An artificial intelligence-based perioperative hypothermia prediction system

By adopting artificial intelligence technology in the perioperative hypothermia prediction system, body temperature data is collected and processed in real time and relevant features are extracted, data heterogeneity and noise problems are solved, and prediction accuracy and robustness are improved.

CN119993506BActive Publication Date: 2025-07-01DAZHU COUNTY PEOPLES HOSPITAL
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202510458406.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-04-14
Publication Date
2025-07-01
Estimated Expiration
2045-04-14

AI Technical Summary

Technical Problem

In the prior art, perioperative hypothermia prediction systems face data heterogeneity, absence and noise problems, resulting in low prediction accuracy.

Method used

The prediction system based on artificial intelligence is adopted, including a data acquisition module, a body temperature feature extraction module, a related feature extraction module and a body temperature prediction module. The intelligent temperature sensor collects body temperature data in real time, calculates body temperature perturbation coefficient and trend factors, and extracts the characteristics of multidimensional factor data through deep learning to improve prediction accuracy.

Benefits of technology

Effectively identify abnormal fluctuations in perioperative body temperature changes, remove noise data, quantify body temperature trends, improve data quality and prediction accuracy, and enhance the robustness of the prediction system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119993506B_ABST
    Figure CN119993506B_ABST
Patent Text Reader

Abstract

The present application provides a perioperative hypothermia prediction system based on artificial intelligence, which relates to the technical field of body temperature prediction. It collects the body temperature sample data of a target object in the perioperative period in real time; determines the body temperature perturbation coefficient of each sample data point in the body temperature sample data, and converts the body temperature sample data into a body temperature characterization sequence of the target object in the perioperative period according to all the body temperature perturbation coefficients, and determines the body temperature trend factor of the target object in the perioperative period through the body temperature characterization sequence; obtains multi-dimensional factor data related to the target object in the perioperative period, extracts features from the multi-dimensional factor data to obtain a perioperative related feature vector of the target object; performs perioperative hypothermia prediction according to the body temperature trend factor and the perioperative related feature vector to obtain a perioperative hypothermia prediction result of the target object. The present application can improve the quality of relevant data of the target object, and extract relevant features therefrom for body temperature prediction to improve the prediction accuracy of the prediction system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the technical field of body temperature prediction. More specifically, this application relates to a perioperative hypothermia prediction system based on artificial intelligence. Background Art

[0002] Perioperative Hypothermia (PH) refers to the phenomenon that the core body temperature of a patient drops below 36°C during the surgical procedure or within a short period after surgery. Hypothermia not only increases the risks of postoperative infection, coagulation dysfunction, cardiovascular complications, and hospital stay, but also affects the postoperative recovery of the patient. Therefore, accurate prediction and prevention of perioperative hypothermia are crucial for anesthesia management and perioperative care. In recent years, the development of Artificial Intelligence (AI) technology has provided new possibilities for the prediction of perioperative hypothermia. In particular, methods based on machine learning and deep learning can utilize large-scale surgical data to improve the accuracy and real-time performance of prediction.

[0003] In the prior art, the performance of artificial intelligence models highly depends on the quality and quantity of data. However, perioperative data often has significant heterogeneity. For example, the methods of body temperature measurement (such as esophageal temperature, bladder temperature, axillary temperature, etc.) may be different in different hospitals, and there are also significant differences in data formats and recording methods, making data integration and standardization a major problem. In addition, the perioperative data of some patients may be missing and noisy, reducing the accuracy of model prediction. Therefore, how to improve the quality of relevant data of the target object and extract relevant features therefrom for body temperature prediction to enhance the prediction accuracy of the prediction system is a difficult problem faced by the industry. Summary of the Invention

[0004] This application provides a perioperative hypothermia prediction system based on artificial intelligence, which can improve the quality of relevant data of the target object and extract relevant features therefrom for body temperature prediction to enhance the prediction accuracy of the prediction system.

