Methods and Systems for Constructing PICC Catheter-Related Thrombosis Risk Prediction Models

Abstract

This invention relates to the field of medical risk modeling technology, specifically to a method and system for constructing a PICC catheter-related thrombosis risk prediction model. The method involves: acquiring relevant indicators of PICC catheter patients; determining the thrombosis risk discrimination ability of different feature values ​​under the relevant indicators based on the ability of these indicators to distinguish thrombosis risk; correcting the thrombosis risk discrimination ability based on the differences in different feature values ​​under the same relevant indicator, thus obtaining the classification ability of the relevant indicators; determining the classification cost of the relevant indicators by utilizing the similarity between the label distribution characteristics before and after classification; and constructing corresponding PICC-related thrombosis prediction models based on the classification costs of different relevant indicators. This invention selects the optimal attribute of the decision tree at the current node position based on the classification cost, thereby completing the construction of the decision tree and making the judgment of PICC thrombosis risk more accurate.

Description

Technical Field

[0001] This invention relates to the field of medical risk modeling technology, specifically to a method and system for constructing a PICC catheter-related thrombosis risk prediction model. Background Technology

[0002] A peripherally inserted central catheter (PICC) is a catheter inserted via puncture into the basilic vein, median cubital vein, cephalic vein, brachial vein, or external jugular vein (in newborns, it can also be inserted via the great saphenous vein, temporal vein, or posterior auricular vein). The tip of the catheter is located in the superior or inferior vena cava. Infused medications can be rapidly circulated through this catheter into a central venous vein with high blood flow, providing a medium- to long-term intravenous therapy route. This reduces the damage to peripheral blood vessels caused by prolonged infusion or the administration of hypertonic or irritating drugs, thus avoiding patient suffering and ensuring the smooth administration of important treatments such as parenteral nutrition and chemotherapy, thereby improving patient satisfaction. PICC-related thrombosis occurs when a blood clot forms on the inner wall of the vessel containing the PICC and on the catheter's surface after insertion. Most patients with PICC-related thrombosis do not have obvious clinical symptoms, which may manifest as occasional swelling or numbness in the limb on the insertion side. These symptoms are easily overlooked, leading to further thrombosis and potentially pulmonary embolism, which can be life-threatening. Postthrombotic syndrome resulting from PICC catheters impairs venous valve function, leading to limb pain, swelling, and functional impairment, affecting quality of life and delaying hospitalization. Therefore, risk prediction for PICC catheter-related thrombosis is crucial. Based on the prediction results, the thrombosis risk is assessed, and appropriate preventative measures are taken to ensure patient safety.

[0003] Currently, traditional machine learning algorithms such as random forests are commonly used to construct PICC catheter-related thrombosis prediction models. However, using traditional random forest algorithms to construct PICC catheter-related thrombosis prediction models has the following problems: Because the incidence of PICC catheter-related thrombosis is relatively low, the training set includes few cases, resulting in inaccurate prediction models and poor applicability; the timing of thrombosis is not analyzed as a variable, preventing nurses from focusing on high-incidence periods of PICC catheter-related thrombosis, leading to poor management efficiency. Summary of the Invention

[0004] To address the technical problems inherent in the construction of decision trees using traditional random forest algorithms, such as inaccurate prediction of PICC catheter-related thrombosis and inability to focus on high-incidence periods, thus affecting the thrombosis prediction capability of random forests, this invention aims to provide a method and system for constructing a PICC catheter-related thrombosis risk prediction model. The specific technical solution adopted is as follows:

[0005] In a first aspect, embodiments of the present invention provide a method for constructing a PICC catheter-related thrombosis risk prediction model, the method comprising:

[0006] Relevant indicators for patients with PICC catheters;

[0007] Based on the ability of the relevant indicators to distinguish thrombosis risk, the ability of different characteristic values ​​under the relevant indicators to distinguish thrombosis risk is determined.

[0008] Based on the differences in different feature values ​​under the same relevant indicator, the ability to distinguish thrombosis risk is corrected to obtain the classification ability of the relevant indicator.

[0009] Combining the thrombosis risk discrimination ability and the classification ability, the classification cost of relevant indicators is determined by utilizing the similarity between the label distribution characteristics before and after the classification of relevant indicators.

[0010] Based on the cost of classifying different related indicators, a corresponding PICC-related thrombosis prediction model is constructed.

[0011] Furthermore, the step of determining the ability to distinguish thrombosis risk based on the relevant indicators, and determining the ability to distinguish thrombosis risk for different feature values ​​under the relevant indicators, includes:

[0012] The characteristic values ​​of relevant indicators are labeled with corresponding PICC thrombosis risk tags;

[0013] Based on the types of PICC thrombosis risk labels corresponding to each feature value under the same type of related indicators and the proportion of different types of PICC thrombosis risk labels, the category characteristics of each feature value under the same type of related indicators in the corresponding different types of PICC thrombosis risk labels are determined.

[0014] A data feature vector is constructed for each feature value corresponding to the category features under all types of PICC thrombosis risk labels, based on each feature value under the same type of related indicators.

[0015] Based on the data feature vector, the ability to distinguish thrombosis risk under different feature values ​​of relevant indicators is determined.

[0016] Furthermore, the determination of the category characteristics of each feature value under the same type of related indicator in corresponding PICC thrombosis risk labels based on the types of PICC thrombosis risk labels corresponding to each feature value under the same type of related indicator and the proportion of different types of PICC thrombosis risk labels includes:

[0017] Use any feature value under any relevant indicator as a reference value; obtain the number of types of PICC thrombosis risk labels corresponding to the same reference value as the feature dispersion.

[0018] Use any PICC thrombosis risk label as a reference thrombosis risk label; calculate the total proportion of other PICC thrombosis risk labels besides the reference thrombosis risk label corresponding to the same reference value, and use it as the feature similarity proportion.

[0019] Based on the feature dispersion and the feature similarity ratio, the category feature of the reference value in the corresponding reference thrombosis risk label is determined.

[0020] Furthermore, the step of determining the ability to distinguish thrombosis risk based on the data feature vector for different feature values ​​under relevant indicators includes:

[0021] Use any characteristic value under any relevant indicator as a reference value;

[0022] Based on the data feature vectors of relevant indicators corresponding to the reference values ​​and the number of reference values, the thrombosis risk discrimination ability of the reference values ​​is determined.

[0023] Furthermore, the determination of the thrombosis risk discrimination ability of the reference value based on the data feature vector of the relevant indicators corresponding to the reference value and the number of reference values ​​includes:

[0024] For each element value in the data feature vector of the relevant indicator corresponding to the reference value, the ratio of the number of element values ​​to the number of parameter values ​​is calculated to form an adjustment feature vector.

