoperative pulmonary complications and accurate acquisition of intervention measures are solved, and the prediction accuracy and the accuracy of the intervention measures are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical care informatics, and more specifically to a postoperative pulmonary complication risk prediction and decision-making support system. Background Art

[0002] Pulmonary complications, especially pulmonary infection, are still the main cause of prolonged postoperative hospitalization and death in thoracic surgery patients. Perioperative lung protection is an important part of accelerated recovery after surgery. Strengthening perioperative lung protection can significantly reduce the occurrence of pulmonary complications and reduce the risk of death. Predicting postoperative pulmonary complications may optimize treatment measures for some patients and promote the rational use of medical resources, thereby strengthening perioperative lung protection and promoting the improvement of medical quality. Similar prior art includes a patent application with publication number CN112017783A, which proposes a prediction model for postoperative pulmonary infection after cardiac surgery and its construction method. In the collection of various clinical data of perioperative patients undergoing cardiac surgery, evaluation indicators related to postoperative pulmonary infection after cardiac surgery are screened out, and the evaluation indicators are incorporated into the Logistic regression model to analyze and calculate the regression coefficients of the risk factors for postoperative pulmonary infection after cardiac surgery, and then the risk score is obtained by the regression coefficients of each risk factor. Finally, the risk score is combined with the pulmonary infection risk prediction function to calculate the probability value of pulmonary infection. Thus, a risk prediction model based on a scoring system can be established. Compared with existing foreign models, it shows better risk prediction ability, can achieve the purpose of early screening of high-risk patients with postoperative pulmonary infection, and play a role in early prevention, early detection, and early treatment, thereby reducing the incidence of postoperative pulmonary infection after cardiac surgery. In addition, similar prior art includes a patent application with publication number US20230016569A1, which proposes a method based on chest X-ray to predict whether a patient suffers from any comorbidity in one or more advanced comorbidity categories. Chest X-ray is a low-cost, minimally invasive source of visual information that can accurately predict whether a subject has diabetes with chronic complications, morbid obesity, congestive heart failure, specific arrhythmias, vascular disease or chronic obstructive pulmonary disease, as well as other advanced comorbidity categories. Both of the above patent applications solve the problem of complication prediction, but do not make predictions for different stages and the prediction accuracy is not high enough. There are no targeted intervention measures for each stage, and there are no effective reference measures for medical staff. The effect of reducing or alleviating complications, especially pulmonary complications, is poor. Summary of the invention

[0003] In order to better solve the above problems, the present invention provides a postoperative pulmonary complication risk prediction and decision-making support system, the system comprising:

[0004] A data acquisition module, used to acquire the admission information of the patient when the patient is admitted to the hospital, acquire the preoperative information of the patient within a first preset time period before the patient undergoes surgery, and acquire the postoperative information of the patient within a second preset time period after the patient undergoes surgery;

[0005] A model training module, used to create and train an admission prediction model, a preoperative prediction model, and a postoperative prediction model, and to obtain a first relevant factor, a second relevant factor, and a third relevant factor based on the trained admission prediction model, the preoperative prediction model, and the postoperative prediction model;

[0006] a feature processing module, configured to extract a first feature, a second feature, and a third feature from the admission information, the preoperative information, and the postoperative information respectively according to the first relevant factor, the second relevant factor, and the third relevant factor, and perform preprocessing;

[0007] a risk prediction module, configured to input the first feature, the second feature, and the third feature into the admission prediction model, the preoperative prediction model, and the postoperative prediction model, respectively, and obtain a first output result, a second output result, and a third output result and a first risk factor, a second risk factor, and a third risk factor;

[0008] An intervention recommendation module is used to obtain each predicted pulmonary complication and corresponding preoperative intervention measures, intraoperative intervention measures and postoperative intervention measures based on the first risk factor, the second risk factor, the third risk factor, the first output result, the second output result and the third output result.

[0009] As a preferred technical solution, the system also includes: a result feedback module, which is used to send each predicted pulmonary complication and the predicted score corresponding to the predicted pulmonary complication, the preoperative intervention measures, the intraoperative intervention measures and the postoperative intervention measures as output results to the medical terminal.

[0010] As a preferred technical solution, the patient's admission information includes the patient's basic information, chief complaint, past medical history, personal medical history, complications, and admission diagnosis information; the preoperative information includes the patient's examination information obtained within the first preset time period before the operation and the patient's admission information; the postoperative information includes the operation duration, blood loss, and anesthetic medication information obtained within the second preset time period after the operation.

