Method and system for predicting mortality risk in patients with peripheral arterial disease

Abstract

The application provides a peripheral arterial disease patient death risk prediction method and system, the method comprising: acquiring preoperative cross-sectional images of the maximum soft tissue of the lower leg of a plurality of peripheral arterial disease patients; acquiring the lower leg skeletal muscle area of the plurality of peripheral arterial disease patients according to the cross-sectional images; acquiring the lower leg muscle index of the plurality of peripheral arterial disease patients according to the height data and the lower leg skeletal muscle area of the plurality of peripheral arterial disease patients; training a gradient boosting machine model using clinical baseline physiological data, lower leg muscle indexes and postoperative survival data of the plurality of peripheral arterial disease patients as training samples to obtain a death risk prediction model; and predicting the postoperative death risk of a target patient with a target peripheral arterial disease by using the death risk prediction model to obtain death risk prediction results corresponding to a plurality of time nodes after the operation. The application effectively improves the accuracy of obtaining the death risk prediction results at a plurality of time nodes after the operation.

Description

Technical Field

[0001] This application relates to the technical field of medical image processing and analysis, specifically to a method and system for predicting the risk of death in patients with peripheral artery disease. Background Technology

[0002] Peripheral artery disease (PAD) is the third most common manifestation of atherosclerosis, affecting more than 230 million people worldwide. Patients with PAD face an extremely high risk of major adverse limb events (MALE) and death. Current clinical guidelines (such as the Global Vascular Guidelines) strongly recommend a comprehensive risk assessment before PAD revascularization surgery to facilitate the development of postoperative follow-up plans by healthcare professionals.

[0003] However, current technology can only rely on doctors' clinical experience to judge the probability of death of patients for a period of time after PAD vascular reconstruction surgery. It is greatly affected by doctors' clinical experience and it is difficult to accurately assess the postoperative risk of patients at different time points after PAD vascular reconstruction surgery, which is not conducive to medical staff to scientifically formulate accurate postoperative follow-up and re-examination plans. Summary of the Invention

[0004] The purpose of this application is to overcome the shortcomings and deficiencies of the prior art and provide a method and system for predicting the risk of death in patients with peripheral artery disease.

[0005] The first aspect of this application provides a method for predicting the risk of death in patients with peripheral artery disease, including: Obtain cross-sectional images of the largest soft tissue in the lower leg before surgery in multiple patients with peripheral artery disease; Based on the cross-sectional images, the calf skeletal muscle area of ​​the multiple patients with peripheral artery disease was obtained; Based on the height data and calf skeletal muscle area of ​​the multiple patients with peripheral artery disease, the calf muscle index of the multiple patients with peripheral artery disease was obtained; The clinical baseline physiological data of the multiple peripheral artery disease patients, the calf muscle index, and postoperative survival data were used as training samples. The gradient booster model is trained based on the training samples to obtain a mortality risk prediction model; The mortality risk prediction model is used to predict the postoperative mortality risk of target patients with target peripheral artery disease, and the mortality risk prediction results are obtained at several postoperative time points.

[0006] As one implementation, the step of obtaining the calf muscle index of the multiple peripheral artery disease patients based on their height data and calf skeletal muscle area includes: The calf muscle index is obtained using the following formula:

[0007] in, The calf muscle index is mentioned. The area of ​​the calf skeletal muscle is described. The height data is as described.

[0008] As one implementation method, the step of predicting postoperative mortality risk in target patients with target peripheral artery disease using the mortality risk prediction model, and obtaining mortality risk prediction results at several postoperative time points, includes: Based on the cross-sectional image of the largest soft tissue in the lower leg and the height data of the target patient before surgery, the target lower leg muscle index was obtained; The clinical baseline physiological data of the target patient, the target calf muscle index, and the several time points are input into the mortality risk prediction model to obtain the mortality risk prediction results of the target patient for the several time points.

[0009] As one implementation, the clinical baseline physiological data includes at least creatinine levels and fasting blood glucose levels.