[0005] This application provides a perioperative hypothermia prediction system based on artificial intelligence, and the prediction system includes:

[0006] A data acquisition module, configured to collect real-time body temperature sample data of the target object during the perioperative period;

[0007] A body temperature feature extraction module, configured to determine the body temperature perturbation coefficient of each sample data point in the body temperature sample data, convert the body temperature sample data into a body temperature characterization sequence of the target object during the perioperative period according to all the body temperature perturbation coefficients, and determine the body temperature trend factor of the target object during the perioperative period through the body temperature characterization sequence;

[0008] A relevant feature extraction module, which is used to obtain multi-dimensional factor data related to the target object during the perioperative period, extract features from the multi-dimensional factor data, and then obtain a perioperative related feature vector of the target object;

[0009] A body temperature prediction module, which is used to perform perioperative hypothermia prediction according to the body temperature trend factor and the perioperative related feature vector, and then obtain a perioperative hypothermia prediction result of the target object.

[0010] In this embodiment, the body temperature sample data of the target object during the perioperative period is collected in real time through an intelligent temperature sensor.

[0011] In this embodiment, determining the body temperature perturbation coefficient of each sample data point in the body temperature sample data specifically includes:

[0012] For each sample data point in the body temperature sample data, obtain the neighborhood sample domain of the sample data point;

[0013] Determine the neighborhood similarity sequence of the sample data point through the neighborhood sample domain, and then obtain the neighborhood similarity sequences of each sample data point;

[0014] Determine the global data density of the body temperature sample data through the neighborhood similarity sequences of each sample data point;

[0015] Determine the body temperature perturbation coefficient of each sample data point in the body temperature sample data according to the corresponding neighborhood similarity sequence and the global data density respectively.

[0016] In this embodiment, converting the body temperature sample data into a body temperature characterization sequence of the target object during the perioperative period according to all the body temperature perturbation coefficients specifically includes:

[0017] Obtain a preset perturbation threshold;

[0018] Screen out all body temperature characterization data points in the body temperature sample data according to each body temperature perturbation coefficient and the perturbation threshold;

[0019] Construct a body temperature characterization sequence of the target object during the perioperative period through all the body temperature characterization data points.

[0020] In this embodiment, determining the body temperature trend factor of the target object during the perioperative period through the body temperature characterization sequence specifically includes:

[0021] Divide the body temperature characterization sequence to obtain body temperature characterization segments at different stages of the target object during the perioperative period;

[0022] Determine the body temperature trend fluctuation values at different stages of the target object during the perioperative period according to the corresponding body temperature characterization segments;

[0023] Determine the perioperative body temperature trend factor of the target object based on all body temperature trend fluctuation values.

[0024] In this embodiment, the body temperature trend factor represents the severity of the fluctuation of the body temperature data trend of the target object during the perioperative period.

[0025] In this embodiment, the multi-dimensional factor data includes the electronic medical record, anesthesia information, imaging examination data, and real-time physiological monitoring data of the target object.

[0026] In this embodiment, performing feature extraction on the multi-dimensional factor data to obtain the perioperative related feature vector of the target object specifically includes:

[0027] Preprocess the multi-dimensional factor data to obtain the preprocessed multi-dimensional factor data;

[0028] Use a feature extraction model based on deep learning to perform feature extraction on the preprocessed multi-dimensional factor data to obtain the perioperative related feature vector of the target object.

[0029] In this embodiment, predicting perioperative hypothermia based on the body temperature trend factor and the perioperative related feature vector is to input the body temperature trend factor and the perioperative related feature vector into a machine learning model for perioperative hypothermia prediction.

[0030] In this embodiment, the perioperative hypothermia prediction result includes whether the target object has hypothermia and the predicted body temperature of the target object.

[0031] In this embodiment, the anomaly detection model is a long short-term memory network model.

[0032] The technical solutions provided by the disclosed embodiments of the present application have the following beneficial effects:

[0033] The data acquisition module collects the body temperature sample data of the target object during the perioperative period in real time; the body temperature feature extraction module determines the body temperature perturbation coefficient of each sample data point in the body temperature sample data, and converts the body temperature sample data into the perioperative body temperature characterization sequence of the target object based on all the body temperature perturbation coefficients, and determines the perioperative body temperature trend factor of the target object through the body temperature characterization sequence; the related feature extraction module obtains the multi-dimensional factor data related to the target object during the perioperative period, performs feature extraction on the multi-dimensional factor data, and then obtains the perioperative related feature vector of the target object; the body temperature prediction module predicts perioperative hypothermia based on the body temperature trend factor and the perioperative related feature vector, and then obtains the perioperative hypothermia prediction result of the target object.