[0011] As a preferred technical solution, the model training module is configured as follows: based on the historical admission information, historical preoperative information, historical postoperative information and corresponding multiple historical pulmonary complications and corresponding scores of other patients undergoing thoracic surgery stored in the storage unit, the historical admission information, each of the historical pulmonary complications and the corresponding scores are used as the first training data, the historical preoperative information, each of the historical pulmonary complications and the corresponding scores are used as the second training data, and the historical postoperative information, each of the historical pulmonary complications and the corresponding scores are used as the third training data, and based on the first training data, the second training data and the third training data, the admission prediction model, the preoperative prediction model and the postoperative prediction model are trained respectively, wherein the admission prediction model, the preoperative prediction model and the postoperative prediction model are all neural network models; the first related factor, the second related factor and the third related factor whose weights are greater than a set threshold are extracted from the output layers of the trained admission prediction model, the preoperative prediction model and the postoperative prediction model, wherein the first related factor, the second related factor and the third related factor respectively contain multiple factors.

[0012] As a preferred technical solution, the risk prediction module is configured as follows:

[0013] The first feature is input into the hospitalization prediction model, and a first output result is obtained, wherein the first output result includes N1 first pulmonary complications and a first score of each first pulmonary complication, and a contribution value of each feature in the first feature corresponding to each first score is obtained, and the N features with the largest contribution value and that can be intervened are used as the first risk factors corresponding to the first pulmonary complication. Similarly, the first risk factor of each first pulmonary complication is obtained, wherein the value ranges of N and N1 are both positive integers greater than or equal to 1;

[0014] The intervention recommendation module is used to analyze the first causal relationship between each of the first pulmonary complications and the corresponding first risk factors, and obtain the preoperative intervention measures based on the first causal relationship and the first score corresponding to the first pulmonary complication.

[0015] As a preferred technical solution, the risk prediction module is further configured to: after the implementation of the preoperative intervention measures and before the thoracic surgery, within the first preset time period, input the second feature into the preoperative prediction model, and obtain a second output result, wherein the second output result includes N2 second pulmonary complications and corresponding second scores, and according to the contribution value of each feature of each second feature to each second score, obtain the second risk factor corresponding to each second pulmonary complication;

[0016] The intervention recommendation module is used to analyze the second causal relationship between each of the second pulmonary complication and the corresponding second risk factor, correspond the first pulmonary complication and the second pulmonary complication to the same disease, and use the corresponding same disease as the first target complication, obtain the first score and the second score of each of the first target complication, and calculate the first difference between the first score and the second score, weight the first score and the second score to obtain the first target score, and obtain intraoperative intervention measures based on the second causal relationship corresponding to each of the first target complication and the first target score, wherein the weight of the first score is less than the weight of the second score, and the larger the first difference, the smaller the weight of the first score, and N1 of the first pulmonary complications include N2 of the second pulmonary complications.

[0017] As a preferred technical solution, the intervention means recommendation module is used to input the third feature into the postoperative prediction model within the second preset time period after surgery, and obtain the third output result, wherein the third output result includes N3 third pulmonary complications and corresponding third scores, and obtain the third risk factor corresponding to each third pulmonary complication according to the contribution value of each feature in the third feature to each third score, wherein N3 is less than or equal to N2;

[0018] The intervention means recommendation module is used to analyze the third causal relationship between each of the third pulmonary complication and the third risk factor, and obtain the postoperative intervention measures based on the third causal relationship, take the third pulmonary complication as the second target complication, wherein the first target complication includes the second target complication, calculate the second difference between the first target score and the third score corresponding to each of the second target complication, and set the weight of the first target score according to the second difference, and weight the first target score and the third score to obtain the second target score corresponding to the second target complication, and take the second target complication as the predicted pulmonary complication, and take the second target score as the predicted score of the predicted pulmonary complication, wherein the weight of the third score is greater than the weight of the first target score, and the larger the second difference is, the smaller the weight of the first target score is, and the smaller the second difference is, the larger the weight of the first target score is.

[0019] As a preferred technical solution, the first feature, the second feature and the third feature respectively include multiple features and corresponding feature values.

[0020] The present invention also provides a method for predicting the risk of postoperative pulmonary complications and assisting decision-making, which is implemented based on the above-mentioned system and includes:

[0021] Step S1: creating and training an admission prediction model, a preoperative prediction model, and a postoperative prediction model through a model training module, and acquiring a first related factor, a second related factor, and a third related factor based on the trained admission prediction model, the preoperative prediction model, and the postoperative prediction model;

[0022] Step S2: obtaining the admission information of the patient when he / she is admitted to the hospital through the data acquisition module, obtaining a first feature from the admission information according to the first related factor, inputting the first feature into the admission prediction model, and obtaining a first output result, and obtaining a first risk factor based on the first output result, and obtaining a preoperative intervention measure for each first pulmonary complication in the first output result based on the first risk factor and the first output result;

[0023] Step S3: obtaining the preoperative information of the patient at the time of admission through the data acquisition module, obtaining a second feature from the preoperative information according to the second related factor, inputting the second feature into the preoperative prediction model, and obtaining a second output result, and obtaining a second risk factor based on the second output result, and obtaining a preoperative intervention measure for each second pulmonary complication in the second output result based on the second risk factor, the second output result and the first output result;

[0024] Step S4: Obtain the postoperative information of the patient at the time of admission through the data acquisition module, obtain a third feature from the preoperative information according to the third related factor, input the third feature into the preoperative prediction model, and obtain a third output result, and obtain a third risk factor based on the third output result, and obtain postoperative intervention measures for each third pulmonary complication in the third output result based on the third risk factor, the second output result and the third output result, wherein the third pulmonary complication is the predicted pulmonary complication.