[0010] As one implementation, the step of acquiring preoperative cross-sectional images of the largest soft tissue in the lower leg of multiple patients with peripheral artery disease includes: Acquired lower extremity computed tomography angiography images of the aforementioned patients with peripheral artery disease; Based on the angiography images, the soft tissue cross-sectional area of ​​multiple axial CT images of the lower leg is obtained; The cross-sectional image of the largest soft tissue in the lower leg is obtained based on the cross-sectional area of ​​the soft tissue in multiple axial CT images.

[0011] As one implementation method, the step of obtaining the cross-sectional image of the largest soft tissue in the lower leg based on the soft tissue cross-sectional area of ​​multiple axial CT images includes: A cross-sectional area fitting function is constructed based on the soft tissue cross-sectional area of ​​multiple axial CT images and their positional information relative to the angiography images. Based on the cross-sectional area fitting function, obtain the cross-sectional image of the largest soft tissue in the lower leg.

[0012] Compared to existing technologies, the mortality risk prediction method for peripheral artery disease patients in this application trains a gradient booster model based on the clinical baseline physiological data of multiple peripheral artery disease patients, the calf skeletal muscle area of ​​the preoperative cross-sectional image of the largest soft tissue in the lower leg, and height data. This results in a mortality risk prediction model, which is then used to predict the postoperative mortality risk of target patients with the target peripheral artery disease. This method can accurately obtain mortality risk prediction results at several postoperative time points, reducing the interference of doctors' clinical experience on the mortality risk prediction results and improving the accuracy of obtaining mortality risk prediction results at several postoperative time points. This is beneficial for medical staff to more accurately formulate postoperative follow-up and re-examination plans.

[0013] A second aspect of this application provides a mortality risk prediction system for patients with peripheral artery disease, comprising: The cross-sectional image acquisition module is used to acquire cross-sectional images of the largest soft tissue in the lower leg before surgery in multiple patients with peripheral artery disease. The calf skeletal muscle area acquisition module is used to acquire the calf skeletal muscle area of ​​the multiple patients with peripheral artery disease based on the cross-sectional image. The calf muscle index acquisition module is used to acquire the calf muscle index of the multiple peripheral artery disease patients based on their height data and calf skeletal muscle area. The training sample acquisition module is used to use the clinical baseline physiological data of the multiple peripheral artery disease patients, the calf muscle index, and postoperative survival data as training samples. The model training module is used to train the gradient booster model based on the training samples to obtain a mortality risk prediction model. The postoperative mortality risk prediction module is used to predict the postoperative mortality risk of target patients with target peripheral artery disease through the mortality risk prediction model, and obtain the mortality risk prediction results at several postoperative time points.

[0014] As one implementation method, the calf muscle index acquisition module obtains the calf muscle index using the following formula:

[0015] in, The calf muscle index is mentioned. The area of ​​the calf skeletal muscle is described. The height data is as described.

[0016] As one implementation, the postoperative mortality risk prediction module includes: The target calf muscle index acquisition unit is used to acquire the target calf muscle index based on the cross-sectional image of the largest soft tissue in the calf and the height data of the target patient before surgery. The mortality risk prediction result acquisition unit is used to input the target patient's clinical baseline physiological data, the target calf muscle index, and the several time points into the mortality risk prediction model to obtain the mortality risk prediction results of the target patient corresponding to the several time points.

[0017] As one implementation, the cross-sectional image acquisition module includes: An angiography image acquisition unit is used to acquire lower extremity computed tomography angiography images of the multiple patients with peripheral artery disease. The soft tissue cross-sectional area acquisition unit is used to acquire the soft tissue cross-sectional area of ​​multiple axial CT images of the lower leg based on the angiography images. The cross-sectional image acquisition unit is used to acquire the cross-sectional image of the largest soft tissue of the lower leg based on the soft tissue cross-sectional area of ​​multiple axial CT images.

[0018] Compared with existing technologies, the peripheral artery disease patient mortality risk prediction system of this application can accurately obtain mortality risk prediction results at several postoperative time points, reduce the interference of doctors' clinical experience on mortality risk prediction results, improve the accuracy of obtaining mortality risk prediction results at several postoperative time points, and help medical staff to more accurately formulate postoperative follow-up and re-examination plans.

