Construction method of central venous catheter intelligent decision model based on multi-source data

Abstract

The invention relates to the technical field of central venous catheters, in particular to a construction method of a central venous catheter intelligent decision model based on multi-source data. The method comprises the following steps: acquiring case data of the central venous catheter, processing the data, and obtaining sign data of a patient and structural feature data of the catheter by extracting sign data of a case patient and analyzing structural features of the central venous catheter; acquiring a vein ultrasonic image according to the sign data of the patient, estimating the specification of the central venous catheter through ultrasonic image analysis, and calculating the displacement probability of the catheter; according to the displacement probability data, mistaken touch area collection is carried out, and the abnormal state of the catheter is analyzed in combination with the vein ultrasound image data; performing catheter risk assessment according to the abnormal state data, and constructing a central venous catheter intelligent decision model based on catheter risk data; according to the method, the central venous catheter decision is optimized, so that the central venous catheter decision is more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of central venous catheters, and in particular to a method for constructing a central venous catheter intelligent decision-making model based on multi-source data. Background Art

[0002] Central venous access devices are catheters with the tip of the catheter positioned in the central vein. Currently, there are three types of central venous access devices commonly used in clinical practice: 1. Central venous catheters (CVC) inserted through the internal jugular vein, subclavian vein, and femoral vein. 2. Peripherally central catheters (PlCC) inserted through peripheral veins. 3. Completely implantable venous access ports (IVAPs) inserted through the internal jugular vein or subclavian vein. With the development of medicine, pharmacy, materials science and other disciplines, various new intravascular catheters and auxiliary medical consumables have been gradually applied in clinical practice. Critically ill patients often need to have a central venous catheter (excluding catheters used only for venous access or blood purification) to monitor hemodynamics and serve as an infusion route. At the same time, it is also necessary to take into account the mechanical complications of catheter indwelling and the prevention and treatment of central line-associated blood-stream infection (CLABSI) and thrombosis. Clinicians need to select appropriate catheters according to the patient's condition and adjust the type and location of the catheter at different stages of the disease. In order to standardize the central venous catheter management strategy for critically ill patients, the traditional method of constructing a central venous catheter intelligent decision-making model based on multi-source data has the problem of inaccurate estimation of central venous catheter specifications and inaccurate central venous catheter risk analysis. Summary of the invention

[0003] Based on this, it is necessary to provide a method for constructing a central venous catheter intelligent decision-making model based on multi-source data to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a method for constructing a central venous catheter intelligent decision-making model based on multi-source data includes the following steps:

[0005] Step S1: Acquire central venous catheter case data; extract case patient vital signs based on the central venous catheter case data to obtain case patient vital signs data; analyze central venous catheter structure characteristics based on the central venous catheter case data to obtain central venous catheter structure characteristic data;

[0006] Step S2: collecting venous ultrasound images of case patients according to the case patient's physical sign data to obtain venous ultrasound image data of case patients; estimating the specifications of central venous catheters based on the structural characteristic data of central venous catheters according to the case patient's venous ultrasound image data to obtain central venous catheter specification data; calculating the probability of catheter position displacement according to the central venous catheter specification data and the case patient's venous ultrasound image data to obtain central venous catheter displacement probability data;

[0007] Step S3: collecting the central venous catheter mis-touch area according to the central venous catheter displacement probability data to obtain the central venous catheter mis-touch area data; analyzing the central venous catheter abnormal state according to the central venous catheter mis-touch area data and the venous ultrasound image data of the case patient to obtain the central venous catheter abnormal state data;

[0008] Step S4: Perform a central venous catheter risk assessment based on the central venous catheter abnormal status data to obtain central venous catheter risk data; construct a central venous catheter intelligent decision-making model based on the central venous catheter risk data to obtain a central venous catheter intelligent decision-making model.

[0009] The present invention can comprehensively monitor and evaluate the use status of central venous catheters and their related risks through multi-source data fusion and intelligent analysis, thereby improving the safety and accuracy of clinical treatment. By acquiring central venous catheter case data and performing physical sign extraction and catheter structural feature analysis, it not only provides basic data for subsequent risk assessment, but also comprehensively grasps the patient's physiological characteristics and the structural characteristics of the catheter, laying a good foundation for accurate analysis. Through the collection and analysis of venous ultrasound image data, the catheter specifications can be estimated in combination with the patient's physical sign data, and the specifications and position changes of the central venous catheter can be accurately grasped. This process helps to deeply understand the position and stability of the catheter in the blood vessel, and provides a reliable data basis for the subsequent calculation of the probability of catheter displacement. Based on the catheter displacement probability data, the mis-touch area data is further collected, the area contacted by the catheter is analyzed, and the abnormal state is identified. This analysis can help identify potential risks caused by abnormal catheter position, such as vascular damage, blood flow obstruction and other problems. The risk assessment combined with abnormal state data helps to comprehensively evaluate various problems encountered by the catheter during use, such as thrombosis, infection, vascular damage and other potential complications, and enhance the safety of the treatment process. Therefore, the present invention optimizes a traditional method for constructing a central venous catheter intelligent decision-making model based on multi-source data, solves the problem of inaccurate estimation of central venous catheter specifications and inaccurate central venous catheter risk analysis in the traditional method for constructing a central venous catheter intelligent decision-making model based on multi-source data, and improves the accuracy of central venous catheter specification estimation and central venous catheter risk analysis.