[0025] The present invention also provides a computer storage medium, wherein the storage medium stores program instructions, wherein when the program instructions are executed, the device where the storage medium is located is controlled to execute the above method.

[0026] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0027] The present invention obtains the admission information, preoperative information and postoperative information of the above-mentioned patient when admitted to the hospital, and extracts the above-mentioned first feature, second feature and third feature from the above-mentioned admission information, preoperative information and postoperative information based on the above-mentioned first related factors, the above-mentioned second related factors and the above-mentioned third related factors extracted from the trained above-mentioned admission model, the above-mentioned preoperative model and the above-mentioned postoperative model, respectively, wherein the above-mentioned first related factors, the above-mentioned second related factors and the above-mentioned third related factors are the features with larger weights in the output layers of the above-mentioned admission model, the above-mentioned preoperative model and the above-mentioned postoperative model, respectively, thereby laying a foundation for obtaining more accurate the above-mentioned first feature, second feature and third feature, and then obtaining more accurate prediction of pulmonary complications, by inputting the above-mentioned first feature into the above-mentioned admission prediction model, and obtaining a first output result, and according to each first pulmonary complication and The above-mentioned first risk factor is obtained corresponding to the first score, and the above-mentioned preoperative intervention measures and the corresponding intervention degree are obtained based on each of the above-mentioned first pulmonary complications and the corresponding first risk factor and the first score. The above-mentioned second feature is also input into the above-mentioned preoperative prediction model, and the second output result and the above-mentioned second risk factor are obtained. The intraoperative intervention measures and the corresponding intervention degree are obtained through the above-mentioned first output result, the above-mentioned second output result and the second risk factor. Similarly, the postoperative intervention measures and the corresponding intervention degree are also obtained. Through the mutual coordination of the above-mentioned technical solutions, the effects of the intervention measures at each stage are fully considered and the pulmonary complications and corresponding scores at different stages are recalculated. Not only can the final multiple postoperative pulmonary complications and the scores of each pulmonary complication be predicted more accurately, but also corresponding intervention measures can be given in a more targeted manner, thereby reducing or alleviating the occurrence of pulmonary complications. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a structural diagram of a postoperative pulmonary complication risk prediction and decision-making support system of the present invention;

[0029] Figure 2 This is a flow chart of a postoperative pulmonary complication risk prediction and decision-making assistance method of the present invention. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0031] The present invention provides a postoperative pulmonary complication risk prediction and decision-making support system, such as Figure 1 As shown, including:

[0032] A data acquisition module, used to acquire the admission information of the patient when the patient is admitted to the hospital, acquire the preoperative information of the patient within a first preset time period before the patient undergoes surgery, and acquire the postoperative information of the patient within a second preset time period after the patient undergoes surgery;

[0033] A model training module, used to create and train an admission prediction model, a preoperative prediction model, and a postoperative prediction model, and to obtain a first relevant factor, a second relevant factor, and a third relevant factor based on the trained admission prediction model, the preoperative prediction model, and the postoperative prediction model;

[0034] a feature processing module, configured to extract a first feature, a second feature, and a third feature from the admission information, the preoperative information, and the postoperative information respectively according to the first relevant factor, the second relevant factor, and the third relevant factor, and perform preprocessing;

[0035] a risk prediction module, configured to input the first feature, the second feature, and the third feature into the admission prediction model, the preoperative prediction model, and the postoperative prediction model, respectively, and obtain a first output result, a second output result, and a third output result and a first risk factor, a second risk factor, and a third risk factor;

[0036] An intervention recommendation module is used to obtain each predicted pulmonary complication and corresponding preoperative intervention measures, intraoperative intervention measures and postoperative intervention measures based on the first risk factor, the second risk factor, the third risk factor, the first output result, the second output result and the third output result.

