An intraoperative arterial partial pressure of oxygen prediction method and device, electronic equipment and medium

Abstract

This application provides a method, device, electronic device, and medium for predicting intraoperative arterial oxygen partial pressure (APPS). The method includes: acquiring anchor point data collected intraoperatively to characterize the patient's baseline physiological state and non-invasive monitoring data of the patient at the current moment, wherein the anchor point data includes a reference value for APS; using the non-invasive monitoring data and the anchor point data, capturing the degree of deviation of the patient's intraoperative physiological state from the baseline physiological state to determine derived features; inputting the derived features into an APS prediction model to predict the change in APS at the current moment relative to the reference value for APS; and determining the predicted value of the patient's APS at the current moment based on the change and the reference value for APS. By employing the above-mentioned method, device, electronic device, and medium for predicting intraoperative APS, the problems of low prediction accuracy and poor individualized adaptability of APS are solved.

Description

Technical Field

[0001] This application relates to the field of medical artificial intelligence technology, and more specifically, to a method, device, electronic device, and medium for predicting intraoperative arterial blood oxygen partial pressure. Background Technology

[0002] During surgical procedures, arterial partial pressure of oxygen (PaO2) is a core physiological indicator for assessing a patient's intraoperative oxygenation status. Its changes directly reflect the patient's pulmonary gas exchange function, circulatory perfusion, and oxygen supply level, serving as a crucial basis for anesthesiologists to adjust ventilation parameters and implement oxygen therapy interventions in real time. Currently, the gold standard for obtaining arterial partial pressure of oxygen in clinical practice is arterial blood gas analysis (ABG). This method requires collecting arterial blood samples through puncture for in vitro testing. While the results are accurate, it has significant limitations: First, it is an invasive procedure and is not suitable for critically ill patients requiring continuous intraoperative oxygenation monitoring or patients undergoing prolonged and complex surgeries. Second, there is a time delay in the testing process; the entire process of blood sample collection, delivery, testing, and result feedback takes several minutes, making real-time monitoring of oxygenation impossible. This can lead to physicians being unable to promptly detect sudden changes in the patient's PaO2, delaying intervention.

[0003] To address the aforementioned issues, existing technologies have developed PaO2 prediction methods based on non-invasive monitoring parameters. However, these methods often rely on static data for fitting and prediction, neglecting the individualized physiological fluctuations of patients during surgery. Consequently, they suffer from low prediction accuracy and poor individualization. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a method, device, electronic device and medium for predicting intraoperative arterial oxygen partial pressure, so as to overcome at least one of the above-mentioned defects.

[0005] In a first aspect, embodiments of this application provide a method for predicting intraoperative arterial oxygen partial pressure, including: Acquire intraoperative anchor data to characterize the patient’s baseline physiological status and non-invasive monitoring data of the patient at the current moment. The anchor data includes arterial blood oxygen partial pressure reference value. By using non-invasive monitoring data and anchor point data, the degree of deviation of the patient's physiological state during the operation from the baseline physiological state is captured in order to determine the derived characteristics; The derived features are input into the arterial blood oxygen partial pressure prediction model to predict the change in arterial blood oxygen partial pressure relative to the arterial blood oxygen partial pressure reference value at the current moment. Based on the change and the reference value of arterial blood oxygen partial pressure, the predicted value of the patient's arterial blood oxygen partial pressure at the current moment is determined.

[0006] In an optional implementation, anchor point data is obtained by using the most recent intraoperative arterial blood gas analysis time as the anchor point time, and determining the anchor point data based on the physiological parameter values ​​at the anchor point time.

[0007] In an optional implementation, the derived features include anchor point baseline features, ratio features, and difference features. The step of using non-invasive monitoring data and anchor point data to capture the degree of deviation of the patient's intraoperative physiological state from the baseline physiological state to determine the derived features includes: determining anchor point baseline features based on anchor point data; determining ratio features based on the ratio of corresponding parameters of anchor point data and non-invasive monitoring data; and determining difference features based on the difference of corresponding parameters of anchor data and non-invasive monitoring data.