[0019] To provide a clearer understanding of this application, the specific embodiments of this application will be described below in conjunction with the accompanying drawings. Attached Figure Description

[0020] Figure 1 A flowchart illustrating a method for predicting mortality risk in patients with peripheral artery disease according to an embodiment of this application; Figure 2 This is a schematic diagram of multiple axial CT images according to one embodiment of this application; Figure 3 This is a schematic diagram illustrating the selection of key features using multiple feature selection algorithms according to one embodiment of this application; Figure 4 This is a schematic diagram of an RF feature selection algorithm according to one embodiment of this application, and a Venn diagram comparing it with various feature selection algorithms; Figure 5 This is a schematic diagram of ROC curves for predicting mortality using LMI, Cr, and Glu in one embodiment of this application. Figure 6 This is a schematic diagram of ROC curves for different muscle indices according to an embodiment of this application. Figure 7 A Spearman correlation heatmap showing the relationship between calf muscle index (LMI) and other clinical baseline characteristics in one embodiment of this application. Figure 8 Heatmaps of C-index (consistency index) for various machine learning models at 1, 3, and 5-year time points; Figure 9 AUC curves of various machine learning models over time, showing the changes in training and test sets. Figure 10 The ROC curve of the training set at 36 months; Figure 11 The ROC curve for the test set at 36 months; Figure 12 For the corresponding Figure 10 The calibration curve; Figure 13 For the corresponding Figure 11 The calibration curve; Figure 14 Kaplan-Meier survival curves for the high LMI group and the low LMI group in one embodiment of this application; Figure 15 This is a schematic diagram of LMI-based grouping according to an embodiment of this application; Figure 16 This is a schematic diagram of the module connections of a peripheral artery disease patient mortality risk prediction system according to an embodiment of this application.

[0021] 101. Cross-sectional image acquisition module; 102. Calf skeletal muscle area acquisition module; 103. Calf muscle index acquisition module; 104. Training sample acquisition module; 105. Model training module; 106. Postoperative mortality risk prediction module. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0023] It should be understood that the described embodiments are merely some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of the embodiments of this application.

[0024] In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. In the description of this application, it should be understood that the terms "first," "second," "third," etc., are used only to distinguish similar objects and are not necessarily used to describe a specific order or sequence, nor should they be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances. The singular forms "a," "the," and "the" used in this application and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. The word "if" as used herein can be interpreted as "when," "when," or "in response to determination."

[0025] Furthermore, in the description of this application, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0026] Please see Figure 1 , Figure 1 A flowchart illustrating a method for predicting mortality risk in patients with peripheral artery disease according to a first embodiment of this application is shown. The method includes the following steps: S1: Obtain cross-sectional images of the largest soft tissue in the lower leg before surgery in multiple patients with peripheral artery disease; S1: The step of acquiring cross-sectional images of the largest soft tissue in the lower leg preoperatively in multiple patients with peripheral artery disease includes: S11: Acquire lower extremity computed tomography (CAT) angiography images of the aforementioned patients with peripheral artery disease; S12: Based on the angiography images, obtain the soft tissue cross-sectional area of ​​multiple axial CT images of the lower leg; Specifically, for each axial CT image, the soft tissue cross-sectional area is obtained by segmenting the image using a preset Hounsfield Unit (HU) threshold. Specifically, the threshold range for skeletal muscle is set to -29 to 150 HU, and the threshold range for intramuscular fat is set to -190 to -30 HU.

[0027] S13: Obtain the cross-sectional image of the largest soft tissue in the lower leg based on the cross-sectional area of ​​the soft tissue in multiple axial CT images.

[0028] Please see Figure 2Multiple axial CT images, including representative images of the L3 and L4 lumbar vertebrae, mid-thigh, and mid-lower leg, are presented. The skeletal muscle regions (shown in green / dark green) and intramuscular fat regions (shown in blue) are marked with different colors in the image. This figure visually demonstrates how this application extracts key morphological features from CTA images, showing the segmentation of the muscle area in the lower leg.