[0037] Specifically, by obtaining the admission information, preoperative information and postoperative information of the above-mentioned patient when admitted to the hospital, and extracting the above-mentioned first feature, second feature and third feature from the above-mentioned admission information, preoperative information and postoperative information based on the above-mentioned first related factors, the above-mentioned second related factors and the above-mentioned third related factors extracted from the trained above-mentioned admission model, the above-mentioned preoperative model and the above-mentioned postoperative model, respectively, wherein the above-mentioned first related factors, the above-mentioned second related factors and the above-mentioned third related factors are respectively the features with larger weights in the output layers of the above-mentioned admission model, the above-mentioned preoperative model and the above-mentioned postoperative model, thereby laying a foundation for obtaining more accurate the above-mentioned first feature, the second feature and the third feature, and then obtaining a more accurate prediction of pulmonary complications, by inputting the above-mentioned first feature into the above-mentioned admission prediction model, and obtaining the first output result, and according to each first pulmonary complication and the corresponding The first score obtains the above-mentioned first risk factor, and based on each of the above-mentioned first pulmonary complications and the corresponding above-mentioned first risk factors and the first score, the above-mentioned preoperative intervention measures and the corresponding intervention levels are obtained, and the above-mentioned second feature is also input into the above-mentioned preoperative prediction model, and the second output result and the above-mentioned second risk factor are obtained. The intraoperative intervention measures and the corresponding intervention levels are obtained through the above-mentioned first output result, the above-mentioned second output result and the second risk factor. Similarly, the postoperative intervention measures and the corresponding intervention levels are also obtained. Through the mutual coordination of the above-mentioned technical solutions, the effects of the intervention measures at each stage are fully considered and the predicted pulmonary complications and corresponding prediction scores of the patients at different stages are recalculated. Not only can the final multiple postoperative pulmonary complications and the scores of each pulmonary complication be predicted more accurately, but also corresponding intervention measures can be given in a more targeted manner, thereby reducing or alleviating the occurrence of pulmonary complications.

[0038] Furthermore, the system also includes: a result feedback module, which is used to send each predicted pulmonary complication and the predicted score corresponding to the predicted pulmonary complication, the preoperative intervention measures, the intraoperative intervention measures and the postoperative intervention measures as output results to the medical terminal.

[0039] Specifically, by sending each of the above-mentioned predicted pulmonary complications and the corresponding prediction scores to the above-mentioned medical terminal, medical staff can understand the above-mentioned predicted pulmonary complications and the corresponding prediction scores, wherein the above-mentioned predicted pulmonary complications include the first pulmonary complication, the second pulmonary complication and the third pulmonary complication, and the above-mentioned preoperative intervention measures, the above-mentioned intraoperative intervention measures and the above-mentioned postoperative intervention measures corresponding to each of the above-mentioned predicted pulmonary complications are also sent to the above-mentioned medical terminal as output results for the reference of medical staff, thereby reducing or alleviating the occurrence of pulmonary complications.

[0040] Furthermore, the patient's admission information includes the patient's basic information, chief complaint, past medical history, personal medical history, complications, and admission diagnosis information; the preoperative information includes the patient's examination information obtained within the first preset time period before the operation and the patient's admission information; the postoperative information includes the operation duration, blood loss, and anesthetic medication information obtained within the second preset time period after the operation.

[0041] Further, the model training module is configured as follows: based on the historical admission information, historical preoperative information, historical postoperative information and corresponding multiple historical pulmonary complications and corresponding scores of other patients undergoing thoracic surgery stored in the storage unit, the historical admission information, each of the historical pulmonary complications and the corresponding scores are used as the first training data, the historical preoperative information, each of the historical pulmonary complications and the corresponding scores are used as the second training data, the historical postoperative information, each of the historical pulmonary complications and the corresponding scores are used as the third training data, and based on the first training data, the second training data and the third training data, the admission prediction model, the preoperative prediction model and the postoperative prediction model are trained respectively, wherein the admission prediction model, the preoperative prediction model and the postoperative prediction model are all neural network models; the first related factor, the second related factor and the third related factor whose weights are greater than a set threshold are extracted from the output layer of the trained admission prediction model, the preoperative prediction model and the postoperative prediction model, wherein each of the first related factor, the second related factor and the third related factor respectively contains multiple factors.

[0042] Specifically, the historical admission information, the historical preoperative information and the historical postoperative information of other patients undergoing thoracic surgery stored in the storage unit are the same as the admission information, preoperative information and postoperative information of the patient for whom postoperative pulmonary complications are to be predicted, and the historical admission information and each of the historical pulmonary complications and the corresponding scores are used as the first training data, and the first training data is used to train the admission prediction model. Since the admission prediction model is a neural network model, and since the historical admission information includes the patient's basic information, chief complaint, past medical history, personal medical history, complications, and admission diagnosis information, but not every content of the historical admission information is related to the historical pulmonary complications, the output layer of the admission prediction model extracts the information with a weight greater than the above. The key information for setting the threshold is the key information in the above-mentioned admission information related to each of the above-mentioned historical pulmonary complications, and the above-mentioned key information is used as the above-mentioned first related factor. Similarly, the above-mentioned historical preoperative information and each of the above-mentioned historical complications and the corresponding scores are used as the above-mentioned second training data, and the above-mentioned historical postoperative information and each of the above-mentioned historical complications and the corresponding scores are used as the above-mentioned third training data. The second training data and the above-mentioned third training data are used to train the above-mentioned preoperative prediction model and the above-mentioned postoperative prediction model respectively, and the above-mentioned second related factor and the above-mentioned third related factor are obtained in the same manner as the above-mentioned first related factor. Through the above-mentioned technical scheme, a foundation is laid for further obtaining accurate postoperative pulmonary complications through the above-mentioned admission prediction model, the above-mentioned preoperative prediction model and the above-mentioned postoperative prediction model.