[0034] It can be seen that in this application, first, by calculating the body temperature perturbation coefficient, abnormal fluctuations in perioperative body temperature changes can be effectively identified, and the body temperature sample data can be converted into a body temperature characterization sequence. This process helps to remove noise data, highlight key temperature change points, and make the data more refined and efficient. By further calculating the body temperature trend factor, the body temperature change trend of an individual in different surgical stages can be quantified, thereby providing more representative input features for the artificial intelligence prediction model. Then, by obtaining multi-dimensional factor data related to the target object during the perioperative period and performing feature extraction driven by deep learning on it, a high-dimensional and high-quality perioperative-related feature vector can be constructed, which can provide a more comprehensive description of the patient's physiological state and help the prediction model have stronger generalization ability and higher prediction accuracy. Finally, by inputting the body temperature trend factor and the perioperative-related feature vector into the prediction model for perioperative hypothermia prediction, various factors affecting body temperature can be comprehensively considered, thereby improving the prediction accuracy and robustness.

[0035] In summary, the technical solution adopted in this application can improve the quality of relevant data of the target object and extract relevant features therefrom for body temperature prediction to enhance the prediction accuracy of the prediction system. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0037] Figure 1 is a module structure diagram of the perioperative hypothermia prediction system based on artificial intelligence provided by the present application;

[0038] Figure 2 is an exemplary flowchart for determining the body temperature trend factor of the target object during the perioperative period provided by the present application;

[0039] Figure 3 is an exemplary flowchart for determining the perioperative-related feature vector of the target object provided by the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0040] The following will clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0041] An embodiment of the present application provides a perioperative hypothermia prediction system based on artificial intelligence. Its core is to collect real-time body temperature sample data of a target object during the perioperative period through a data acquisition module; determine the body temperature perturbation coefficient of each sample data point in the body temperature sample data through a body temperature feature extraction module, and convert the body temperature sample data into a body temperature characterization sequence of the target object during the perioperative period based on all the body temperature perturbation coefficients. Determine the body temperature trend factor of the target object during the perioperative period through the body temperature characterization sequence; obtain multi-dimensional factor data related to the target object during the perioperative period through a related feature extraction module, extract features from the multi-dimensional factor data, and then obtain a perioperative related feature vector of the target object; the body temperature prediction module performs perioperative hypothermia prediction based on the body temperature trend factor and the perioperative related feature vector, and then obtains a perioperative hypothermia prediction result of the target object. Adopting the above solution can improve the quality of relevant data of the target object and extract relevant features therefrom for body temperature prediction to improve the prediction accuracy of the prediction system.

[0042] To better understand the above technical solution, the following will combine the accompanying drawings of the specification and specific implementation manners to elaborate on the above technical solution in detail. Refer to Figure 1 As shown, this figure is a module structure diagram of a perioperative hypothermia prediction system based on artificial intelligence according to an embodiment of the present application. The prediction system includes: a data acquisition module 100, a body temperature feature extraction module 200, a related feature extraction module 300, and a body temperature prediction module 400, which are described as follows:

[0043] The data acquisition module 100 collects real-time body temperature sample data of a target object during the perioperative period.

[0044] Specifically, the real-time body temperature sample data of the target object can be collected through an intelligent temperature sensor; it should be noted that in order to accurately predict perioperative hypothermia, using an intelligent temperature sensor for real-time body temperature monitoring can effectively improve the timeliness and accuracy of data. In this application, continuous body temperature monitoring of the target object is realized through a non-contact infrared temperature sensor or an implantable / attachable temperature sensor, so as to obtain the body temperature sample data of the target object during the perioperative period.

[0045] The body temperature feature extraction module 200 is used to determine the body temperature perturbation coefficient of each sample data point in the body temperature sample data, convert the body temperature sample data into a body temperature characterization sequence of the target object during the perioperative period based on all the body temperature perturbation coefficients, and determine the body temperature trend factor of the target object during the perioperative period through the body temperature characterization sequence.

[0046] In this embodiment, the specific method for determining the body temperature perturbation coefficient of each sample data point in the body temperature sample data can be as follows, that is:

[0047] For each sample data point in the body temperature sample data, obtain the neighboring sample domain of the sample data point;

[0048] Determine the neighboring similarity sequence of the sample data point through the neighboring sample domain, and then obtain the neighboring similarity sequences of all sample data points;

[0049] Determine the global data density of the body temperature sample data through the neighboring similarity sequences of all sample data points;

[0050] Determine the body temperature perturbation coefficient of each sample data point in the body temperature sample data according to the corresponding neighboring similarity sequence and the global data density respectively.