[0008] In optional implementations, the anchor reference features include at least one of the following: initial oxygenation index, initial ROX index; the ratio features include at least one of the following: systolic blood pressure ratio, heart rate ratio, pulse rate ratio, ROX index ratio, blood oxygen ratio, mean arterial pressure ratio, end-tidal carbon dioxide ratio, airway pressure ratio, diastolic blood pressure ratio; the difference features include at least one of the following: systolic blood pressure difference, ROX index difference, pulse rate difference, tidal volume difference, heart rate difference, airway pressure difference, blood oxygen difference.

[0009] In an optional implementation, after acquiring the anchor point data collected during the operation to characterize the patient's baseline physiological state and the patient's non-invasive monitoring data at the current moment, the method further includes: when there are missing values ​​in the anchor point data or non-invasive monitoring data, using the median value in the training set to complete the missing values.

[0010] In an optional implementation, the method further includes: acquiring the patient's preoperative demographic data to determine the median value in the training set using the preoperative demographic data.

[0011] In an optional implementation, the parameter types in the non-invasive monitoring data correspond to the parameter types in the anchor point data.

[0012] Secondly, embodiments of this application also provide an intraoperative arterial oxygen partial pressure prediction device, the device comprising: The data acquisition module is used to acquire anchor data collected during the operation to characterize the patient's baseline physiological status and non-invasive monitoring data of the patient at the current moment. The anchor data includes the arterial blood oxygen partial pressure reference value. The feature determination module is used to capture the degree of deviation of the patient's intraoperative physiological state from the baseline physiological state using non-invasive monitoring data and anchor point data, so as to determine the derived features. The change prediction module is used to input the derived features into the arterial blood oxygen partial pressure prediction model to predict the change in arterial blood oxygen partial pressure relative to the arterial blood oxygen partial pressure reference value at the current moment. The partial pressure determination module is used to determine the predicted value of the patient's arterial oxygen partial pressure at the current moment based on the change and the reference value of arterial blood oxygen partial pressure.

[0013] Thirdly, embodiments of this application also provide an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, the steps of the intraoperative arterial blood oxygen partial pressure prediction method described above are performed.

[0014] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the intraoperative arterial oxygen partial pressure prediction method described above.

[0015] The embodiments of this application bring the following beneficial effects: This application provides a method, device, electronic device, and medium for predicting intraoperative arterial oxygen partial pressure (PaO2) using non-invasive monitoring data as the core input. This eliminates the need for multiple intraoperative arterial punctures, avoiding the risks associated with invasive procedures. Furthermore, the non-invasive data acquisition has no time delay. Combined with the model's rapid computational capabilities, real-time PaO2 prediction can be achieved. Simultaneously, anchor point data is introduced as a precise reference for the patient's intraoperative baseline physiological state. By constructing ratio and difference-based derived features, the degree of deviation of the patient's physiological state from their baseline is accurately captured, rather than using a uniform reference standard. This adapts to the individual differences of different patients. Compared with existing intraoperative arterial oxygen partial pressure prediction methods, this solves the problems of low prediction accuracy and poor individualization of PaO2.

[0016] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart of the intraoperative arterial oxygen partial pressure prediction method provided in the embodiments of this application is shown; Figure 2 A schematic diagram of the intraoperative arterial oxygen partial pressure prediction device provided in the embodiments of this application is shown; Figure 3 A schematic diagram of the structure of the electronic device provided in the embodiments of this application is shown. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0020] To facilitate understanding of this embodiment, the following describes each of the exemplary steps provided in this embodiment using the intraoperative arterial oxygen partial pressure prediction method provided in this application embodiment as an example of its application to a terminal device.

[0021] Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for predicting intraoperative arterial oxygen partial pressure as provided in an embodiment of this application. Figure 1 As shown in the embodiments of this application, the intraoperative arterial oxygen partial pressure prediction method includes: Step S101: Acquire the anchor point data collected during the operation to characterize the patient's baseline physiological state and the non-invasive monitoring data of the patient at the current moment.

[0022] The term "patient" can refer to a patient who is currently undergoing surgery.