[0025] Based on the number of types of PICC thrombosis risk labels, the adjusted feature vector is modified to obtain the modified feature vector;

[0026] The modulus of the modified feature vector is used as a reference value to distinguish the risk of thrombosis.

[0027] Furthermore, the ability to differentiate thrombosis risk based on the differences in different feature values ​​under the same relevant indicator is corrected to obtain the classification ability of the relevant indicator, including:

[0028] Using any relevant indicator as a reference relevant indicator, the importance of the classification ability is determined based on the distribution of the thrombosis risk discrimination ability corresponding to different feature values ​​under the same reference relevant indicator.

[0029] Representative values ​​for distinguishing thrombosis risk were selected from the thrombosis risk discrimination capabilities corresponding to different characteristic values ​​of relevant indicators.

[0030] Based on the importance of the relevant indicators, the representative values ​​of the thrombosis risk differentiation ability corresponding to the relevant indicators are adjusted to obtain the classification ability of the relevant indicators.

[0031] Furthermore, the step of selecting representative values ​​for distinguishing thrombosis risk from the thrombosis risk discrimination capabilities corresponding to different characteristic values ​​of relevant reference indicators includes:

[0032] The maximum value among the different characteristic values ​​of the reference relevant indicator corresponding to the thrombosis risk discrimination ability is selected as the representative value of the thrombosis risk discrimination ability of the reference relevant indicator.

[0033] Furthermore, the step of determining the partitioning cost of relevant indicators by utilizing the similarity between the label distribution characteristics before and after the partitioning includes:

[0034] By comparing the thrombosis risk discrimination ability of relevant indicators before classification with the thrombosis risk discrimination ability of different feature values ​​of relevant indicators after classification, the classification improvement value is determined.

[0035] By combining the partitioning improvement value and the partitioning capability, the partitioning cost of the relevant indicators is determined.

[0036] Furthermore, the step of constructing a corresponding PICC-related thrombosis prediction model based on the classification costs of different related indicators includes:

[0037] Based on the classification costs of different types of relevant indicators in PICC catheter patients, multiple decision trees are constructed to obtain a random forest model, which serves as the corresponding PICC-related thrombosis prediction model.

[0038] Secondly, a system for constructing a PICC catheter-related thrombosis risk prediction model is provided, the system comprising:

[0039] The data acquisition module obtains relevant indicators for patients with PICC catheters.

[0040] The capability determination module determines the ability to distinguish thrombosis risk for different feature values ​​under the relevant indicators based on the ability of the relevant indicators to distinguish thrombosis risk.

[0041] The correction module corrects the ability to distinguish thrombosis risk based on the differences in different feature values ​​under the same relevant indicator, thereby obtaining the classification ability of the relevant indicator;

[0042] The cost determination module, combining the thrombosis risk discrimination ability and the classification ability, uses the similarity between the label distribution characteristics before and after the classification of relevant indicators to determine the classification cost of relevant indicators.

[0043] The model building module constructs corresponding PICC-related thrombosis prediction models based on the classification costs of different types of relevant indicators.

[0044] Thirdly, a server is provided, including a memory and a processor. The memory is used to store executable program code, and the processor is used to call and run the executable program code from the memory, causing the device to perform the methods of the first aspect or any possible implementation thereof.

[0045] Fourthly, a computer program product is provided, comprising: computer program code, which, when run on a computer, causes the computer to perform the methods described in the first aspect or any possible implementation thereof.

[0046] Fifthly, a computer-readable storage medium is provided that stores computer program code, which, when executed on a computer, causes the computer to perform the methods described in the first aspect or any possible implementation thereof.

[0047] The embodiments of the present invention have at least the following beneficial effects:

[0048] Relevant indicators for PICC catheter patients are obtained. Since different indicators have varying degrees of importance and impact on thrombosis risk, the ability to classify these indicators needs to be assessed based on their ability to distinguish thrombosis risk. Therefore, the ability of different characteristic values ​​under these indicators to distinguish thrombosis risk is further determined. Because relying solely on the distribution characteristics of a single characteristic value under the same indicator to determine the ability to distinguish PICC thrombosis risk has low accuracy, the ability to distinguish thrombosis risk is corrected by examining the differences in different characteristic values ​​under the same indicator. This avoids the situation where characteristic values ​​under the same indicator show differences in their distribution from the average distribution, but such concentrated distribution under a single characteristic value of the same indicator is normal and indicates a low correlation with PICC thrombosis risk, and cannot be used as the primary basis for judging thrombosis risk. The similarity between the label distribution characteristics before and after the classification of relevant indicators is used to determine the classification cost of the relevant indicators. Based on the classification costs of different relevant indicators, corresponding PICC-related thrombosis prediction models are constructed. The present invention selects the best attribute of the decision tree at the current node position based on the partitioning cost, thereby completing the construction of the decision tree and making the judgment of PICC thrombosis risk more accurate. Attached Figure Description

[0049] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1A flowchart illustrating a method for constructing a PICC catheter-related thrombosis risk prediction model according to an embodiment of the present invention;

[0051] Figure 2 Another flowchart of a method for constructing a PICC catheter-related thrombosis risk prediction model provided in one embodiment of the present invention;

[0052] Figure 3 This is another flowchart illustrating a method for constructing a PICC catheter-related thrombosis risk prediction model, as provided in an embodiment of the present invention.

[0053] Figure 4 A flowchart illustrating another method for constructing a PICC catheter-related thrombosis risk prediction model, as provided in an embodiment of the present invention.

[0054] Figure 5 This is a schematic diagram of the composition structure of a system for constructing a PICC catheter-related thrombosis risk prediction model provided in this application embodiment;

[0055] Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0056] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method and system for constructing a PICC catheter-related thrombosis risk prediction model proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0058] This invention provides a method for constructing a PICC catheter-related thrombosis risk prediction model and a specific implementation method for the system. This method is applicable to PICC-related thrombosis prediction model scenarios. By analyzing the importance of relevant indicators in PICC thrombosis characteristics and the differentiation of thrombosis risk labels before and after segmentation, this method determines the appropriate segmentation criteria for the decision tree used for thrombosis risk assessment, thereby improving the accuracy of the thrombosis risk model assessment.

[0059] The following description, in conjunction with the accompanying drawings, details the specific scheme of the method and system for constructing a PICC catheter-related thrombosis risk prediction model provided by this invention.

[0060] Please see Figure 1 The diagram illustrates a flowchart of a method for constructing a PICC catheter-related thrombosis risk prediction model according to an embodiment of the present invention. The method includes the following steps:

[0061] Step S100: Obtain relevant indicators for patients with PICC catheters.