[0051] In specific implementation, first, for each sample data point in the body temperature sample data, the neighboring sample domain of the sample data point can be obtained, that is, calculate the neighboring similarity between the sample data point and other sample data points in the body temperature sample data by means of Euclidean distance. This neighboring similarity represents the similarity degree between sample data points, so as to select the M sample data points closest to this sample data point, where the value of M can be preset according to historical experience, and the data set composed of all selected sample data points is used as the neighboring sample domain of this sample data point; then, the neighboring similarity sequence of the sample data point can be determined through the neighboring sample domain, that is, arrange the neighboring similarities between each sample data point in the neighboring sample domain and this sample data point in ascending order, and thus the obtained sequence is used as the neighboring similarity sequence of this sample data point. Through the above method, the neighboring similarity sequences of all sample data points can be obtained.

[0052] In addition, in specific implementation, the global data density of the body temperature sample data can be determined through the neighboring similarity sequences of all sample data points. Among them, the global data density represents the overall data density of the body temperature sample data. The mean value of the neighboring similarity sequences of all sample data points can be calculated, and thus the mean value of all calculation results is used as the global data density of the body temperature sample data; then, the body temperature perturbation coefficient of each sample data point in the body temperature sample data can be determined according to the corresponding neighboring similarity sequence and the global data density respectively. Among them, the body temperature perturbation coefficient is an index indicating the degree of perturbation of the sample data point to the overall data. For each sample data point, the ratio of each neighboring similarity in the neighboring similarity sequence of the sample data point to the global data density can be calculated, and thus the mean value of all calculation results is used as the body temperature perturbation coefficient of this sample data point. Through the above method, the body temperature perturbation coefficient of each sample data point in the body temperature sample data can be obtained.

[0053] In this embodiment, according to all the body temperature perturbation coefficients, converting the body temperature sample data into the body temperature characterization sequence of the target object during the perioperative period can be specifically implemented in the following manner, that is:

[0054] Obtain a preset perturbation threshold;

[0055] Based on each body temperature perturbation coefficient and the perturbation threshold, screen out all body temperature characterization data points in the body temperature sample data;

[0056] Construct a body temperature characterization sequence of the target object during the perioperative period through all the body temperature characterization data points.

[0057] In specific implementation, first, a preset perturbation threshold can be obtained. This perturbation threshold can be set based on historical data and expert experience, which will not be elaborated here. Then, based on each body temperature perturbation coefficient and the perturbation threshold, all body temperature characterization data points in the body temperature sample data can be screened out, that is, the sample data points in the body temperature sample data with a body temperature perturbation coefficient less than the perturbation threshold are screened out, and the screened sample data points are used as body temperature characterization data points. In this way, all body temperature characterization data points in the body temperature sample data can be obtained. It should be noted that when the body temperature perturbation coefficient is not less than the perturbation threshold, it indicates that the corresponding sample data point is a perturbation data point, that is, noise and interference. Finally, a body temperature characterization sequence of the target object during the perioperative period can be constructed through all the body temperature characterization data points, that is, all the body temperature characterization data points are combined in the order of acquisition time, and the combined sequence is used as the body temperature characterization sequence of the target object during the perioperative period.

[0058] Preferably, in this embodiment, refer to Figure 2 As shown, this figure is an exemplary flowchart for determining the body temperature trend factor of the target object during the perioperative period in the embodiment of the present application. In this embodiment, the following steps can be specifically adopted to determine the body temperature trend factor of the target object during the perioperative period through the body temperature characterization sequence:

[0059] In step S21, divide the body temperature characterization sequence to obtain body temperature characterization segments at different stages of the target object during the perioperative period;

[0060] In step S22, determine the body temperature trend fluctuation values at different stages of the target object during the perioperative period according to the corresponding body temperature characterization segments;

[0061] In step S23, determine the body temperature trend factor of the target object during the perioperative period through all the body temperature trend fluctuation values.