[0043] Furthermore, the risk prediction module is configured as follows:

[0044] The first feature is input into the hospitalization prediction model, and a first output result is obtained, wherein the first output result includes N1 first pulmonary complications and a first score of each first pulmonary complication, and a contribution value of each feature in the first feature corresponding to each first score is obtained, and the N features with the largest contribution value and that can be intervened are used as the first risk factors corresponding to the first pulmonary complication. Similarly, the first risk factor of each first pulmonary complication is obtained, wherein the value ranges of N and N1 are both positive integers greater than or equal to 1;

[0045] The intervention recommendation module is used to analyze the first causal relationship between each of the first pulmonary complications and the corresponding first risk factors, and obtain the preoperative intervention measures based on the first causal relationship and the first score corresponding to the first pulmonary complication.

[0046] Specifically, since the above-mentioned first feature is the feature information related to the above-mentioned first pulmonary complication extracted from the above-mentioned admission information based on the above-mentioned first related factor, and the above-mentioned first feature is input into the above-mentioned admission prediction model to obtain the above-mentioned first output result, the above-mentioned first output result includes N1 different types of the above-mentioned first pulmonary complications, and the first score of each of the above-mentioned first pulmonary complications, and the higher the first score, the higher the probability of occurrence of the above-mentioned first pulmonary complication and the greater the severity. In order to minimize the occurrence of the above-mentioned first pulmonary complication through precise preoperative intervention measures before surgery, therefore, when the above-mentioned first output result is output through the above-mentioned admission prediction model, the contribution value of each feature of the above-mentioned first feature to the above-mentioned first score is output through the output layer. Since it is necessary to obtain precise preoperative intervention measures before surgery, Intervention measures, the above characteristics must be able to intervene, for example: age, medical history, although related to the above first pulmonary complication, but can not be intervened by preoperative intervention measures, so the above N characteristics with the largest contribution value and can be intervened are used as the first risk factors corresponding to the above first pulmonary complication, and the same method is used to obtain the first risk factor corresponding to each of the above pulmonary complication, and the first causal relationship between each of the above first pulmonary complication and the corresponding first risk factor is analyzed through the above intervention measure recommendation module, the above preoperative intervention measures should be obtained based on the above first causal relationship, and the intervention degree of the above preoperative intervention measures is obtained according to the first score of the above first pulmonary complication. Through the above technical scheme, the above preoperative intervention measures can be accurately obtained, thereby reducing the probability of occurrence of the above first pulmonary complication.

[0047] Further, the risk prediction module is further configured to: after the implementation of the preoperative intervention measures and before the thoracic surgery within the first preset time period, input the second feature into the preoperative prediction model, and obtain a second output result, wherein the second output result includes N2 second pulmonary complications and corresponding second scores, and according to the contribution value of each feature of each second feature to each second score, obtain the second risk factor corresponding to each second pulmonary complication;

[0048] The intervention recommendation module is used to analyze the second causal relationship between each of the second pulmonary complication and the corresponding second risk factor, correspond the first pulmonary complication and the second pulmonary complication to the same disease, and use the corresponding same disease as the first target complication, obtain the first score and the second score of each of the first target complication, and calculate the first difference between the first score and the second score, weight the first score and the second score to obtain the first target score, and obtain intraoperative intervention measures based on the second causal relationship corresponding to each of the first target complication and the first target score, wherein the weight of the first score is less than the weight of the second score, and the larger the first difference, the smaller the weight of the first score, and N1 of the first pulmonary complications include N2 of the second pulmonary complications.

[0049] Specifically, after the implementation of the above-mentioned preoperative intervention measures and before the above-mentioned thoracic surgery, the above-mentioned second feature is input into the above-mentioned preoperative prediction model, and the above-mentioned second output result is obtained, wherein the above-mentioned second output result includes multiple different types of the above-mentioned second pulmonary complications and the corresponding second scores, and based on the contribution value of each feature of the above-mentioned second feature to the above-mentioned second score, the above-mentioned second risk factor corresponding to each of the above-mentioned second pulmonary complications is obtained, wherein the method for obtaining the above-mentioned second risk factor is the same as the method for obtaining the above-mentioned first risk factor. Since the above-mentioned admission prediction model can predict the occurrence of the above-mentioned first pulmonary complication more accurately, and it has been intervened through more precise intervention measures, the occurrence of some of the above-mentioned first pulmonary complications or the reduction of the score will be avoided, that is, N1 of the above-mentioned first pulmonary complications include N2 of the above-mentioned second pulmonary complications. By analyzing the second causal relationship between each of the above-mentioned first target complications, that is, the above-mentioned second pulmonary complications and the corresponding second risk factors, the above-mentioned intraoperative intervention measures are obtained, and also according to the above-mentioned first score and the above-mentioned second score corresponding to the above-mentioned first target complication. The first difference between the two, and the weight of the first score is set according to the first difference. The first difference reflects the effect of the preoperative intervention measures. The larger the first difference is, the better the effect of the preoperative intervention measures is, that is, the smaller the weight of the first score is. Conversely, the worse the effect of the preoperative intervention measures is, that is, the larger the weight of the first score is. Since the corresponding time of the preoperative information is closer to the postoperative period, the second score corresponding to the second pulmonary complication is more accurate and smaller than the first score. Therefore, the weight of the second score is greater than the weight of the first score. The first target score is obtained by weighting the first score and the second score. The first target score reflects the probability of occurrence of the corresponding first target complication, and the intervention degree of the intraoperative intervention measures corresponding to the first target complication is set according to the first target score. For example, when drug intervention is needed, the drug dosage is adjusted within an appropriate range according to the intervention degree. Through the above technical solution, the score of each of the above first target complications can be accurately obtained, and accurate intraoperative intervention measures can be obtained for each of the above first target complications.