[0062] Collect relevant information on patients with indwelling PICC catheters, including demographics, disease information, medication information, catheter placement, clinical laboratory information, maintenance, and PICC catheter-related thrombosis information.

[0063] Demographic information includes the patient's gender, age, BMI, and education level.

[0064] Disease information includes the patient's primary diagnosis upon admission, tumor stage, whether the patient has hypertension, diabetes, coronary heart disease, deep vein thrombosis and thrombophlebitis, history of deep vein catheterization, and reason for catheter placement.

[0065] Medication information includes the chemotherapy cycle and the types of chemotherapy drugs.

[0066] Clinical laboratory information includes values ​​for white blood cells, hemoglobin, platelets, D-dimer, prothrombin time, fibrinogen, activated partial thromboplastin time, albumin, and C-reactive protein.

[0067] The catheter placement information includes the PICC catheter model, number of lumens, orifice type, placement method, catheter / vessel diameter ratio, puncture site, punctured vessel, number of punctures, catheter tip positioning method, and catheter tip position.

[0068] Maintenance information includes catheter fixation method, whether the sealing fluid is pre-filled, type of sealing fluid, and type of connector.

[0069] Information on PICC catheter-related thrombosis includes whether PICC catheter-related thrombosis occurred and the date of occurrence.

[0070] Based on univariate analysis, Cox survival analysis, DeepSurv, DeepHit and other algorithms, a PICC-related thrombosis prediction model was constructed and the weight of each factor was obtained.

[0071] The relevant indicators for PICC catheter patients referred to here include: age, gender, body mass index, D-dimer level, hemoglobin, demographic information, disease information, medication information, clinical laboratory information, catheter placement information, and maintenance information, as well as other indicators related to body shape. These indicators can be collected through the hospital's medical record system. These indicators can also be understood as the PICC thrombosis characteristics of PICC catheter patients. Essentially, they refer to the patient's age, gender, body mass index, D-dimer level, hemoglobin level, demographic information, disease information, medication information, clinical laboratory information, catheter placement information, and maintenance information as relevant indicators, but categorized into different types, such as age as one indicator, gender as another, etc. It should be noted that, for example, if gender is a characteristic described in words, it can be replaced with a pre-set label value, or in other embodiments, only numerically represented indicators may be considered, excluding those described in words.

[0072] Each relevant indicator has its own corresponding feature value, and the data is diverse, meaning that each relevant indicator has multiple feature values ​​of different sizes.

[0073] As an embodiment of the present invention, relevant indicators of PICC catheter patients can be input into the deepsurv model to obtain PICC thrombosis risk labels; the deepsurv model is the PICC-related thrombosis prediction model.

[0074] Specifically: statistical analysis was used to select significantly relevant variables, including but not limited to: gender, age, height, education level, initial diagnosis at admission, activated partial thromboplastin time (APTT), albumin level, catheter fixation, type of sealing solution, type of connector, type of catheter opening, dressing, and certain chemotherapy drugs (such as platinum-based drugs, anthracyclines, alkylating agents, and plant alkaloids).

[0075] Relevant indicators are input into the Deepsurv model, along with the patient's PICC thrombosis risk assessment results, i.e., PICC thrombosis risk labels. These PICC thrombosis risk labels are then used as category labels for the corresponding data. By using the PICC thrombosis risk labels obtained from the Deepsurv model as labels for the decision tree data, a supervised learning model of decision trees is used to construct a PICC-related thrombosis prediction model.

[0076] The reasons for using the evaluation results of the DeepSurv model in the decision tree model are as follows: 1. In the first stage, the DeepSurv model was used for risk assessment, extracting preliminary risk features. In the second stage, the decision tree model was used to further learn and optimize these risk features to improve the accuracy of risk assessment. 2. In this way, secondary optimization of risk assessment can be achieved, that is, based on the DeepSurv model, the decision tree model is used to further refine and improve the accuracy of risk prediction. 3. Decision trees are supervised learning models, and by learning classification labels, they can improve the accuracy and robustness of PICC thrombosis risk assessment.

[0077] Then, the PICC thrombosis risk label for each PICC catheter patient is determined, thereby determining the PICC thrombosis risk label L corresponding to each feature value V of each relevant indicator. Thus, the feature value of the relevant indicator and the corresponding PICC thrombosis risk label can be expressed as X={V,L}; where V is the feature value of the relevant indicator; and L is the PICC thrombosis risk label corresponding to V.

[0078] Step S200: Based on the ability of the relevant indicators to distinguish thrombosis risk, determine the ability of different feature values ​​under the relevant indicators to distinguish thrombosis risk.

[0079] By assessing the ability and accuracy of various relevant indicators representing thrombosis characteristics in classifying thrombosis risk, we obtain the thrombosis risk discrimination capability of the relevant indicators. By determining whether the accuracy of thrombosis risk judgment is improved before and after classification, we can ascertain the classification cost of the relevant indicators and select the attributes of the decision tree based on the classification cost.

[0080] Data obtained from hospitals, especially patient data, is inherently biased. A significant portion of the population carries a risk of PICC line thrombosis, while a smaller portion does not. This results in an imbalanced sample size for relevant indicators. Furthermore, different PICC line thrombosis-related indicators have varying degrees of importance in assessing thrombosis risk, leading to inaccurate decision tree models and impacting risk evaluation.

[0081] Therefore, by utilizing the accuracy of different relevant indicators in classifying thrombosis risk and the changes between normal data and data with thrombosis risk after attribute classification, the thrombosis risk discrimination ability of the current relevant indicators can be determined.

[0082] When constructing a PICC-related thrombosis prediction model, the traditional decision tree approach for selecting relevant indicators relies on purity, represented by entropy. However, the samples obtained are clinical data from hospitals, meaning most users have some risk of PICC thrombosis, leaving only a small number without this risk. Furthermore, the relevant indicators from PICC catheter patients are complex and diverse; not every indicator serves as the primary basis for judging thrombosis risk. Therefore, the attribute selection method of traditional decision tree algorithms may introduce significant errors in constructing a PICC catheter-related thrombosis risk prediction model. Consequently, in learning thrombosis-related indicators, highly pure indicators may represent common population traits and cannot be used as the primary basis for distinguishing thrombosis risk, thus affecting the construction of the decision tree.

[0083] On the other hand, different indicators representing thrombosis characteristics have varying degrees of importance. For example, while factors such as age have some influence on the risk of PICC thrombosis—the older one gets, the weaker one's immune system, and the greater the risk of thrombosis—age cannot be used as the primary basis for thrombosis risk. It is still necessary to look at various hemoglobin and other directly related indicators associated with PICC thrombosis. Therefore, it is necessary to judge the ability to classify relevant indicators based on the ability to distinguish thrombosis risk.