[0023] Anchor point data is data collected during the operation of a patient to characterize the patient's baseline physiological state during the operation. Anchor point data and non-invasive monitoring data are collected at different times. Anchor point data includes the arterial partial pressure of oxygen collected at the anchor point time, and the arterial partial pressure of oxygen collected at the anchor point time is called the arterial partial pressure of oxygen reference value.

[0024] The core of this step is to acquire two types of homologous physiological data, laying the foundation for subsequent feature construction and model prediction.

[0025] In one embodiment, the parameter types of non-invasive monitoring data correspond one-to-one with the parameter types of anchor point data to ensure data comparability.

[0026] In this embodiment, before surgery, the patient's preoperative demographic data is retrieved from the hospital information system to determine the median value in the training set. This demographic data includes, but is not limited to: age, gender, body mass index (BMI), preoperative lung function classification, type of surgery, and history of underlying diseases, providing a stratified basis for subsequent missing value completion.

[0027] Then, during the surgery, the most recent arterial blood gas analysis time preceding the current moment is used as the anchor point time, and anchor point data is determined based on the physiological parameter values ​​at the anchor point time. For example, using the most recent arterial blood gas analysis time during the operation as the anchor point time, arterial blood gas analysis data at the anchor point time and simultaneously acquired non-invasive monitoring physiological parameter values ​​are collected, and the two are integrated into anchor point data.

[0028] The anchor point data includes, but is not limited to: arterial oxygen partial pressure, inhaled oxygen concentration, pulse oxygen saturation, respiratory rate, systolic blood pressure, diastolic blood pressure, mean arterial pressure, heart rate, pulse rate, end-expiratory carbon dioxide, peak airway pressure, and tidal volume.

[0029] For example, the anchor point time is T0 during the operation. The measured PaO2 of arterial blood gas analysis at this anchor point time is 136 mmHg. Non-invasive monitoring data at the anchor point time is also recorded, such as the inhaled oxygen concentration of 50%. All of the above data are used as anchor point data.

[0030] As the surgery continues, non-invasive physiological parameters can be collected in real time at any point after the anchor point (e.g., T1, T2, T3) using routine intraoperative non-invasive monitoring devices (such as electrocardiogram monitors and ventilator monitoring modules). When the physician needs to determine the arterial blood oxygen partial pressure at the current time T3, in response to the prediction command, the non-invasive monitoring data at the current time T3 is obtained from the non-invasive monitoring device. This non-invasive monitoring data is collected non-invasively, has no time delay, and the parameter types completely correspond to the anchor point data, ensuring that ratio and difference calculations can be performed with the anchor point data.

[0031] In one embodiment, the anchor point data and the non-invasive monitoring data at the current moment may have missing values, which can be filled in.

[0032] For example, after acquiring anchor data and non-invasive monitoring data of the patient at the current moment, if there are missing values ​​in the anchor data or non-invasive monitoring data, for the missing item (such as heart rate), based on the preoperative demographic data, multiple candidate data matching the patient's age, gender, body mass index and surgical type are selected from the training set. The median value of the multiple candidate data is used as the target data, and the missing value is filled in using the target data.

[0033] Unlike traditional global mean completion, this application uses stratified completion based on preoperative demographic data, which makes the completed values ​​more closely match the individual characteristics of patients, reduces the impact of missing data on the prediction results, and improves the robustness of the model.

[0034] Step S102: Using non-invasive monitoring data and anchor point data, the degree of deviation of the patient's physiological state during the operation from the baseline physiological state is captured in order to determine the derived characteristics.

[0035] This step is the core of achieving personalized prediction. By constructing derived features, absolute physiological parameter values ​​can be transformed into relative changes relative to the patient's own baseline, accurately capturing the dynamic deviation of the patient's physiological state during surgery, rather than using a unified industry reference standard. This solves the problem of poor individualization adaptability of traditional methods from the feature level.

[0036] Derived features are a set of features constructed through specific operations based on intraoperative anchor point data and current non-invasive monitoring data. They are used to reflect the relative changes in the patient's physiological state during surgery and are the core features for achieving individualized and accurate prediction of intraoperative arterial oxygen partial pressure. Unlike directly collected raw physiological parameters, they provide effective inputs for the arterial oxygen partial pressure prediction model that fit the dynamic changes in intraoperative physiology.