[0010] Preferably, step S1 comprises the following steps:

[0011] Step S11: Acquire central venous catheter case data;

[0012] Step S12: performing case desensitization processing according to the central venous catheter case data to obtain central venous catheter case desensitization data;

[0013] Step S13: extracting the case patient's vital signs based on the central venous catheter case desensitization data to obtain the case patient's vital signs data;

[0014] Step S14: performing a central venous catheter structural feature analysis based on the central venous catheter case desensitized data to obtain central venous catheter structural feature data.

[0015] The present invention can ensure the privacy protection of patients while providing a high-quality data basis for subsequent analysis by obtaining central venous catheter case data and performing desensitization processing. The desensitized case data protects the patient's personal information and ensures the security of the data. Based on these desensitized data, the patient's vital signs information is further extracted to fully understand the patient's health status and provide reliable physiological data support for subsequent analysis. Combining the desensitized data of the central venous catheter case to analyze the catheter structure characteristics helps to accurately grasp the relevant characteristics of the catheter, including the shape, specifications and position characteristics of the catheter, which is of great significance for the use and maintenance of the catheter. Through a detailed analysis of the catheter structure characteristics, the relationship between the catheter and the patient's blood vessels is better understood, and the risk factors or potential problems that exist are found, providing a scientific basis for subsequent treatment and operation. On the whole, this series of steps can provide more accurate basic data for the safe use and management of central venous catheters by extracting valuable information from the case data, ensuring that potential problems during the use of the catheter are identified and avoided in advance.

[0016] Preferably, step S2 comprises the following steps:

[0017] Step S21: collecting venous ultrasound images of the case patient according to the case patient's physical sign data to obtain venous ultrasound image data of the case patient;

[0018] Step S22: performing vein feature analysis based on the venous ultrasound image data of the case patient to obtain ultrasound image vein feature data;

[0019] Step S23: estimating the specifications of the central venous catheter based on the ultrasound image vein feature data and the central venous catheter structure feature data to obtain the central venous catheter specification data;

[0020] Step S24: Calculate the probability of catheter position displacement according to the central venous catheter specification data and the ultrasound image vein feature data to obtain central venous catheter displacement probability data.

[0021] The present invention obtains detailed venous imaging data by collecting venous ultrasound image data of case patients, providing direct image data support for subsequent venous feature analysis. Venous feature analysis helps to deeply understand the morphology, structure and dynamic changes of veins by extracting key features in the image, thereby providing a reliable basis for accurately evaluating the venous status. These venous feature data are used to estimate the specifications of the central venous catheter structural characteristics, accurately infer the adaptability and specification requirements of the catheter, and provide a reference basis for selecting a suitable catheter model. Furthermore, the probability of catheter position displacement is calculated in combination with the catheter specification data and the venous feature data, which helps to predict and evaluate the risk of catheter displacement. Through this series of steps, the adaptability and safety of the catheter are comprehensively evaluated, providing a scientific basis for risk management in the subsequent treatment process. Overall, these steps can improve the accuracy of the operation during the use of the catheter and reduce the potential risk of complications through in-depth analysis of venous ultrasound image data.

[0022] Preferably, step S22 includes the following steps:

[0023] Step S221: collecting image texture features according to the venous ultrasound image data of the case patient to obtain ultrasound image texture feature data;

[0024] Step S222: performing venous blood vessel edge feature analysis based on the ultrasound image texture feature data and the venous ultrasound image data of the case patient to obtain venous blood vessel edge feature data;

[0025] Step S223: performing contour feature analysis according to the venous blood vessel edge feature data and the ultrasonic image texture feature data to obtain venous blood vessel contour data;

[0026] Step S224: Performing vein feature analysis based on the vein contour data and the vein edge feature data to obtain ultrasound image vein feature data.