[0084] Therefore, in this embodiment of the invention, the ability of different characteristic values ​​under different relevant indicators to distinguish thrombosis risk is determined.

[0085] In some embodiments, based on the ability of the relevant indicators to distinguish thrombosis risk, the ability of different characteristic values ​​under the relevant indicators to distinguish thrombosis risk is determined, that is, step 200 above can be achieved by... Figure 2 The steps shown are to be implemented as follows:

[0086] Step S210: Label the characteristic values ​​of the relevant indicators with the corresponding PICC thrombosis risk labels.

[0087] It should be noted that the PICC thrombosis risk label corresponding to the feature value of this relevant indicator is the PICC thrombosis risk label predicted by the deepsurv model for each PICC catheter patient. That is, if the PICC thrombosis risk label corresponding to patient c1 is c2, then the feature value of all relevant indicators corresponding to patient c1 is the PICC thrombosis risk label c2 corresponding to patient c1.

[0088] Step S220: Based on the types of PICC thrombosis risk labels corresponding to each feature value under the same type of related indicators and the proportion of different types of PICC thrombosis risk labels, determine the category characteristics of each feature value under the same type of related indicators in the corresponding different types of PICC thrombosis risk labels.

[0089] In some possible implementations, by analyzing the number of PICC thrombosis risk labels and the proportion of PICC thrombosis risk labels corresponding to each feature value under the same type of related indicator, the category characteristics of each feature value under the same type of related indicator in different types of PICC thrombosis risk labels can be quickly calculated. That is, the above step S220 can be achieved by the following steps S221, S222 and S223 (not shown in the figure):

[0090] Step S221: Take any feature value under any relevant indicator as a reference value; obtain the number of types of PICC thrombosis risk labels corresponding to the same reference value as the feature dispersion.

[0091] For example, for the relevant indicator, age, age can correspond to multiple feature values. Any feature value can be used as a reference value. Each feature value has its own corresponding PICC thrombosis risk label. The number of types of PICC thrombosis risk labels corresponding to the reference value is counted.

[0092] Step S222: Use any PICC thrombosis risk label as a reference thrombosis risk label; calculate the total proportion of other PICC thrombosis risk labels besides the reference thrombosis risk label corresponding to the same reference value, as the feature similarity proportion.

[0093] As an embodiment of the present invention, the proportion of reference thrombosis risk labels corresponding to the same reference value can be calculated first and denoted as η. j , mix the preset complete value with η j The difference is taken as the feature similarity ratio, i.e., 1-η. j This feature similarity ratio is used to compensate for the sample imbalance problem that exists in the initial data collection. Since this is a data ratio calculation, the total number ratio of all PICC thrombosis risk labels is 1, and the corresponding preset complete value is 1. As another embodiment of the present invention, the total number ratio of other PICC thrombosis risk labels besides the reference thrombosis risk label corresponding to the same reference value can also be directly calculated as the feature similarity ratio.

[0094] Step S223: Based on the feature dispersion and the feature similarity ratio, determine the category feature of the reference value when corresponding to the reference thrombosis risk label.

[0095] As an embodiment of the present invention, the product of the feature dispersion and the feature similarity ratio can be used as the category feature of the reference value when corresponding to the reference thrombosis risk label. It can also be understood that the feature similarity ratio is used as the weight of the feature dispersion, and the feature dispersion is weighted to obtain the category feature of the reference value when corresponding to the reference thrombosis risk label. For example, when the k-th feature value of the i-th related indicator is used as the reference value, and the j-th PICC thrombosis risk label is used as the reference thrombosis risk label, the category feature can be expressed as: α(v ik ,L j )=N(v ik ,L j )×(1-η j ). Wherein, α(v ik ,L j ) represents the k-th eigenvalue of the i-th type. ik Below, corresponding to the j-th type of PICC thrombosis risk label L j Category features at time; η j v is the k-th feature value of the i-th correlation index. ik The percentage of the number of PICC thrombosis risk labels of type j.

[0096] Step S230: Construct a data feature vector for each feature value by corresponding to the category features under all types of PICC thrombosis risk labels for each feature value under the same type of related indicators.

[0097] This data feature vector consists of one feature vector for each feature value under the same category of related indicators. The number of elements in this feature vector is n, where n is the total number of feature values ​​under the same category of related indicators. It should be noted that identical feature values ​​under the same category of related indicators are grouped into the same category. The total number of feature values ​​here refers to the total number of feature values ​​with different values. For example, if all the feature values ​​under the same category of related indicators are 5, 2, 2, and 5, then the total number of feature values ​​under that category of related indicators is 2, and the corresponding feature values ​​are 2 and 5 respectively.

[0098] Step S240: Based on the data feature vector, determine the ability to distinguish thrombosis risk for different feature values ​​under relevant indicators.

[0099] While still using any feature value under any relevant indicator as a reference value, the thrombosis risk discrimination ability of the reference value is determined based on the data feature vector of the relevant indicator corresponding to the reference value and the number of reference values.

[0100] In some possible implementations, by analyzing the data feature vectors of relevant indicators corresponding to the reference values ​​and the number of reference values, the thrombosis risk discrimination ability of the reference values ​​can be quickly calculated. That is, the above step S240 can be achieved through the following steps S241, S242 and S243 (not shown in the figure):

[0101] Step S241: For the element values ​​in the data feature vector of the relevant indicators corresponding to the reference value, calculate the ratio of the number of element values ​​to the number of parameter values ​​to form an adjustment feature vector.

[0102] It can also be understood as calculating the ratio cn of the number of all element values ​​and parameter values ​​in the data feature vector of the relevant indicators corresponding to the reference value, and using the vector composed of all ratios cn as the adjustment feature vector.

[0103] Step S242: Based on the number of types of PICC thrombosis risk labels, the adjusted feature vector is modified to obtain the modified feature vector.

[0104] As an embodiment of the present invention, the reciprocal of the number of types of PICC thrombosis risk labels is taken as the difference between each element in the adjusted feature vector, and all the differences are used as a vector composed of elements, which is used as the corrected feature vector.

[0105] Step S243: Use the modulus of the modified feature vector as a reference value for the ability to distinguish thrombosis risk.