[0029] S2: Based on the cross-sectional image, obtain the calf skeletal muscle area of ​​the multiple patients with peripheral artery disease; S3: Based on the height data of the multiple patients with peripheral artery disease and the calf skeletal muscle area, obtain the calf muscle index of the multiple patients with peripheral artery disease; S4: Use the clinical baseline physiological data of the multiple peripheral artery disease patients, the calf muscle index, and postoperative survival data as training samples.

[0030] The clinical baseline physiological data include at least creatinine (Cr) levels and fasting blood glucose (Glu) levels.

[0031] Please see Figure 3 and Figure 4 This embodiment reveals the process of selecting core predictive features through multi-algorithm cross-validation. The specific steps are as follows: First, all the acquired candidate variables (including patient demographics, comorbidities, multiple clinical blood indicators, and muscle / fat imaging indices at various anatomical levels) were used as the initial feature set.

[0032] Subsequently, four independent machine learning feature selection algorithms were used to evaluate the importance of the initial feature set in parallel: (1) LASSO regression (minimum absolute contraction and selection operator): L1 regularization penalty term is used to remove features with multicollinearity, and features with non-zero regression coefficients are selected to obtain the first set of most important features; (2) mRMR algorithm (maximum relevance and minimum redundancy): find the variable with the highest relevance to the target variable "mortality" in the feature subset and the minimum redundancy between the selected features to obtain the second most important feature set; (3) XGBoost and (4) Random Forest (RF): Using these two tree-based algorithms, the information gain and Gini impurity reduction brought about by each feature during node splitting are calculated, the variable importance is ranked, and the top important features are extracted to obtain the third and fourth most important feature sets.

[0033] Finally, as Figure 4 Part E is shown as a Venn diagram. The intersection of the first, second, third, and fourth most important feature sets selected by the four algorithms was performed. Cross-validation revealed that calf muscle index (LMI), creatinine (Cr), and fasting blood glucose (Glu) were consistently identified as the three most valuable prognostic features across all algorithms. It should be noted that this multi-algorithm fusion screening mechanism effectively eliminates biases introduced by a single algorithm, ensuring the robustness of the selected features.

[0034] Please see Figure 5 , Figure 6 and Figure 7 , Figure 5 The ROC curves for predicting mortality rates using multiple key features are presented, showing that LMI has the largest area under the curve (AUC). Figure 6 The ROC curves of different muscle indices (such as L3-SMI, L3-PMI, etc.) were compared, confirming that LMI is superior to the traditional lumbar muscle index. Figure 7 A Spearman correlation heatmap showing the relationship between calf muscle index (LMI) and other clinical baseline characteristics is presented. This heatmap confirms that LMI is not merely a simple anatomical parameter, but also systematically reflects the patient's overall pathological state. Specifically, LMI is significantly negatively correlated with age and systemic inflammatory markers (such as white blood cell count, WBC), while it is significantly positively correlated with body mass index (BMI), nutritional markers (such as albumin), and anemia markers (such as red blood cell count, RBC, and hemoglobin, Hb). This indicates that the degree of calf muscle atrophy in the affected limb (i.e., low LMI) profoundly reflects the inflammatory storm, nutritional depletion, and ischemic metabolic abnormalities in patients with peripheral artery disease (PAD), explaining its rationale as a core feature for predicting mortality risk from a pathological mechanism perspective.

[0035] S5: Train the gradient booster model based on the training samples to obtain the mortality risk prediction model; The Gradient Boosting Machine (GBM) model refers to the GBM algorithm, a supervised learning algorithm based on ensemble learning. It gradually reduces prediction errors by training multiple weak learners (usually decision trees) sequentially, thereby building a strong prediction model. It belongs to the Boosting class of methods, and its core idea is to optimize the loss function using gradient descent, fitting the residual (negative gradient direction) of the previous model in each iteration.

[0036] Specifically, the input data of the input layer of the gradient booster model includes: LMI values ​​(unit 10-4), Cr values ​​(unit μmol / L), and Glu values ​​(unit mmol / L). The input data is preprocessed, and the preprocessing methods include: missing value handling (filling or removal), outlier identification and correction, normalization / standardization of continuous variables, and one-hot encoding of categorical variables.