[0037] In one embodiment, the derived features include anchor point baseline features, ratio features, and difference features. Anchor point baseline features can be determined based on anchor point data; ratio features can be determined based on the ratio of corresponding parameters between the anchor point data and the non-invasive monitoring data; and difference features can be determined based on the difference between the corresponding parameters between the anchor point data and the non-invasive monitoring data. The construction process of the derived features is entirely based on the corresponding parameters of the anchor point data and the non-invasive monitoring data at the current time.

[0038] When determining the anchor point baseline features, core parameters reflecting the patient's basic oxygenation capacity during surgery can be directly extracted from the anchor point data as anchor point baseline features, which can serve as the basis for subsequent judgment of changes in oxygenation status.

[0039] Anchor benchmark characteristics include at least one of the following: initial oxygenation index and initial ROX index. The initial oxygenation index is the ratio of arterial blood oxygen partial pressure reference value to inhaled oxygen concentration, and the initial ROX index is the ratio of pulse blood oxygen saturation to inhaled oxygen concentration divided by respiratory rate. Both are core indicators for clinical assessment of oxygenation status.

[0040] When determining the ratio characteristics, the actual calculation is the ratio of the parameters corresponding to the anchor point data and the non-invasive monitoring data at the current time. That is, the ratio characteristics are the parameter values ​​at the anchor point time divided by the parameter values ​​corresponding to the non-invasive monitoring data at the current time, which are used to reflect the relative change ratio of the parameter values ​​at the anchor point time and the current time. The ratio characteristics include at least one of the following: systolic blood pressure ratio, heart rate ratio, pulse rate ratio, ROX index ratio, blood oxygen ratio, mean arterial pressure ratio, end-tidal carbon dioxide ratio, airway pressure ratio, and diastolic blood pressure ratio.

[0041] For example, if the systolic blood pressure in the current non-invasive monitoring data is 125 mmHg, and the systolic blood pressure at the anchor point is 128 mmHg, then the systolic blood pressure ratio is... .

[0042] When determining the difference characteristics, the actual calculation is to compare the difference between the corresponding parameters of the anchor point data and the non-invasive monitoring data at the current time. This reflects the absolute change in the parameter values ​​between the anchor point time and the current time. The difference characteristics include at least one of the following: systolic pressure difference, ROX index difference, pulse rate difference, tidal volume difference, heart rate difference, airway pressure difference, and blood oxygen difference.

[0043] For example, if the current heart rate is 78 beats / min and the heart rate at the anchor point is 82 beats / min, then the heart rate difference is 4 beats / min.

[0044] Step S103: Input the derived features into the arterial blood oxygen partial pressure prediction model to predict the change in arterial blood oxygen partial pressure at the current moment relative to the arterial blood oxygen partial pressure reference value.

[0045] The arterial blood oxygen partial pressure prediction model is a regression model built based on machine learning algorithms. For example, the arterial blood oxygen partial pressure prediction model can be an XGBoost regression model, which has the characteristics of strong anti-overfitting ability, good fitting effect on nonlinear features, and fast calculation speed, and is suitable for the needs of real-time prediction during surgery.

[0046] The derived features are input into the pre-trained arterial oxygen partial pressure prediction model. The arterial oxygen partial pressure prediction model outputs the change in arterial oxygen partial pressure at the current time relative to the reference value of arterial oxygen partial pressure at the anchor point time through nonlinear fitting calculation of the multi-dimensional derived features.