[0106] When the k-th feature value of the i-th relevant indicator is used as the reference value, the formula for calculating the thrombosis risk discrimination ability of the reference value is as follows: Where, ||γ(v ik )‖ is the modulus of the modified eigenvector, γ(v ik ) represents the modified feature vector; β(v) ik ) represents the data feature vector corresponding to the k-th feature value of the i-th related indicator; num(v ik ) represents the number of the k-th feature values ​​of the i-th correlation index; [num(v ik )] [n,1] Let be a quantity vector of n elements, consisting of the number of the k-th eigenvalues ​​of the i-th correlation index, where each element in the quantity vector is num(v). ik ); is the adjusted eigenvector of the kth eigenvalue of the i-th type of correlation index; n is the total number of eigenvalues ​​under the i-th type of correlation index. Let be an inverse vector of n elements, which is formed by the reciprocal of the total number of feature values ​​under the i-th correlation index, where the element value in the inverse vector is 1 / n.

[0107] In the formula for calculating the ability to distinguish thrombosis risk, the data feature vector β(v ik This reflects the data characteristics corresponding to the k-th feature value of the i-th related indicator. These data characteristics are determined by the features under each PICC thrombosis risk label under the k-th feature value of the i-th related indicator. It reflects the distribution characteristics of the data under the kth feature value of the i-th related indicator; This indicates that, under equilibrium conditions, the feature values ​​corresponding to each PICC thrombosis risk label exhibit a uniform distribution. Adjusting the element values ​​in the feature vector to be closer to... The higher the value, the weaker the ability of the relevant indicators corresponding to that feature value to distinguish PICC thrombosis risk, because it fails to clearly distinguish PICC thrombosis risk based on that feature; therefore, the smaller the difference between the data distribution feature of this relevant indicator and the uniform distribution, the more similar the data distribution at that node is to the uniform distribution, and the weaker the ability to distinguish PICC thrombosis risk, as it does not show a clear identification of a certain type of PICC thrombosis risk; ||γ(v ik The value indicates the ability of the characteristic value of the relevant indicator to distinguish thrombosis risk. It reflects the concentration of PICC thrombosis risk labels in the data under the characteristic value of the relevant indicator. The larger the value, the greater the difference between the distribution of PICC thrombosis risk labels and the uniform distribution. The data under the characteristic value of the relevant indicator has a significant ability to distinguish certain categories of thrombosis risk.

[0108] Step S300: Based on the differences in different feature values ​​under the same relevant indicator, the ability to distinguish thrombosis risk is corrected to obtain the classification ability of the relevant indicator.

[0109] Since different related indicators representing thrombosis characteristics have varying degrees of importance for thrombosis, relying solely on the distribution characteristics of a single characteristic value under the same related indicator to determine the risk of PICC thrombosis has low accuracy. This is because there is no comparison with data of other characteristic values ​​under the same related indicator. Characteristic values ​​under the same related indicator may show differences between their distribution and the average distribution to some extent. However, in reality, such concentrated distribution under a single characteristic value of the same related indicator is a normal phenomenon and is an attribute with low correlation to PICC thrombosis risk, and cannot be used as the main basis for judging thrombosis risk.

[0110] Each feature value of the same relevant indicator corresponds to a label distribution feature, which is also known as the modified feature vector γ(v). ikThe corrected feature vector is an n-dimensional vector, and the importance of relevant indicators is determined based on the similarity between vectors. If the label distributions of different feature values ​​under the same relevant indicator are similar, it is considered a normal distribution characteristic unrelated to PICC thrombosis risk, and therefore this relevant indicator cannot be used as the main basis for judging thrombosis risk. Therefore, the ability to distinguish thrombosis risk is corrected by the differences in data points under different feature values ​​of the same relevant indicator.

[0111] In some embodiments, the ability to distinguish thrombosis risk can be corrected based on the differences between different characteristic values ​​of relevant indicators to obtain the classification ability of relevant indicators. Here, different characteristic values ​​of relevant indicators can also reflect the similarity between different characteristic values; that is, step S300 above can be achieved through… Figure 3 The steps shown are to be implemented as follows:

[0112] Step S310: Using any relevant indicator as a reference relevant indicator, determine the importance of the classification ability based on the distribution of thrombosis risk differentiation ability corresponding to different feature values ​​under the same reference relevant indicator.

[0113] In mathematics, variance is typically used to reflect the dispersion or volatility of a set of data. A larger variance indicates greater volatility and instability in the data; conversely, a smaller variance indicates less volatility and stability. Specifically, variance helps understand how data points are distributed around their mean, i.e., the degree of centralization and dispersion, volatility and stability of the data. A smaller variance means the data points are more concentrated, with most data points close to the mean, while a larger variance means the data points are more dispersed, fluctuating significantly around the mean. Therefore, as an embodiment of this invention, the distribution of the ability to distinguish thrombosis risk corresponding to different feature values ​​under the same reference correlation index is reflected by variance; that is, the variance of the ability to distinguish thrombosis risk corresponding to different feature values ​​under the same reference correlation index is used as the degree of importance of the distinguishing ability.

[0114] Step S320: Select representative values ​​of thrombosis risk discrimination ability from the thrombosis risk discrimination ability corresponding to different characteristic values ​​of relevant indicators.

[0115] In one embodiment of the present invention, the maximum value among the thrombosis risk discrimination capabilities corresponding to different characteristic values ​​of the reference relevant indicator is selected as the representative value of the thrombosis risk discrimination capability of the reference relevant indicator. In another embodiment of the present invention, the mean value of the thrombosis risk discrimination capability corresponding to different characteristic values ​​of the reference relevant indicator can also be calculated as the representative value of the thrombosis risk discrimination capability of the reference relevant indicator. In yet another embodiment of the present invention, the median value of the thrombosis risk discrimination capability corresponding to different characteristic values ​​of the reference relevant indicator can also be used as the representative value of the thrombosis risk discrimination capability of the reference relevant indicator.

[0116] Step S330: Based on the importance of the relevant indicators, the representative value of the thrombosis risk differentiation ability corresponding to the relevant indicators is corrected to obtain the classification ability of the relevant indicators.

[0117] In this embodiment of the invention, the product of the importance of the classification ability of the reference relevant indicator and the representative value of the thrombosis risk discrimination ability is calculated as the classification ability of the reference relevant indicator. This is to achieve the purpose of correcting the representative value of the thrombosis risk discrimination ability corresponding to the reference relevant indicator by the importance of the reference relevant indicator, thereby obtaining the classification ability of the reference relevant indicator.

[0118] In steps S310 to S330 above, the ability to distinguish thrombosis risk is corrected based on the differences in different feature values ​​under the same relevant indicator, thus obtaining the classification ability of the relevant indicator. The classification ability of the relevant indicator can also be obtained through the calculation formula of the classification ability of the relevant indicator.