[0037] Specifically, the processing layer of the gradient booster model is based on ensemble learning using the GBM algorithm, and the loss function is minimized by iteratively training the decision tree.

[0038] Please see Figure 8 , Figure 9 and Figure 10 Eight machine learning models, including GBM and XGBoost, were trained using training samples. The C-index (consistency index) heatmaps for each model at 1, 3, and 5 years are shown below. Figure 8 As shown, Figure 9 The AUC curve of the training set over time is shown. Figure 10 The AUC curves of the test set over time are shown, demonstrating that the GBM model including LMI has robust and excellent long-term predictive power.

[0039] Please continue reading. Figures 10-13 Detailed performance validation data for the gradient booster model on the training and independent test sets are as follows: Figure 10 and Figure 11 As shown. Among them, Figure 10 and Figure 11 The ROC curves for the training and test sets at 36 months are shown respectively. The AUC values ​​both exceed 0.84, proving that the model has extremely high discriminative power. Figure 12 and Figure 13 The corresponding calibration curves show that the predicted probability highly overlaps with the actual observed probability (close to the diagonal), proving that the mortality risk probability value output by this application is accurate and reliable, and has high clinical reference value.

[0040] In a specific example, the configuration parameters of the gradient boosting machine model are as follows: Distribution selection: Cox proportional hazards model (distribution='coxph'); Number of decision trees: 1200 (n.trees=1200); Maximum depth of a single tree: 1 (interaction.depth=1, i.e., a stubby tree weak learner to prevent overfitting); Minimum leaf node sample size: 3 (n.minobsinnode=3); Learning rate (step size): 0.01 (shrinkage=0.01); Cross-validation folds: 10 (cv.folds=10, used to optimize parameters and evaluate generalization ability); Parallel computing core count: 10 (n.cores=10, to improve computational efficiency).

[0041] Specifically, during the training process, multiple decision trees are iteratively constructed using training samples. Each tree learns the residuals from the previous round, gradually optimizing the overall prediction performance. The Cox partial likelihood loss function is used to address the survival time issue involving censored data. 10-fold cross-validation is used to automatically select the optimal number of trees and parameters.

[0042] The final prediction output of the Gradient Boosting Machine (GBM) model can be represented by the following additive model function:

[0043] in, For the process After rounds of iterative training, the model adapts to the input. The output is the postoperative mortality risk prediction result (expressed as logarithmic hazard ratio or relative risk). This is the input feature vector. During the training phase, Corresponding to the characteristic data of each patient in the training sample, including the patient's clinical baseline physiological data (creatinine level, fasting blood glucose level) and calf muscle index (LMI); in the prediction phase, Target characteristic data corresponding to the target patient; This represents the total number of iterations (i.e., the total number of decision trees, for example, 1200). For the first The basic weak learner (usually a regression decision tree) generated by rounds of training; For the first The weights (step size / learning rate) corresponding to each decision tree.

[0044] It is understandable that the mechanism by which training samples play a role in model training is as follows: The training process of this model is an iterative process of minimizing a loss function based on training samples. The training sample set is configured to include patient features. And real postoperative survival data (including survival status and time). In each iteration In the middle, the new decision tree Instead of directly fitting the actual survival data, it fits the previous round of prediction model. The residuals generated on the training samples (i.e., the negative gradient direction of the loss function). By repeatedly calculating the residuals using the training samples and generating new decision trees to compensate for them, the model continuously corrects previous prediction errors, ultimately achieving... The weighted sum of the outputs of each tree yields a mortality risk prediction model that accurately fits the true survival distribution of the training sample data. .

[0045] S6: The mortality risk prediction model is used to predict the postoperative mortality risk of target patients with target peripheral artery disease, and the mortality risk prediction results at several postoperative time points are obtained.

[0046] Specifically, several time points include 12 months, 24 months, 36 months, 48 ​​months, 60 months, etc.