[0047] In one embodiment, when training the arterial oxygen partial pressure (OPS) prediction model, intraoperative clinical data from multiple surgical patients can be collected, including multiple anchor point data for each patient, non-invasive monitoring data at different times after the anchor point, measured arterial OPS values ​​from arterial blood gas analysis at the corresponding times, and preoperative demographic data. The clinical data is preprocessed (outlier cleaning, missing value completion) and derivative features are constructed to obtain training set derivative features. The actual change in arterial OPS for each sample is calculated, which is equal to the difference between the measured arterial OPS value and the reference value of arterial OPS at the anchor point. Then, using the training set derivative features as input and the actual change in arterial OPS as output, the machine learning model is trained and hyperparameters are optimized (e.g., learning rate, tree depth, number of iterations). Finally, the trained model is validated using a test set. If the model's prediction accuracy meets the preset clinical threshold, it is determined to be a pre-trained arterial OPS prediction model and can be deployed to intraoperative monitoring equipment.

[0048] In one embodiment, the number of derived features is fixed, for example, 24. However, the importance of these 24 derived features in predicting arterial partial pressure of oxygen varies and changes dynamically. Therefore, the derived features are stratified according to their clinical oxygenation relevance, and dynamic feature weights are assigned to different levels of derived features based on the real-time trends of physiological parameters collected intraoperatively. Simultaneously, feature kernel clustering is introduced to allow the arterial partial pressure of oxygen prediction model to focus on core feature clusters strongly correlated with oxygenation changes during prediction, thus improving prediction accuracy from both feature weighting and core focus dimensions. Furthermore, with the introduction of dynamic feature weights and feature kernel clustering, the XGBoost model also needs to undergo weighted feature adaptation training when training the arterial partial pressure of oxygen prediction model.

[0049] Specifically, combining clinical knowledge of anesthesiology and respiratory medicine, multiple derived features were stratified into three levels of oxygenation correlation to clarify the degree of correlation between each derived feature and changes in arterial blood oxygen partial pressure.

[0050] For example, parameters such as initial oxygenation index, initial ROX index, airway pressure ratio, airway pressure difference, ROX index ratio, and ROX index difference directly reflect the characteristics of lung gas exchange, oxygenation efficiency, and airway ventilation status. They are core influencing factors of changes in arterial blood oxygen partial pressure. Therefore, these derived characteristics can be classified as first-level characteristics; heart rate ratio, heart rate difference, systolic blood pressure ratio, and systolic blood pressure difference can be classified as second-level characteristics; and the remaining derived characteristics can be classified as third-level characteristics.

[0051] Each derived feature is assigned a feature weight according to its level, and the feature weights are dynamically adjusted based on the real-time trend of parameter changes.

[0052] For example, after determining the values ​​of 24 derived features, the changing trends (e.g., increasing, decreasing, or stable) of the first-level and second-level features are determined. The feature weights are then dynamically adjusted based on the magnitude of feature changes, giving higher weights to strongly correlated features that show significant abnormal changes, further enhancing the model's ability to capture changes in the oxygenation core. Specifically, if the magnitude of a feature change is greater than or equal to a set threshold, indicating a significant change in the derived feature, its feature weight is increased. Conversely, if the magnitude of a feature change in a derived feature at the same level as a significantly changed feature is less than the set threshold, indicating no significant change in the derived feature, its feature weight is decreased to ensure the total weight sums to 1. Furthermore, the feature weights of the third-level features remain unchanged and are not included in the adjustment.

[0053] After completing the dynamic correction of feature weights, a feature kernel clustering and filtering logic is introduced to extract core clusters from the weighted features and remove low-weight interfering features, so that the XGBoost model only focuses on high-weight core feature clusters for computation, reducing prediction errors caused by feature redundancy.

[0054] For example, a weight screening threshold is set, with 0.1 as the weight screening critical value. Derivative features with feature weights less than 0.1 are removed. The high-weight derivative features after screening are aggregated into three core feature clusters (such as airway ventilation feature cluster, oxygenation efficiency feature cluster, and circulation perfusion feature cluster) according to the oxygenation path. The original weighted weights of the features within each cluster are retained. The three core feature clusters and their corresponding dynamic weights are then input into the feature input layer of the XGBoost model to determine the amount of change.