[0119] The magnitude of eigenvalues ​​reflects the similarity of feature distributions. The larger the variance of the thrombosis risk discrimination ability corresponding to these eigenvalues, the smaller the similarity of the feature distributions, and thus the greater the importance of the corresponding attribute. The greater the importance of the discrimination ability, the smaller the similarity, and the greater the importance of the relevant indicator. Taking the i-th relevant indicator as a reference, the formula for calculating the discrimination ability of the constructed relevant indicator is: W(V i )=p(V i )×mt k ‖γ(v ik )‖;wherein, W(V i ) represents the ability to classify the i-th relevant indicator; p(V) i ) represents the importance of the ability to classify the i-th relevant indicator; mt k ‖γ(v ik ‖ represents the representative value of the thrombosis risk discrimination ability of the i-th related indicator. The greater the significance of the discrimination ability of the related indicator, the greater its importance in distinguishing the thrombosis risk of PICC; mt k ‖γ(v ikThe value p(V) represents the ability to classify the relevant indicator. i W(V) serves as a correction factor for the dividing ability. i ) represents the i-th related indicator V i The corrected partitioning capability.

[0120] Thus, the ability to classify an attribute is obtained by using the difference in thrombosis risk under the same attribute representing thrombosis characteristics.

[0121] Step S400: Combining the thrombosis risk discrimination ability and the classification ability, the classification cost of the relevant indicators is determined by utilizing the similarity between the label distribution characteristics before and after the classification of the relevant indicators.

[0122] However, the following problem arises when constructing a decision tree for thrombosis risk: the strong segmentation ability of the obtained relevant indicator is due to the fact that thrombosis risk was well segmented before the relevant indicator was used, thus achieving good segmentation results in the subsequent segmentation. However, the actual correlation with the segmentation ability of the relevant indicator is weak; therefore, the obtained segmentation ability is inaccurate. It is necessary to utilize the similarity in data distribution before and after the segmentation to determine the segmentation cost of the relevant indicator.

[0123] If there is no significant improvement in the assessment of thrombosis risk before and after the partitioning, then the relevant indicator is considered to have no good effect on the assessment of thrombosis risk. The relevant indicator has no significant impact on the decision tree, requiring a deeper decision tree to differentiate PICC thrombosis risk. For the decision tree, the importance of the relevant indicator is relatively low at this position, and the partitioning cost of the relevant indicator is also lower. In this embodiment of the invention, the selected relevant indicators are made to significantly improve the assessment of thrombosis risk as much as possible.

[0124] In some embodiments, the similarity between the label distribution characteristics before and after the relevant index segmentation can be utilized to more accurately analyze the segmentation cost of the relevant index. That is, the above step S400 can be achieved through... Figure 4 The steps shown are to be implemented as follows:

[0125] Step S410: Compare the thrombosis risk discrimination ability of relevant indicators before classification with the thrombosis risk discrimination ability of different feature values ​​of relevant indicators after classification, and determine the classification improvement value.

[0126] First, the thrombosis risk discrimination ability of the relevant indicators is calculated, which is used as the discrimination ability to be classified. For each feature value, the difference between the thrombosis risk discrimination ability of each feature value under the relevant indicator and the discrimination ability to be classified is calculated, which is used as the single classification improvement value for each feature value under the relevant indicator. A representative value is selected from the single classification improvement values ​​for each feature value under the relevant indicator as the classification improvement value. The changes in discrimination ability before and after the classification of the relevant indicator are compared, that is, the changes in the discrimination ability to be classified and the discrimination ability are compared. The difference between the two reflects the magnitude of the improvement in classification ability. The greater the improvement, the greater the classification cost of the relevant indicator. The difference with the largest difference between the thrombosis risk discrimination ability of the relevant indicator and the thrombosis risk discrimination ability of the feature value is selected as the final improvement in the classification ability of the relevant indicator, which is recorded as the classification improvement value.

[0127] Therefore, in one embodiment of the present invention, the maximum value among the single division improvement values ​​is selected as the division improvement value corresponding to the relevant index.

[0128] It should be noted that the method for obtaining the thrombosis risk discrimination ability of relevant indicators is the same as the method for obtaining the thrombosis risk discrimination ability of each feature value under the relevant indicators. More specifically: The relevant indicators are labeled with corresponding PICC thrombosis risk tags; based on the types of PICC thrombosis risk tags corresponding to the same type of relevant indicators and the proportion of different types of PICC thrombosis risk tags, the unclassified category features of the same type of relevant indicators under different types of PICC thrombosis risk tags are determined; based on the unclassified category features of the same type of relevant indicators under different types of PICC thrombosis risk tags, a data feature vector for each feature value is constructed; based on the unclassified data feature vectors, the thrombosis risk discrimination ability of the relevant indicators is determined.

[0129] Specifically, based on the types of PICC thrombosis risk labels corresponding to the same type of related indicators and the proportion of different types of PICC thrombosis risk labels, the unclassified category features of the same type of related indicators corresponding to different types of PICC thrombosis risk labels are determined. Specifically: the number of types of PICC thrombosis risk labels corresponding to the same type of related indicators is obtained as the dispersion of the unclassified features; any PICC thrombosis risk label is used as a reference thrombosis risk label; the total proportion of other PICC thrombosis risk labels besides the reference thrombosis risk label corresponding to the same type of related indicators is calculated as the similarity proportion of the unclassified features; based on the dispersion of the unclassified features and the similarity proportion of the unclassified features, the unclassified category features of the same type of related indicators corresponding to the reference thrombosis risk label are determined.

[0130] Specifically, based on the feature vector of the data to be segmented, the thrombosis risk discrimination ability of relevant indicators is determined. Specifically: for each element value within the feature vector of the data to be segmented corresponding to the same type of relevant indicator, the ratio of the element value to the number of the same type of relevant indicator is calculated to form an adjusted feature vector to be segmented; based on the number of types of PICC thrombosis risk labels, the adjusted feature vector to be segmented is corrected to obtain a corrected feature vector to be segmented; the modulus of the corrected feature vector to be segmented is used as the reference value for thrombosis risk discrimination ability. It should be noted that the method for obtaining the thrombosis risk discrimination ability of relevant indicators is the same as that for the thrombosis risk discrimination ability of the feature values ​​of relevant indicators. The only difference is that when calculating the thrombosis risk discrimination ability of the feature values ​​of relevant indicators, the PICC thrombosis risk labels corresponding to the feature values ​​are analyzed separately, while when calculating the thrombosis risk discrimination ability of relevant indicators, the PICC thrombosis risk labels corresponding to the relevant indicators are analyzed and calculated. Therefore, this will not be elaborated further here.

[0131] Step S420: Combine the partitioning improvement value and the partitioning capability to determine the partitioning cost of relevant indicators.

[0132] Both the partitioning improvement value and the partitioning capability are positively correlated with the partitioning cost of the relevant indicator. In one embodiment of the present invention, the square root of the sum of the squares of the partitioning improvement value and the partitioning capability is used as the partitioning cost of the relevant indicator.