[0047] Please continue reading. Figure 14 and Figure 15 , Figure 14 This is a schematic diagram of Kaplan-Meier survival curves for the high LMI group and the low LMI group. Figure 15 This is a schematic diagram of LMI-based grouping. Combined with... Figure 14 and Figure 15 This application can also group target patients based on their target calf muscle index and LMI threshold; for example, patients with a target calf muscle index greater than or equal to the LMI threshold are assigned to the low-risk group, and patients with a target calf muscle index less than the LMI threshold are assigned to the high-risk group, and generate corresponding survival probability curves to assist doctors in developing postoperative follow-up plans: such as shortening the follow-up interval and strengthening nutritional intervention for high-risk patients.

[0048] Compared to existing technologies, the mortality risk prediction method for peripheral artery disease patients in this application trains a gradient booster model based on the clinical baseline physiological data of multiple peripheral artery disease patients, the calf skeletal muscle area of ​​the preoperative cross-sectional image of the largest soft tissue in the lower leg, and height data. This results in a mortality risk prediction model, which is then used to predict the postoperative mortality risk of target patients with the target peripheral artery disease. This method can accurately obtain mortality risk prediction results at several postoperative time points, reducing the interference of doctors' clinical experience on the mortality risk prediction results and improving the accuracy of obtaining mortality risk prediction results at several postoperative time points. This is beneficial for medical staff to more accurately formulate postoperative follow-up and re-examination plans.

[0049] In a feasible embodiment, the step of obtaining the calf muscle index of the plurality of peripheral artery disease patients based on their height data and calf skeletal muscle area includes: The calf muscle index is obtained using the following formula:

[0050] in, The calf muscle index is mentioned. The area of ​​the calf skeletal muscle is described. The height data is as described.

[0051] In a feasible embodiment, the step of predicting postoperative mortality risk in target patients with target peripheral artery disease using the mortality risk prediction model, and obtaining mortality risk prediction results at several postoperative time points, includes: Based on the cross-sectional image of the largest soft tissue in the lower leg and the height data of the target patient before surgery, the target lower leg muscle index was obtained; The clinical baseline physiological data of the target patient, the target calf muscle index, and the several time points are input into the mortality risk prediction model to obtain the mortality risk prediction results of the target patient for the several time points.

[0052] In a feasible embodiment, step S13: obtaining the cross-sectional image of the largest soft tissue of the lower leg based on the soft tissue cross-sectional area of ​​multiple axial CT images includes: S131: Construct a cross-sectional area fitting function based on the soft tissue cross-sectional area of ​​multiple axial CT images and the positional information relative to the angiography image; S132: Obtain the cross-sectional image of the largest soft tissue of the lower leg according to the cross-sectional area fitting function.

[0053] In summary, this application has the following effects: Significantly Improved Prediction Accuracy: This invention demonstrates that the predictive power of LMI (AUC=0.71) is significantly superior to the traditional abdominal muscle index. When combined with the GBM model, in the independent validation set, its C-index for 1 year, 3 years, and 5 years reached 0.847, 0.809, and 0.823, respectively. This is particularly evident in long-term predictions over 36 months (e.g., Figure 8 As shown in the figure, the model exhibits extremely high discrimination (AUC=0.844) and excellent calibration (low Brier score), which is superior to existing clinical scoring systems.

[0054] Precisely reflecting the pathological mechanism: Unlike systemic indicators, LMI directly reflects the degree of ischemic muscle atrophy in the affected limbs of patients with PAD. Studies have shown that LMI has a stronger correlation with inflammatory markers, nutritional status, and anemia markers than the lumbar muscle index, making it a "biomarker" for the severity of PAD.

[0055] Highly clinically applicable and non-invasive: This method requires no additional examination equipment and can be completed using only routine preoperative CTA images. The selected features (LMI, creatinine, blood glucose) are all clinically readily available data, facilitating integration and promotion within hospital information systems (HIS).

[0056] Optimize resource allocation: Validated by decision curve analysis (DCA) (as described in the instruction manual), the model has a positive net clinical benefit across a wide range of threshold probabilities, which can help doctors identify high-risk groups that truly need intensive intervention and avoid wasting medical resources.