[0055] It should be noted that the oxygenation pathway is the complete physiological process by which the body absorbs oxygen from the external environment and binds it to the blood. It is the physiological basis for the formation of arterial blood oxygen partial pressure (APPS), while APS is a direct quantitative indicator of the oxygen exchange process from the alveoli to the blood within the oxygenation pathway. The two are strongly correlated in terms of physiological process and outcome representation. Any abnormality in any link of the oxygenation pathway during surgery will directly or indirectly lead to fluctuations in APS. The decomposition and feature clustering of the oxygenation pathway in this application can accurately capture the impact of abnormalities in each link on APS, thereby improving the prediction accuracy of APS from a physiological perspective.

[0056] Step S104: Based on the change and the reference value of arterial blood oxygen partial pressure, determine the predicted value of the patient's arterial blood oxygen partial pressure at the current moment.

[0057] Based on the change output of the arterial blood oxygen partial pressure prediction model, combined with the arterial blood oxygen partial pressure reference value in the anchor data, the predicted arterial blood oxygen partial pressure value at the current moment can be determined by linear calculation. The predicted arterial blood oxygen partial pressure value at the current moment is equal to the sum of the arterial blood oxygen partial pressure reference value and the change.

[0058] For example, if the reference value for arterial partial pressure of oxygen is 118 mmHg and the change predicted by the arterial partial pressure of oxygen prediction model is 3 mmHg, then the predicted value for arterial partial pressure of oxygen at the current moment is 121 mmHg.

[0059] If a new arterial blood gas analysis is performed after the current moment during the operation, the new blood gas analysis time will be used as the new anchor time to update the anchor data, reconstruct the derived features of subsequent monitoring times, ensure the timeliness and accuracy of the baseline reference, and adapt to the drastic changes in the patient's physiological state during the operation (such as massive hemorrhage, sudden drop in lung compliance, and adjustment of ventilation parameters).

[0060] The intraoperative arterial oxygen partial pressure prediction method provided in this application uses non-invasive monitoring data as the core input, eliminating the need for multiple intraoperative arterial punctures and avoiding the risks associated with invasive procedures. Furthermore, the non-invasive data acquisition has no time delay, and combined with the model's rapid computing capabilities, it can achieve real-time prediction of PaO2. At the same time, it introduces anchor point data as an accurate reference for the patient's intraoperative baseline physiological state. By constructing ratio and difference-type derived features, it accurately captures the degree of deviation of the patient's physiological state from their own baseline, rather than using a uniform reference standard. This adapts to the individual differences of different patients, solving the problems of low prediction accuracy and poor individualization of PaO2.

[0061] Based on the same inventive concept, this application also provides an intraoperative arterial oxygen partial pressure prediction device corresponding to the intraoperative arterial oxygen partial pressure prediction method. Since the principle of the device in this application is similar to the intraoperative arterial oxygen partial pressure prediction method described above in this application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0062] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of an intraoperative arterial oxygen partial pressure prediction device provided in an embodiment of this application. Figure 2 As shown, the intraoperative arterial oxygen partial pressure prediction device 200 includes: The data acquisition module 201 is used to acquire anchor point data collected during the operation to characterize the patient's baseline physiological state and non-invasive monitoring data of the patient at the current moment. The anchor point data includes the arterial blood oxygen partial pressure reference value. The feature determination module 202 is used to capture the degree of deviation of the patient's physiological state during the operation from the baseline physiological state using non-invasive monitoring data and anchor point data, so as to determine the derived features. The change prediction module 203 is used to input the derived features into the arterial blood oxygen partial pressure prediction model to predict the change in arterial blood oxygen partial pressure relative to the arterial blood oxygen partial pressure reference value at the current moment. Partial pressure determination module 204 is used to determine the predicted value of the patient's arterial oxygen partial pressure at the current moment based on the change and the reference value of arterial blood oxygen partial pressure.

[0063] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 3 As shown, the electronic device 300 includes a processor 310, a memory 320, and a bus 330.

[0064] The memory 320 stores machine-readable instructions executable by the processor 310. When the electronic device 300 is running, the processor 310 and the memory 320 communicate via the bus 330. When the machine-readable instructions are executed by the processor 310, they can perform the operations described above. Figure 1 The steps of the intraoperative arterial oxygen partial pressure prediction method in the illustrated method embodiment can be found in the method embodiment for specific implementation methods, which will not be repeated here.