[0133] In steps S410 to S420 above, the similarity between PICC thrombosis risk labels before and after the relevant indicator classification is used to determine whether the classification of the relevant indicator has a significant effect on the assessment of thrombosis risk, thereby determining the classification cost t(V) of the relevant indicator. i Using the i-th relevant indicator as the reference relevant indicator, construct the partitioning cost (V) of the reference relevant indicator. i The calculation formula for ) is as follows: Wherein, t(V) i W(V) represents the enhancement value for the i-th related indicator; i ) represents the ability to classify the i-th related indicator.

[0134] The partitioning cost was obtained by analyzing the changes in relevant indicators before and after the partitioning. Thus, the partitioning cost was determined through the analysis of relevant indicators, and the selection of decision tree attributes was further implemented based on the partitioning cost.

[0135] Step S500: Based on the classification cost of different related indicators, construct the corresponding PICC-related thrombosis prediction model.

[0136] After obtaining the segmentation cost based on the changes in body shape before and after segmentation in steps S100-S400, multiple decision trees are constructed based on the segmentation costs of different types of relevant indicators for PICC catheter patients. More specifically, the relevant indicator with the highest segmentation cost is selected for constructing the decision tree; this can also be understood as using the relevant indicator with the highest segmentation cost as the segmentation criterion for the decision tree.

[0137] By using different random sampling results, N selected samples are used as the root node samples of a decision tree to train a decision tree. Multiple decision trees can be constructed using different random sampling training sets, resulting in a random forest, which serves as the corresponding PICC-related thrombosis prediction model, thus completing the construction of the PICC catheter-related thrombosis risk prediction model. In this embodiment of the invention, each relevant index is equivalent to a parent node of the decision tree, and the feature values ​​under the relevant index are equivalent to child nodes further subdivided from the parent node of the decision tree.

[0138] By inputting relevant indicators of thrombosis risk, the PICC thrombosis risk level can be obtained. Healthcare professionals can then use this risk level to provide different interventions to individuals undergoing PICC thrombosis examinations. For example, preventive advice can be given to healthy individuals; low-risk patients may only require routine care and monitoring; medium-risk patients may require more frequent monitoring and preventive measures; and high-risk patients may require more aggressive prevention and treatment measures.

[0139] As an embodiment of the present invention, after obtaining the PICC-related thrombosis prediction model, the C-index and Brier score can be used to evaluate the models constructed by different algorithms to obtain the best prediction model.

[0140] This application provides a system for constructing a PICC catheter-related thrombosis risk prediction model, such as... Figure 5 As shown, system 600 includes:

[0141] The data acquisition module 610 acquires relevant indicators of patients with PICC catheters.

[0142] The capability determination module 620 determines the ability to distinguish thrombosis risk under different characteristic values ​​of the relevant indicators based on the ability of the relevant indicators to distinguish thrombosis risk.

[0143] The correction module 630 corrects the ability to distinguish thrombosis risk based on the differences in different feature values ​​under the same relevant indicator, thereby obtaining the classification ability of the relevant indicator;

[0144] The cost determination module 640 combines the thrombosis risk discrimination ability and the classification ability, and uses the similarity between the label distribution characteristics before and after the classification of relevant indicators to determine the classification cost of relevant indicators.

[0145] The model building module 650 constructs corresponding PICC-related thrombosis prediction models based on the classification costs of different types of relevant indicators.

[0146] As an embodiment of the present invention, the probability of thrombosis in each PICC catheter patient can be calculated based on the weight values ​​of each factor using the implanted PICC catheter-related thrombosis prediction model.

[0147] As an embodiment of the present invention, a result output module 660 may be added to determine the risk level based on the constructed PICC catheter-related thrombosis prediction model for each PICC catheter patient's case and output the result; and provide key reminders for the high-incidence period of PICC catheter-related thrombosis.

[0148] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the system for constructing a PICC catheter-related thrombosis risk prediction model and the method for constructing a PICC catheter-related thrombosis risk prediction model provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0149] Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. For example, as shown... Figure 6 As shown, the computer device 700 includes: a memory 701, a processor 702, and a computer program 703 stored in the memory 701 and running on the processor 702, wherein when the processor 702 executes the computer program 703, the computer device can execute any of the aforementioned methods for constructing a PICC catheter-related thrombosis risk prediction model.

[0150] Furthermore, this application also protects an apparatus that may include a memory and a processor, wherein the memory stores executable program code, and the processor is used to call and execute the executable program code to execute a method for constructing a PICC catheter-related thrombosis risk prediction model provided in this application.

[0151] This embodiment can divide the device into functional modules based on the above method example. For example, each module can correspond to a separate function, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0152] When each module is divided according to its function, the device may also include a signal uploading module, a determination module, and an adjustment module. It should be noted that all relevant content of each step involved in the above method embodiments can be referenced from the functional descriptions of the corresponding functional modules, and will not be repeated here.

[0153] It should be understood that the device provided in this embodiment is used to execute the above-described method for constructing a PICC catheter-related thrombosis risk prediction model, and therefore can achieve the same effect as the above-described implementation method.

[0154] When using integrated units, the device may include a processing module and a storage module. When applied to a workpiece, the processing module can be used to control and manage the workpiece's operations. The storage module can be used to support the execution of program code by the workpiece.

[0155] The processing module may be a processor or a controller, which can implement or execute various exemplary logic blocks, modules, and circuits as disclosed in this application. The processor may also be a combination of functions that implement computing capabilities, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc., and the storage module may be a memory.

[0156] In addition, the device provided in the embodiments of this application may specifically be a chip, component or module. The chip may include a connected processor and a memory. The memory is used to store instructions. When the processor calls and executes the instructions, the chip can execute the method for constructing a PICC catheter-related thrombosis risk prediction model provided in the above embodiments.

[0157] This embodiment also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the above-described related method steps to implement the method for constructing a PICC catheter-related thrombosis risk prediction model provided in the above embodiment.

[0158] This embodiment also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned steps to realize the method for constructing a PICC catheter-related thrombosis risk prediction model provided in the above embodiment.