[0057] Please see Figure 16 A second embodiment of this application provides a mortality risk prediction system for patients with peripheral artery disease, comprising: The cross-sectional image acquisition module is used to acquire cross-sectional images of the largest soft tissue in the lower leg before surgery in multiple patients with peripheral artery disease. The calf skeletal muscle area acquisition module is used to acquire the calf skeletal muscle area of ​​the multiple patients with peripheral artery disease based on the cross-sectional image. The calf muscle index acquisition module is used to acquire the calf muscle index of the multiple peripheral artery disease patients based on their height data and calf skeletal muscle area. The training sample acquisition module is used to use the clinical baseline physiological data of the multiple peripheral artery disease patients, the calf muscle index, and postoperative survival data as training samples. The model training module is used to train the gradient booster model based on the training samples to obtain a mortality risk prediction model. The postoperative mortality risk prediction module is used to predict the postoperative mortality risk of target patients with target peripheral artery disease through the mortality risk prediction model, and obtain the mortality risk prediction results at several postoperative time points.

[0058] In one feasible embodiment, the calf muscle index acquisition module obtains the calf muscle index using the following formula:

[0059] in, The calf muscle index is mentioned. The area of ​​the calf skeletal muscle is described. The height data is as described.

[0060] In one feasible embodiment, the postoperative mortality risk prediction module includes: The target calf muscle index acquisition unit is used to acquire the target calf muscle index based on the cross-sectional image of the largest soft tissue in the calf and the height data of the target patient before surgery. The mortality risk prediction result acquisition unit is used to input the target patient's clinical baseline physiological data, the target calf muscle index, and the several time points into the mortality risk prediction model to obtain the mortality risk prediction results of the target patient corresponding to the several time points.

[0061] In one feasible embodiment, the cross-sectional image acquisition module includes: An angiography image acquisition unit is used to acquire lower extremity computed tomography angiography images of the multiple patients with peripheral artery disease. The soft tissue cross-sectional area acquisition unit is used to acquire the soft tissue cross-sectional area of ​​multiple axial CT images of the lower leg based on the angiography images. The cross-sectional image acquisition unit is used to acquire the cross-sectional image of the largest soft tissue of the lower leg based on the soft tissue cross-sectional area of ​​multiple axial CT images.

[0062] It should be noted that the peripheral artery disease patient mortality risk prediction system provided in the second embodiment of this application is only illustrated by the above-described division of functional modules when executing the peripheral artery disease patient mortality risk prediction method. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. Furthermore, the peripheral artery disease patient mortality risk prediction system provided in the second embodiment of this application and the peripheral artery disease patient mortality risk prediction method in the first embodiment of this application belong to the same concept, and its implementation process is detailed in the method embodiment, which will not be repeated here.

[0063] The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and 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 modules can be selected to achieve the purpose of this application according to actual needs. Those skilled in the art can understand and implement this without any inventive effort.

[0064] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0065] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 The computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function selected in one or more boxes.

[0066] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function selected in one or more boxes.

[0067] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0068] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0069] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0070] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0071] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for predicting mortality risk in patients with peripheral artery disease, characterized in that, include: Obtain cross-sectional images of the largest soft tissue in the lower leg before surgery in multiple patients with peripheral artery disease; Based on the cross-sectional images, the calf skeletal muscle area of ​​the multiple patients with peripheral artery disease was obtained; Based on the height data and calf skeletal muscle area of ​​the multiple patients with peripheral artery disease, the calf muscle index of the multiple patients with peripheral artery disease was obtained; The clinical baseline physiological data of the multiple peripheral artery disease patients, the calf muscle index, and postoperative survival data were used as training samples. The gradient booster model is trained based on the training samples to obtain a mortality risk prediction model; The mortality risk prediction model is used to predict the postoperative mortality risk of target patients with target peripheral artery disease, and the mortality risk prediction results are obtained at several postoperative time points.

2. The method for predicting mortality risk in patients with peripheral artery disease according to claim 1, characterized in that, The step of obtaining the calf muscle index of the multiple peripheral artery disease patients based on their height data and calf skeletal muscle area includes: The calf muscle index is obtained using the following formula: in, The calf muscle index is mentioned. The area of ​​the calf skeletal muscle is described. The height data is as described.