[0050] Further, the intervention means recommendation module is used to input the third feature into the postoperative prediction model within the second preset time period after surgery, and obtain the third output result, wherein the third output result includes N3 third pulmonary complications and corresponding third scores, and obtain the third risk factor corresponding to each third pulmonary complication according to the contribution value of each feature in the third feature to each third score, wherein N3 is less than or equal to N2;

[0051] The intervention means recommendation module is used to analyze the third causal relationship between each of the third pulmonary complication and the third risk factor, and obtain the postoperative intervention measures based on the third causal relationship, taking the third pulmonary complication as the second target complication, wherein the first target complication includes the second target complication, calculate the second difference between the first target score and the third score corresponding to each of the second target complications, and set the weight of the first target score according to the second difference, and weight the first target score and the third score to obtain the second target score corresponding to the second target complication, wherein the weight of the third score is greater than the weight of the first target score, and the larger the second difference is, the smaller the weight of the first target score is, and the smaller the second difference is, the larger the weight of the first target score is.

[0052] Specifically, the third feature is input into the postoperative prediction model within the second preset time period after the operation, and the third output result is obtained, and according to the contribution value of each feature in the third feature to the third score corresponding to each third pulmonary complication in the third output result, the third risk factor corresponding to each third pulmonary complication is obtained according to the method of obtaining the first risk factor, and the third causal relationship between each third pulmonary complication and the third risk factor is analyzed, and the postoperative intervention measures are obtained based on the third causal relationship. Since the postoperative information can best reflect the patient's postoperative status, the third score can more accurately reflect each third score. The probability and severity of the occurrence of the second target complication, and because the first target score of the second target complication is the score after the above-mentioned intraoperative intervention measures are taken and reflects the patient's postoperative pulmonary complications, the third score is more accurate than the first target score. Therefore, the weight of the third score is greater than the weight of the first target score. The weight setting of the first target score is the same as the weight setting method of the first score, which will not be repeated here. Through the above-mentioned technical solution, not only the above-mentioned postoperative intervention measures corresponding to the above-mentioned second target complications can be accurately obtained, but also the corresponding second target score can be accurately obtained, and then the accurate intervention degree of the above-mentioned postoperative intervention measures can be obtained.

[0053] The present invention also provides a method for predicting the risk of postoperative pulmonary complications and assisting decision making, the method is implemented based on the above system, such as Figure 2 As shown, the method includes:

[0054] Step S1: creating and training an admission prediction model, a preoperative prediction model, and a postoperative prediction model through a model training module, and acquiring a first related factor, a second related factor, and a third related factor based on the trained admission prediction model, the preoperative prediction model, and the postoperative prediction model;

[0055] Step S2: obtaining the admission information of the patient when he / she is admitted to the hospital through the data acquisition module, obtaining a first feature from the admission information according to the first related factor, inputting the first feature into the admission prediction model, and obtaining a first output result, and obtaining a first risk factor based on the first output result, and obtaining a preoperative intervention measure for each first pulmonary complication in the first output result based on the first risk factor and the first output result;

[0056] Step S3: obtaining the preoperative information of the patient at the time of admission through the data acquisition module, obtaining a second feature from the preoperative information according to the second related factor, inputting the second feature into the preoperative prediction model, and obtaining a second output result, and obtaining a second risk factor based on the second output result, and obtaining a preoperative intervention measure for each second pulmonary complication in the second output result based on the second risk factor, the second output result and the first output result;

[0057] Step S4: Obtain the postoperative information of the patient at the time of admission through the data acquisition module, obtain a third feature from the preoperative information according to the third related factor, input the third feature into the preoperative prediction model, and obtain a third output result, and obtain a third risk factor based on the third output result, and obtain postoperative intervention measures for each third pulmonary complication in the third output result based on the third risk factor, the second output result and the third output result, wherein the third pulmonary complication is the predicted pulmonary complication.