Claims

1. A method for constructing a PICC catheter-related thrombosis risk prediction model, characterized in that, The method includes the following steps: Obtain relevant indicators for patients with PICC catheters; Based on the ability of the relevant indicators to distinguish thrombosis risk, the ability of different characteristic values ​​under the relevant indicators to distinguish thrombosis risk is determined. Based on the distribution of thrombosis risk discrimination ability corresponding to different feature values ​​under the same relevant indicator, the thrombosis risk discrimination ability is corrected to obtain the classification ability of the relevant indicator. The classification enhancement value is determined based on the difference between the thrombosis risk discrimination ability of relevant indicators and the thrombosis risk discrimination ability of different feature values ​​under the classified relevant indicators; the classification cost of relevant indicators is determined based on the classification enhancement value and classification ability. Based on the classification costs of different related indicators, a corresponding PICC-related thrombosis prediction model is constructed. Specifically, based on the ability of the relevant indicators to distinguish thrombosis risk, the ability to distinguish thrombosis risk for different feature values ​​under the relevant indicators is determined, including: assigning corresponding PICC thrombosis risk labels to the feature values ​​of the relevant indicators; determining the category features of different types of PICC thrombosis risk labels corresponding to each feature value under the same type of relevant indicators, based on the types of PICC thrombosis risk labels corresponding to each feature value under the same type of relevant indicators and the proportion of different types of PICC thrombosis risk labels; constructing a data feature vector for each feature value based on the category features of all types of PICC thrombosis risk labels corresponding to each feature value under the same type of relevant indicators; and determining the ability to distinguish thrombosis risk for different feature values ​​under the relevant indicators based on the data feature vectors.

2. The method for constructing a PICC catheter-related thrombosis risk prediction model according to claim 1, characterized in that, The method of determining the category characteristics of different types of PICC thrombosis risk labels corresponding to each feature value under the same type of related indicator, and the proportion of different types of PICC thrombosis risk labels, includes: Use any feature value under any relevant indicator as a reference value; obtain the number of types of PICC thrombosis risk labels corresponding to the same reference value as the feature dispersion. Use any PICC thrombosis risk label as a reference thrombosis risk label; calculate the total proportion of other PICC thrombosis risk labels besides the reference thrombosis risk label corresponding to the same reference value, and use it as the feature similarity proportion. Based on the feature dispersion and the feature similarity ratio, the category feature of the reference value in the corresponding reference thrombosis risk label is determined.

3. The method for constructing a PICC catheter-related thrombosis risk prediction model according to claim 1, characterized in that, The determination of the ability to distinguish thrombosis risk based on the data feature vector under different feature values ​​of relevant indicators includes: Use any characteristic value under any relevant indicator as a reference value; Based on the data feature vectors of relevant indicators corresponding to the reference values ​​and the number of reference values, the thrombosis risk discrimination ability of the reference values ​​is determined.

4. The method for constructing a PICC catheter-related thrombosis risk prediction model according to claim 3, characterized in that, The determination of the thrombosis risk discrimination ability of the reference value based on the data feature vector of the relevant indicators corresponding to the reference value and the number of reference values ​​includes: For each element value in the data feature vector of the relevant indicator corresponding to the reference value, the ratio of the number of element values ​​to the number of reference values ​​is calculated to form an adjustment feature vector. Based on the number of types of PICC thrombosis risk labels, the adjusted feature vector is modified to obtain the modified feature vector; The modulus of the modified feature vector is used as a reference value to distinguish the risk of thrombosis.

5. The method for constructing a PICC catheter-related thrombosis risk prediction model according to claim 1, characterized in that, The distribution of thrombosis risk discrimination ability based on the different feature values ​​corresponding to the same relevant indicator is used to correct the thrombosis risk discrimination ability, thereby obtaining the classification ability of the relevant indicator, including: Using any relevant indicator as a reference relevant indicator, the importance of the classification ability of the reference relevant indicator is determined based on the distribution of the ability to distinguish thrombosis risk corresponding to different feature values ​​under the same reference relevant indicator. Representative values ​​for distinguishing thrombosis risk were selected from the thrombosis risk discrimination capabilities corresponding to different characteristic values ​​of relevant indicators. Based on the importance of the classification ability of the reference relevant indicators, the representative value of the thrombosis risk differentiation ability corresponding to the reference relevant indicators is modified to obtain the classification ability of the reference relevant indicators.

6. The method for constructing a PICC catheter-related thrombosis risk prediction model according to claim 5, characterized in that, The process of selecting representative values ​​for distinguishing thrombosis risk from the different characteristic values ​​of relevant reference indicators includes: The maximum value among the different characteristic values ​​of the reference relevant indicator corresponding to the thrombosis risk discrimination ability is selected as the representative value of the thrombosis risk discrimination ability of the reference relevant indicator.

7. The method for constructing a PICC catheter-related thrombosis risk prediction model according to claim 1, characterized in that, The difference between the thrombosis risk discrimination ability based on relevant indicators and the thrombosis risk discrimination ability of different feature values ​​under the relevant indicators after classification is used to determine the classification improvement value, including: By comparing the difference between the thrombosis risk discrimination ability of relevant indicators before classification and the thrombosis risk discrimination ability of different characteristic values ​​of relevant indicators after classification, the classification improvement value is determined.

8. The method for constructing a PICC catheter-related thrombosis risk prediction model according to claim 1, characterized in that, The step of constructing a corresponding PICC-related thrombosis prediction model based on the classification costs of different related indicators includes: Based on the classification costs of different types of relevant indicators in PICC catheter patients, multiple decision trees are constructed to obtain a random forest model, which serves as the corresponding PICC-related thrombosis prediction model.

9. A system for constructing a PICC catheter-related thrombosis risk prediction model, characterized in that, The system includes: The data acquisition module obtains relevant indicators for patients with PICC catheters. The capability determination module, based on the ability to distinguish thrombosis risk according to the relevant indicators, determines the thrombosis risk distinguishing ability of different feature values ​​under the relevant indicators, including: assigning corresponding PICC thrombosis risk labels to the feature values ​​of the relevant indicators; determining the category features of different types of PICC thrombosis risk labels corresponding to each feature value under the same type of relevant indicators, based on the types of PICC thrombosis risk labels corresponding to each feature value under the same type of relevant indicators and the proportion of different types of PICC thrombosis risk labels; constructing a data feature vector for each feature value from the category features of all types of PICC thrombosis risk labels corresponding to each feature value under the same type of relevant indicators; and determining the thrombosis risk distinguishing ability of different feature values ​​under the relevant indicators based on the data feature vectors. The correction module corrects the thrombosis risk discrimination ability based on the distribution of thrombosis risk discrimination ability corresponding to different feature values ​​under the same relevant indicator, thereby obtaining the classification ability of the relevant indicator. The cost determination module determines the segmentation improvement value based on the difference between the thrombosis risk discrimination ability of relevant indicators and the thrombosis risk discrimination ability of different feature values ​​under the segmented relevant indicators; and determines the segmentation cost of relevant indicators based on the segmentation improvement value and segmentation ability. The model building module constructs corresponding PICC-related thrombosis prediction models based on the classification costs of different types of relevant indicators.

Images