3. The method for predicting mortality risk in patients with peripheral artery disease according to claim 1, characterized in that, The steps of predicting postoperative mortality risk in target patients with target peripheral artery disease using the aforementioned mortality risk prediction model, and obtaining mortality risk prediction results at several postoperative time points, include: Based on the cross-sectional image of the largest soft tissue in the lower leg and the height data of the target patient before surgery, the target lower leg muscle index was obtained; The clinical baseline physiological data of the target patient, the target calf muscle index, and the several time points are input into the mortality risk prediction model to obtain the mortality risk prediction results of the target patient for the several time points.

4. The method for predicting mortality risk in patients with peripheral artery disease according to claim 1, characterized in that, The clinical baseline physiological data include at least creatinine levels and fasting blood glucose levels.

5. The method for predicting mortality risk in patients with peripheral artery disease according to claim 1, characterized in that, The step of acquiring preoperative cross-sectional images of the largest soft tissue in the lower leg of multiple patients with peripheral artery disease includes: Acquired lower extremity computed tomography angiography images of the aforementioned patients with peripheral artery disease; Based on the angiography images, the soft tissue cross-sectional area of ​​multiple axial CT images of the lower leg is obtained; The cross-sectional image of the largest soft tissue in the lower leg is obtained based on the cross-sectional area of ​​the soft tissue in multiple axial CT images.

6. The method for predicting mortality risk in patients with peripheral artery disease according to claim 5, characterized in that, The step of obtaining the cross-sectional image of the largest soft tissue in the lower leg based on the cross-sectional area of ​​soft tissue from multiple axial CT images includes: A cross-sectional area fitting function is constructed based on the soft tissue cross-sectional area of ​​multiple axial CT images and their positional information relative to the angiography images. Based on the cross-sectional area fitting function, obtain the cross-sectional image of the largest soft tissue in the lower leg.

7. A mortality risk prediction system for patients with peripheral artery disease, characterized in that, include: The cross-sectional image acquisition module is used to acquire cross-sectional images of the largest soft tissue in the lower leg before surgery in multiple patients with peripheral artery disease. The calf skeletal muscle area acquisition module is used to acquire the calf skeletal muscle area of ​​the multiple patients with peripheral artery disease based on the cross-sectional image. The calf muscle index acquisition module is used to acquire the calf muscle index of the multiple peripheral artery disease patients based on their height data and calf skeletal muscle area. The training sample acquisition module is used to use the clinical baseline physiological data of the multiple peripheral artery disease patients, the calf muscle index, and postoperative survival data as training samples. The model training module is used to train the gradient booster model based on the training samples to obtain a mortality risk prediction model. The postoperative mortality risk prediction module is used to predict the postoperative mortality risk of target patients with target peripheral artery disease through the mortality risk prediction model, and obtain the mortality risk prediction results at several postoperative time points.

8. The peripheral artery disease patient mortality risk prediction system according to claim 7, characterized in that, The calf muscle index acquisition module obtains the calf muscle index using the following formula: in, The calf muscle index is mentioned. The area of ​​the calf skeletal muscle is described. The height data is as described.

9. The peripheral artery disease patient mortality risk prediction system according to claim 7, characterized in that, The postoperative mortality risk prediction module includes: The target calf muscle index acquisition unit is used to acquire the target calf muscle index based on the cross-sectional image of the largest soft tissue in the calf and the height data of the target patient before surgery. The mortality risk prediction result acquisition unit is used to input the target patient's clinical baseline physiological data, the target calf muscle index, and the several time points into the mortality risk prediction model to obtain the mortality risk prediction results of the target patient corresponding to the several time points.

10. The peripheral artery disease patient mortality risk prediction system according to claim 7, characterized in that, The cross-sectional image acquisition module includes: An angiography image acquisition unit is used to acquire lower extremity computed tomography angiography images of the multiple patients with peripheral artery disease. The soft tissue cross-sectional area acquisition unit is used to acquire the soft tissue cross-sectional area of ​​multiple axial CT images of the lower leg based on the angiography images. The cross-sectional image acquisition unit is used to acquire the cross-sectional image of the largest soft tissue of the lower leg based on the soft tissue cross-sectional area of ​​multiple axial CT images.

Images