[0058] The present invention also provides a computer storage medium, wherein the storage medium stores program instructions, wherein when the program instructions are executed, the device where the storage medium is located is controlled to execute the above method.

[0059] In summary, the present invention obtains the admission information, preoperative information and postoperative information of the above-mentioned patient when admitted to the hospital, and extracts the above-mentioned first feature, second feature and third feature from the above-mentioned admission information, preoperative information and postoperative information based on the above-mentioned first related factors, the above-mentioned second related factors and the above-mentioned third related factors extracted from the trained above-mentioned admission model, the above-mentioned preoperative model and the above-mentioned postoperative model, respectively, thereby laying a foundation for obtaining more accurate the above-mentioned first feature, second feature and third feature, and then obtaining more accurate prediction of pulmonary complications, by inputting the above-mentioned first feature into the above-mentioned admission prediction model and obtaining the first output result, and obtaining the above-mentioned first risk factor according to each first pulmonary complication and the corresponding first score in the above-mentioned first output result, and based on each above-mentioned first pulmonary complication The above-mentioned preoperative intervention measures and the corresponding intervention degree are obtained according to the above-mentioned symptoms and the corresponding first risk factor and the first score. The above-mentioned second feature is also input into the above-mentioned preoperative prediction model, and the second output result and the above-mentioned second risk factor are obtained. The intraoperative intervention measures and the corresponding intervention degree are obtained through the above-mentioned first output result, the above-mentioned second output result and the second risk factor. Similarly, the postoperative intervention measures and the corresponding intervention degree are also obtained. Through the mutual coordination of the above-mentioned technical solutions, the effects of the intervention measures at each stage are fully considered and the pulmonary complications of the patients at different stages and the corresponding scores are recalculated. Not only can the final multiple postoperative pulmonary complications and the scores of each pulmonary complication be predicted more accurately, but also the corresponding intervention measures can be given more targeted, thereby reducing or alleviating the occurrence of pulmonary complications.

[0060] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0061] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

[0062] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A postoperative pulmonary complication risk prediction and decision support system, characterized in that: The system comprises: A data acquisition module, used to acquire the admission information of the patient when the patient is admitted to the hospital, acquire the preoperative information of the patient within a first preset time period before the patient undergoes surgery, and acquire the postoperative information of the patient within a second preset time period after the patient undergoes surgery; A model training module, used to create and train an admission prediction model, a preoperative prediction model, and a postoperative prediction model, and to obtain a first relevant factor, a second relevant factor, and a third relevant factor based on the trained admission prediction model, the preoperative prediction model, and the postoperative prediction model; a feature processing module, configured to extract a first feature, a second feature, and a third feature from the admission information, the preoperative information, and the postoperative information respectively according to the first relevant factor, the second relevant factor, and the third relevant factor, and perform preprocessing; a risk prediction module, configured to input the first feature, the second feature, and the third feature into the admission prediction model, the preoperative prediction model, and the postoperative prediction model, respectively, and obtain a first output result, a second output result, and a third output result and a first risk factor, a second risk factor, and a third risk factor; An intervention recommendation module is used to obtain each predicted pulmonary complication and corresponding preoperative intervention measures, intraoperative intervention measures and postoperative intervention measures based on the first risk factor, the second risk factor, the third risk factor, the first output result, the second output result and the third output result.

2. The system according to claim 1, characterized in that The system also includes: a result feedback module, which is used to send each predicted pulmonary complication and the predicted score corresponding to the predicted pulmonary complication, the preoperative intervention measure, the intraoperative intervention measure and the postoperative intervention measure as output results to the medical terminal.

3. The system according to claim 1, characterized in that The patient's admission information includes the patient's basic information, chief complaint, past medical history, personal medical history, complications, and admission diagnosis information; the preoperative information includes the patient's examination information obtained within the first preset time period before the operation and the patient's admission information; the postoperative information includes the operation duration, blood loss, and anesthetic medication information obtained within the second preset time period after the operation.

4. The system according to claim 1, characterized in that The model training module is configured as follows: based on the historical admission information, historical preoperative information, historical postoperative information and corresponding multiple historical pulmonary complications and corresponding scores of other patients undergoing thoracic surgery stored in the storage unit, the historical admission information, each of the historical pulmonary complications and the corresponding scores are used as the first training data, the historical preoperative information, each of the historical pulmonary complications and the corresponding scores are used as the second training data, and the historical postoperative information, each of the historical pulmonary complications and the corresponding scores are used as the third training data, and based on the first training data, the second training data and the third training data, the admission prediction model, the preoperative prediction model and the postoperative prediction model are trained respectively, wherein the admission prediction model, the preoperative prediction model and the postoperative prediction model are all neural network models; the first related factor, the second related factor and the third related factor whose weights are greater than a set threshold are extracted from the output layers of the trained admission prediction model, the preoperative prediction model and the postoperative prediction model, wherein the first related factor, the second related factor and the third related factor respectively contain multiple factors.