[0065] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described actions. Figure 1 The steps of the intraoperative arterial oxygen partial pressure prediction method in the illustrated method embodiment can be found in the method embodiment for specific implementation methods, which will not be repeated here.

[0066] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0067] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0068] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0069] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0070] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0071] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for predicting intraoperative arterial oxygen partial pressure, characterized in that, include: Acquire intraoperative anchor data to characterize the patient’s baseline physiological status and non-invasive monitoring data of the patient at the current moment, wherein the anchor data includes arterial blood oxygen partial pressure reference value; Using the non-invasive monitoring data and the anchor point data, the degree of deviation of the patient's intraoperative physiological state from the baseline physiological state is captured to determine derived characteristics; The derived features are input into the arterial blood oxygen partial pressure prediction model to predict the change in arterial blood oxygen partial pressure at the current moment relative to the arterial blood oxygen partial pressure reference value. Based on the change and the arterial blood oxygen partial pressure reference value, the predicted value of the patient's arterial blood oxygen partial pressure at the current moment is determined.

2. The method according to claim 1, characterized in that, Anchor point data can be obtained in the following ways: The most recent intraoperative arterial blood gas analysis time was used as the anchor point time, and the anchor point data was determined based on the physiological parameter values ​​at the anchor point time.

3. The method according to claim 1, characterized in that, The derived features include anchor point baseline features, ratio features, and difference features. The step of using the non-invasive monitoring data and the anchor point data to capture the degree of deviation of the patient's intraoperative physiological state from the baseline physiological state to determine the derived features includes: Based on the anchor point data, determine the anchor point reference features; Based on the ratio of the corresponding parameters of the anchor point data and the non-invasive monitoring data, the ratio characteristics are determined; The difference characteristics are determined based on the difference between the corresponding parameters of the anchor point data and the non-invasive monitoring data.

4. The method according to claim 3, characterized in that, The anchor point reference characteristics include at least one of the following: initial oxygenation index, initial ROX index; The ratio characteristics include at least one of the following: systolic blood pressure ratio, heart rate ratio, pulse rate ratio, ROX index ratio, blood oxygen ratio, mean arterial pressure ratio, end-tidal carbon dioxide ratio, airway pressure ratio, and diastolic blood pressure ratio; The difference features include at least one of the following: systolic pressure difference, ROX index difference, pulse rate difference, tidal volume difference, heart rate difference, airway pressure difference, and blood oxygen difference.

5. The method according to claim 1, characterized in that, After acquiring the intraoperative anchor point data characterizing the patient's baseline physiological state and the patient's non-invasive monitoring data at the current moment, the procedure also includes: When there are missing values ​​in the anchor point data or the non-invasive monitoring data, the missing values ​​are filled in using the median value in the training set.

6. The method according to claim 5, characterized in that, The method further includes: Obtain the preoperative demographic data of the patients to determine the median value in the training set using the preoperative demographic data.

7. The method according to claim 1, characterized in that, The parameter types in the non-invasive monitoring data correspond to the parameter types in the anchor point data.

8. A device for predicting intraoperative arterial oxygen partial pressure, characterized in that, include: The data acquisition module is used to acquire anchor point data collected during the operation to characterize the patient's baseline physiological state and the patient's non-invasive monitoring data at the current moment. The anchor point data includes arterial blood oxygen partial pressure reference values. The feature determination module is used to capture the degree of deviation of the patient's intraoperative physiological state from the baseline physiological state using the non-invasive monitoring data and the anchor point data, so as to determine the derived features; The change prediction module is used to input the derived features into the arterial blood oxygen partial pressure prediction model to predict the change in arterial blood oxygen partial pressure at the current moment relative to the arterial blood oxygen partial pressure reference value. The partial pressure determination module is used to determine the predicted value of the patient's arterial oxygen partial pressure at the current moment based on the change and the arterial blood oxygen partial pressure reference value.

9. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the steps of the intraoperative arterial oxygen partial pressure prediction method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the intraoperative arterial oxygen partial pressure prediction method as described in any one of claims 1 to 7.

Images