[0027] The present invention collects texture features from the venous ultrasound images of case patients, extracts subtle structural changes in the images, and provides important data support for the morphological analysis of veins. Texture feature data helps to reveal the tissue characteristics of venous blood vessels, such as wall smoothness, uniformity of vascular walls, etc., and thus reflects the health status of veins. Using these texture feature data and venous image information for edge feature analysis helps to accurately identify the boundaries of blood vessels, thereby improving the positioning accuracy of venous blood vessels and reducing errors. Next, contour feature analysis is performed in combination with edge feature data and texture feature data to further clarify the morphological contour of the blood vessels, provide support for a comprehensive assessment of the venous morphology, and perform comprehensive venous feature analysis based on venous vascular contour feature data and edge feature data, which can extract the key characteristics of veins from all directions and angles, thereby providing more comprehensive and accurate image data for clinical applications. The implementation of these steps improves the accuracy and depth of venous image analysis and helps to effectively identify potential risks.

[0028] Preferably, step S222 includes the following steps:

[0029] Calculate the grayscale value of the ultrasound image according to the venous ultrasound image data of the case patient, so as to obtain the grayscale value data of the venous ultrasound image;

[0030] Calculate the global grayscale value standard deviation of the ultrasound image according to the grayscale value data of the venous ultrasound image to obtain the global grayscale value standard deviation data of the ultrasound image;

[0031] Performing ultrasound image grayscale contrast analysis based on ultrasound image global grayscale value standard deviation data to obtain ultrasound image grayscale contrast data;

[0032] Calculate the gradient change of texture feature direction according to the texture feature data of the ultrasonic image to obtain the gradient change data of the texture feature direction of the image;

[0033] The venous blood vessel edge feature data is obtained by performing venous blood vessel edge feature analysis based on the image texture feature directional gradient change data and the ultrasound image gray value contrast data.

[0034] The present invention provides detailed quantitative data for the intensity distribution of the image by calculating the grayscale value of the venous ultrasound image, which helps to analyze the subtle differences in the venous wall and the dynamic changes in blood flow. Further, by calculating the global grayscale value standard deviation of the image, the discrete degree of grayscale in the image is revealed, thereby evaluating the clarity and contrast of the image. This process helps to remove noise and improve image quality. Next, the grayscale value contrast analysis can quantify the local differences in the image, help accurately identify the boundary between the venous structure and the surrounding tissue, and improve the accuracy of blood vessel identification. The calculation of the directional gradient change of the texture feature reveals the changing rules of the texture in different directions in the image, further extracts detailed information, and reflects the density changes of the tissue around the blood vessels. Combining these data to perform venous vascular edge feature analysis helps to accurately outline the edge contour of the vein, improve the resolution of the image, and provide more accurate vein position and morphology data.

[0035] Preferably, step S23 includes the following steps:

[0036] Step S231: performing image vein fine-grained analysis according to the ultrasound image vein feature data to obtain ultrasound image vein fine-grained data;

[0037] Step S232: performing image vein wall thickness detection based on the ultrasound image vein fine-grained data to obtain ultrasound image vein wall thickness data;

[0038] Step S233: performing image vein position measurement on the ultrasound image vein feature data, thereby obtaining ultrasound image vein position data;

[0039] Step S234: performing image vein depth detection according to the ultrasound image vein position data to obtain ultrasound image vein depth data;

[0040] Step S235: estimating the specifications of the central venous catheter based on the central venous catheter structural feature data based on the ultrasound image vein depth data to obtain the central venous catheter specification data.

[0041] The present invention deeply explores the subtle differences in venous images through fine-grained analysis of the veins in ultrasound images, improves the clarity and detail of the image, and thus provides more accurate basic data for subsequent analysis. Venous wall thickness detection helps determine the health status of blood vessels and promptly identifies whether the vascular wall is thickened, which is very important for evaluating the elasticity and potential lesions of blood vessels. Venous position measurement can accurately determine the spatial position of the vein in the image, provide key data for positioning, and further improve the visibility and positioning accuracy of the vascular structure in the image. Depth detection ensures the position of the vein in three-dimensional space and its relative relationship with the surrounding tissue, which helps to fully understand the spatial distribution of blood vessels and their state changes. By estimating the specifications of the central venous catheter in combination with the venous depth data, not only can more accurate catheter specification information be obtained, but also the adaptability and safety of the catheter can be further optimized.

[0042] Preferably, step S3 comprises the following steps:

[0043] Step S31: collecting the central venous catheter mis-touch area according to the central venous catheter displacement probability data to obtain the central venous catheter mis-touch area data;

[0044] Step S32: estimating the probability of local thrombosis according to the central venous catheter mis-touch area data and the venous ultrasound image data of the case patient, and obtaining ultrasound image local thrombosis probability data;

[0045] Step S33: performing central venous catheter abnormal state analysis based on the local thrombosis probability data of the ultrasound image and the central venous catheter mis-touch area data to obtain central venous catheter abnormal state data.