5. The system according to claim 1, characterized in that The risk prediction module is configured as follows: The first feature is input into the hospitalization prediction model, and a first output result is obtained, wherein the first output result includes N1 first pulmonary complications and a first score of each first pulmonary complication, and a contribution value of each feature in the first feature corresponding to each first score is obtained, and the N features with the largest contribution value and that can be intervened are used as the first risk factors corresponding to the first pulmonary complication. Similarly, the first risk factor of each first pulmonary complication is obtained, wherein the value ranges of N and N1 are both positive integers greater than or equal to 1; The intervention recommendation module is used to analyze the first causal relationship between each of the first pulmonary complications and the corresponding first risk factors, and obtain the preoperative intervention measures based on the first causal relationship and the first score corresponding to the first pulmonary complication.

6. The system according to claim 5, characterized in that The risk prediction module is further configured to: after the implementation of the preoperative intervention measure and within the first preset time period before the thoracic surgery, input the second feature into the preoperative prediction model and obtain a second output result, wherein the second output result includes N2 second pulmonary complications and corresponding second scores, and obtain the second risk factor corresponding to each second pulmonary complication according to the contribution value of each feature of each second feature to each second score; The intervention recommendation module is used to analyze the second causal relationship between each of the second pulmonary complication and the corresponding second risk factor, correspond the first pulmonary complication and the second pulmonary complication to the same disease, and use the corresponding same disease as the first target complication, obtain the first score and the second score of each of the first target complication, and calculate the first difference between the first score and the second score, weight the first score and the second score to obtain the first target score, and obtain intraoperative intervention measures based on the second causal relationship corresponding to each of the first target complication and the first target score, wherein the weight of the first score is less than the weight of the second score, and the larger the first difference, the smaller the weight of the first score, and N1 of the first pulmonary complications include N2 of the second pulmonary complications.

7. The system according to claim 6, characterized in that The intervention means recommendation module is used to input the third feature into the postoperative prediction model within the second preset time period after surgery, and obtain the third output result, wherein the third output result includes N3 third pulmonary complications and corresponding third scores, and obtain the third risk factor corresponding to each third pulmonary complication according to the contribution value of each feature in the third feature to each third score, wherein N3 is less than or equal to N2; The intervention means recommendation module is used to analyze the third causal relationship between each of the third pulmonary complication and the third risk factor, and obtain the postoperative intervention measures based on the third causal relationship, take the third pulmonary complication as the second target complication, wherein the first target complication includes the second target complication, calculate the second difference between the first target score and the third score corresponding to each of the second target complication, and set the weight of the first target score according to the second difference, and weight the first target score and the third score to obtain the second target score corresponding to the second target complication, and take the second target complication as the predicted pulmonary complication, and take the second target score as the predicted score of the predicted pulmonary complication, wherein the weight of the third score is greater than the weight of the first target score, and the larger the second difference is, the smaller the weight of the first target score is, and the smaller the second difference is, the larger the weight of the first target score is.

8. The system according to claim 1, characterized in that The first feature, the second feature and the third feature respectively include multiple features and corresponding feature values.

9. A method for predicting the risk of postoperative pulmonary complications and assisting decision making, the method being implemented based on the system according to any one of claims 1 to 8, characterized in that: The method comprises: Step S1: creating and training an admission prediction model, a preoperative prediction model, and a postoperative prediction model through a model training module, and acquiring a first related factor, a second related factor, and a third related factor based on the trained admission prediction model, the preoperative prediction model, and the postoperative prediction model; Step S2: obtaining the admission information of the patient when he / she is admitted to the hospital through the data acquisition module, obtaining a first feature from the admission information according to the first related factor, inputting the first feature into the admission prediction model, and obtaining a first output result, and obtaining a first risk factor based on the first output result, and obtaining a preoperative intervention measure for each first pulmonary complication in the first output result based on the first risk factor and the first output result; Step S3: obtaining the preoperative information of the patient at the time of admission through the data acquisition module, obtaining a second feature from the preoperative information according to the second related factor, inputting the second feature into the preoperative prediction model, and obtaining a second output result, and obtaining a second risk factor based on the second output result, and obtaining an intraoperative intervention measure for each second pulmonary complication in the second output result based on the second risk factor, the second output result and the first output result; Step S4: Obtain the postoperative information of the patient at the time of admission through the data acquisition module, obtain a third feature from the preoperative information according to the third related factor, input the third feature into the preoperative prediction model, and obtain a third output result, and obtain a third risk factor based on the third output result, and obtain postoperative intervention measures for each third pulmonary complication in the third output result based on the third risk factor, the second output result and the third output result, wherein the third pulmonary complication is the predicted pulmonary complication.

10. A computer storage medium, characterized in that: The storage medium stores program instructions, wherein when the program instructions are executed, the device where the storage medium is located is controlled to execute the method according to claim 9.

Images