[0062] In specific implementation, first, the body temperature characterization sequence can be partitioned, that is, the body temperature characterization sequence is partitioned according to the corresponding stages of the target object during the perioperative period, so as to obtain the body temperature characterization segments of the target object at different stages during the perioperative period. The perioperative period includes the preoperative stage, the intraoperative stage, and the postoperative stage. Then, the body temperature trend fluctuation values of the target object at different stages during the perioperative period can be determined according to the corresponding body temperature characterization segments. The body temperature trend fluctuation value represents the value of the severity of the trend fluctuation of the body temperature data of the target object during the perioperative period. The body temperature characterization data points in the body temperature characterization segment can be subjected to binomial fitting to obtain the corresponding body temperature characterization curve, and the average curvature of the body temperature characterization curve can be calculated, and the calculation result is used as the body temperature trend fluctuation value of this stage. Through the above method, the body temperature trend fluctuation values of the target object at different stages during the perioperative period can be obtained. Finally, the body temperature trend factor of the target object during the perioperative period can be determined through all the body temperature trend fluctuation values. The body temperature trend factor represents the severity of the trend fluctuation of the body temperature data of the target object during the perioperative period, that is, each body temperature trend fluctuation value can be used as the value of the body temperature trend factor at the corresponding stage. Through the above method, the body temperature trend factor of the target object during the perioperative period can be obtained.

[0063] It should be noted that by calculating the body temperature perturbation coefficient, abnormal fluctuations in body temperature changes during the perioperative period can be effectively identified, and the body temperature sample data can be converted into a body temperature characterization sequence. This process helps to remove noise data, highlight key temperature change points, and make the data more refined and efficient. By further calculating the body temperature trend factor, the body temperature change trend of an individual at different surgical stages can be quantified, thereby providing more representative input features for the artificial intelligence prediction model. Through the above method, the data quality can be significantly improved, enabling the model to pay more attention to the key factors actually affecting the occurrence of hypothermia, and at the same time reducing the interference of redundant information on the prediction accuracy.

[0064] The related feature extraction module 300 is used to obtain multi-dimensional factor data related to the target object during the perioperative period, extract features from the multi-dimensional factor data, and further obtain the perioperative related feature vector of the target object.

[0065] It should be noted that in this application, the multi-dimensional factor data includes the electronic medical record, anesthesia information, imaging examination data, and real-time physiological monitoring data of the target object. In specific implementation, the electronic medical record of the target object can be obtained through the hospital information system, the anesthesia information of the target object can be obtained through the anesthesia information management system, the imaging examination data of the target object can be obtained through the picture archiving and communication system, and the real-time physiological monitoring data of the target object can be obtained through the patient monitoring system.

[0066] Preferably, in this embodiment, refer to Figure 3As shown, this figure is an exemplary flowchart for determining the perioperative-related feature vector of the target object in the embodiment of the present application. In this embodiment, feature extraction is performed on the multi-dimensional factor data, and the following steps can be specifically used to obtain the perioperative-related feature vector of the target object:

[0067] In step S31, the multi-dimensional factor data is preprocessed to obtain the preprocessed multi-dimensional factor data;

[0068] In step S32, a feature extraction model based on deep learning is used to perform feature extraction on the preprocessed multi-dimensional factor data, thereby obtaining the perioperative-related feature vector of the target object.

[0069] Specifically, first, the multi-dimensional factor data is preprocessed, that is, the numerical data in the multi-dimensional factor data is subjected to data cleaning and missing value processing, and the time series data in the multi-dimensional factor data is aligned to ensure that the timestamps of the intraoperative physiological monitoring data, anesthesia information, and preoperative examination data are consistent. The missing time points of the physiological data at different times are filled using linear interpolation. And to reduce the impact of different data distributions on the model, the numerical features are normalized. Then, for different types of data, different deep learning models are used for feature extraction to automatically learn high-dimensional feature representations, that is, a feature extraction model based on deep learning can be used to perform feature extraction on the preprocessed multi-dimensional factor data. For example, a feature extraction model based on natural language processing (NLP) + Transformer / BERT model is used to extract features from the electronic medical records of the target object, a time series model based on LSTM (long short-term memory network) is used to extract features from the anesthesia information of the target object, a CNN deep learning network based on ResNet / EfficientNet is used to extract features from the imaging examination data, and a time series model based on LSTM / GRU + Transformer is used to extract features from the real-time physiological monitoring data. Thus, the vector composed of all the extracted features can be used as the perioperative-related feature vector of the target object.

[0070] It should be noted that by obtaining multi-dimensional factor data related to the target object during the perioperative period (such as electronic medical records, anesthesia information, imaging examination data, and real-time physiological monitoring data) and performing deep learning-driven feature extraction on it, a high-dimensional and high-quality perioperative-related feature vector (PFV) can be constructed. The main advantages of doing so are as follows: First, multi-source data fusion can provide a more comprehensive description of the patient's physiological state, improve the integrity and consistency of the data, and avoid information loss or bias that may be caused by a single data source. Second, the deep learning model can automatically extract complex latent patterns, especially time-series features with long-term dependencies, such as the intraoperative body temperature change trend and the impact of anesthetic drugs on body temperature, ensuring that the model can learn key features. In addition, the combination of imaging data and text medical record information can help identify potential risk factors for patient hypothermia (such as thyroid dysfunction or hemodynamic instability), thereby enhancing the robustness of the prediction. Finally, through multi-modal data fusion and feature vector construction, the body temperature prediction model has stronger generalization ability and higher prediction accuracy, can identify perioperative hypothermia risks earlier and more accurately, optimize surgical management, and improve patient safety.

[0071] The body temperature prediction module 400 is used to perform perioperative hypothermia prediction based on the body temperature trend factor and the perioperative-related feature vector, and then obtain the perioperative hypothermia prediction result of the target object.

[0072] In this embodiment, performing perioperative hypothermia prediction based on the body temperature trend factor and the perioperative-related feature vector is to input the body temperature trend factor and the perioperative-related feature vector into a machine learning model for perioperative hypothermia prediction.

[0073] It should be noted that the perioperative hypothermia prediction result includes whether the target object has hypothermia and the predicted body temperature of the target object; in specific implementation, the body temperature trend factor and the perioperative-related feature vector can be input into a suitable machine learning model. After the body temperature trend factor and the perioperative-related feature vector undergo feature fusion (such as feature splicing or weighted fusion), a unified input vector is formed and transmitted to the machine learning model. The machine learning model used in this application is a support vector machine (SVM). In actual implementation, other machine learning models can also be selected, which is not limited here. The machine learning model will make a prediction based on the real-time input body temperature trend factor and perioperative-related feature vector, and the machine learning model will output the perioperative hypothermia prediction result, which includes whether the target object has hypothermia and the predicted body temperature of the target object.

[0074] In addition, it should be noted that by inputting the body temperature trend factor and perioperative-related feature vectors into the prediction model for perioperative hypothermia prediction, various factors affecting body temperature can be comprehensively considered, such as the patient's basic information, anesthesia data, real-time physiological monitoring, imaging examinations, etc., improving the comprehensiveness and accuracy of the prediction. Through the fusion and feature extraction of multi-dimensional data, not only the quality of the data related to the target object is improved, but also the prediction system can automatically identify key factors from a large amount of complex data, reducing the error of human intervention, thereby improving the prediction accuracy and robustness.

[0075] Thus, in this application, first, by calculating the body temperature perturbation coefficient, abnormal fluctuations in perioperative body temperature changes can be effectively identified, and the body temperature sample data can be converted into a body temperature characterization sequence. This process helps to remove noise data, highlight key temperature change points, make the data more refined and efficient. By further calculating the body temperature trend factor, the body temperature change trend of an individual in different surgical stages can be quantified, thus providing more representative input features for the artificial intelligence prediction model; then, by obtaining multi-dimensional factor data related to the target object during the perioperative period and performing feature extraction driven by deep learning, a high-dimensional and high-quality perioperative-related feature vector can be constructed, which can provide a more comprehensive description of the patient's physiological state and help the prediction model have stronger generalization ability and higher prediction accuracy; finally, by inputting the body temperature trend factor and perioperative-related feature vectors into the prediction model for perioperative hypothermia prediction, various factors affecting body temperature can be comprehensively considered, thereby improving the prediction accuracy and robustness.

[0076] In summary, the technical solution adopted in this application can improve the quality of the data related to the target object and extract relevant features therefrom for body temperature prediction to improve the prediction accuracy of the prediction system.

[0077] This application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0078] Those of ordinary skill 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 relevant hardware through a program, and the program can be stored in a computer-readable storage medium. The storage medium includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc memories, magnetic disk memories, tape memories, or any other medium that can be used to carry or store data and is computer-readable.

[0079] It should also be noted that the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or also includes elements inherent in such a process, method, commodity or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, commodity or device including the element.

Claims

1. An artificial intelligence-based perioperative hypothermia prediction system, characterized in that: The prediction system comprises: A data collection module is used to collect temperature sample data of the target object in the perioperative period in real time; A body temperature feature extraction module, used to determine the body temperature perturbation coefficient of each sample data point in the body temperature sample data, convert the body temperature sample data into a body temperature representation sequence of the target object in the perioperative period according to all the body temperature perturbation coefficients, and determine the body temperature trend factor of the target object in the perioperative period through the body temperature representation sequence; A related feature extraction module is used to obtain multidimensional factor data related to the target object during the perioperative period, perform feature extraction on the multidimensional factor data, and then obtain a perioperative related feature vector of the target object; A body temperature prediction module, used to predict perioperative hypothermia according to the body temperature trend factor and the perioperative period-related feature vector, and then obtain a perioperative hypothermia prediction result of the target object; Wherein, determining the body temperature disturbance coefficient of each sample data point in the body temperature sample data specifically includes: For each sample data point in the body temperature sample data, obtaining a neighboring sample domain of the sample data point; Determine the neighbor similarity sequence of the sample data point through the neighbor sample domain, and then obtain the neighbor similarity sequence of each sample data point; Determine the global data density of the body temperature sample data through the nearest similarity sequence of each sample data point; Determine the body temperature perturbation coefficient of each sample data point in the body temperature sample data according to the corresponding nearest neighbor similarity sequence and the global data density, wherein the body temperature perturbation coefficient is an indicator indicating the degree of perturbation of the sample data point to the overall data; The method of converting the body temperature sample data into a body temperature representation sequence of the target object during the perioperative period according to all body temperature disturbance coefficients specifically includes: Obtaining a preset disturbance threshold; Filtering out all body temperature characterization data points in the body temperature sample data according to each body temperature disturbance coefficient and the disturbance threshold; Constructing a temperature representation sequence of the target object during the perioperative period through all the temperature representation data points; Wherein, determining the temperature trend factor of the target object during the perioperative period through the temperature characterization sequence specifically includes: Dividing the body temperature representation sequence to obtain body temperature representation segments of the target subject at different stages during the perioperative period; Determine the temperature trend fluctuation value of the target object at different stages during the perioperative period according to the corresponding temperature representation segment; The body temperature trend factor of the target object in the perioperative period is determined by all the body temperature trend fluctuation values, wherein the body temperature trend factor represents the severity of the body temperature data trend fluctuation of the target object in the perioperative period.

2. The artificial intelligence-based perioperative hypothermia prediction system according to claim 1, characterized in that: The intelligent temperature sensor is used to collect the target object’s body temperature sample data in real time during the perioperative period.

3. The artificial intelligence-based perioperative hypothermia prediction system according to claim 1, characterized in that: The multidimensional factor data includes the target subject's electronic medical record, anesthesia information, imaging examination data and real-time physiological monitoring data.

4. The artificial intelligence-based perioperative hypothermia prediction system according to claim 1, characterized in that: Extracting features from the multidimensional factor data to obtain perioperative-related feature vectors of the target object specifically includes: Preprocessing the multidimensional factor data to obtain preprocessed multidimensional factor data; A deep learning-based feature extraction model is used to extract features from the preprocessed multidimensional factor data to obtain the perioperative-related feature vectors of the target object.

5. The artificial intelligence-based perioperative hypothermia prediction system according to claim 1, characterized in that: Predicting perioperative hypothermia according to the body temperature trend factor and the perioperative period-related feature vector is to input the body temperature trend factor and the perioperative period-related feature vector into a machine learning model to predict perioperative hypothermia.

6. The artificial intelligence-based perioperative hypothermia prediction system according to claim 1, characterized in that: The perioperative hypothermia prediction result includes whether the target subject has hypothermia and the predicted body temperature of the target subject.

Citation Information

Patent Citations

  • Source network load self-optimization intelligent regulation and control system based on wide-area perception prediction and real-time control optimization technical method

    CN114243727A

  • Multiple calibration single classification method for unsupervised time sequence anomaly detection

    CN117807470A