[0046] The present invention accurately identifies potential changes in the position of the catheter by analyzing the probability data of central venous catheter displacement, thereby helping to locate the area of ​​mis-touch and avoid damage to surrounding tissues. By collecting the mis-touch area, the dangerous area where the blood vessel and the catheter are in contact can be identified in time, providing an effective basis for subsequent preventive work. Combining ultrasound image data with contact area information, the probability of local thrombosis is evaluated, so that potential thrombosis risks can be discovered in advance, which is of great significance for preventing thrombotic complications. Further analysis of the local thrombosis probability data helps to evaluate whether the central venous catheter is in an abnormal state and reveal problems of blood flow obstruction or catheter blockage. The abnormal state analysis of the central venous catheter can provide a scientific basis for identifying abnormal catheter function.

[0047] Preferably, step S32 includes the following steps:

[0048] Collecting vein morphology changes to obtain ultrasound image vein morphology change data;

[0049] Step S322: performing venous puncture detection according to the ultrasound image venous morphology change data to obtain ultrasound image venous puncture data;

[0050] Step S323: performing venous vessel injury detection according to the ultrasonic image venous vessel puncture data to obtain ultrasonic image venous vessel injury data;

[0051] Step S324: performing vasospasm analysis based on the ultrasound image venous vascular injury data to obtain ultrasound image vasospasm probability data;

[0052] Step S325: estimating venous blood flow obstruction according to the central venous catheter mis-touch area data and the ultrasound image venous vascular injury data to obtain venous blood flow obstruction data;

[0053] Step S326: estimating the probability of local thrombosis based on the ultrasound image vascular spasm probability data and the venous blood flow obstruction data, and obtaining the ultrasound image local thrombosis probability data.

[0054] The present invention can accurately capture the morphological characteristics of veins in different states by collecting changes in vein morphology, which provides basic data for further analysis of the health status of veins. By performing vascular puncture detection on the data of changes in vein morphology in ultrasound images, it is effectively determined whether the catheter touches the blood vessel, thereby identifying the risk of puncture and reducing unnecessary damage. Based on the vascular puncture data, vascular injury detection can identify minor damage or rupture of the blood vessel at an early stage, preventing further damage to the blood vessel or causing more serious complications. Further analysis of the vascular injury situation, probability analysis of vascular spasm, and then predicted blood flow problems provide guidance for prevention and intervention. At the same time, by combining the central venous catheter mis-touch area data with vascular injury data, the risk of blood flow obstruction can be predicted in advance, thereby reducing complications caused by blood flow obstruction. Finally, based on the combination of vascular spasm probability and blood flow obstruction data, the probability of local thrombosis can be accurately estimated, providing strong data support for subsequent treatment and intervention measures.

[0055] Preferably, step S33 includes the following steps:

[0056] Step S331: performing venous thrombosis enlargement analysis based on the local thrombosis probability data of the ultrasound image to obtain venous thrombosis enlargement data;

[0057] Step S332: performing venous vessel blockage detection based on the venous vessel thrombus enlargement data to obtain venous vessel blockage data;

[0058] Step S333: performing central venous catheter function loss detection according to the venous vessel blockage data and the venous vessel thrombosis enlargement data to obtain central venous catheter function loss data;

[0059] Step S334: estimating the probability of venous infection according to the local thrombosis probability data of the ultrasound image and the central venous catheter mis-touch area data to obtain venous infection probability data;

[0060] Step S335: performing central venous catheter abnormal state analysis based on the venous infection probability data and the central venous catheter function loss data to obtain central venous catheter abnormal state data.

[0061] The present invention monitors the development and changes of thrombi in veins through the analysis of venous thrombus enlargement, thereby warning the growth trend of thrombi and providing early indications for timely treatment. By combining the data of thrombus enlargement for venous vessel blockage detection, it is possible to accurately identify whether thrombi cause blood flow obstruction, thereby further determining whether intervention measures are needed to avoid aggravation of blockage. By combining the data of thrombus enlargement and vascular blockage, central venous catheter function loss detection can be performed to timely detect the decline in catheter function caused by thrombus or blockage, and avoid the occurrence of serious complications. In addition, by combining the local thrombosis probability data with the central venous catheter mistouch area data, the risk of venous infection is assessed, signs of infection are detected early, and the infection is avoided from spreading and endangering the health of patients. Finally, based on the venous infection probability data and the catheter function loss data, the abnormal state analysis of the central venous catheter is performed, which can comprehensively evaluate the catheter state and provide a reliable basis for subsequent treatment strategies.

[0062] Preferably, step S4 comprises the following steps:

[0063] Step S41: Perform venous function impairment detection based on the abnormal state data of the central venous catheter to obtain venous function impairment data;

[0064] Step S42: Estimating venous catheter complications according to the venous function impairment data and the central venous catheter abnormal state data to obtain central venous catheter complication data;

[0065] Step S43: performing a central venous catheter risk assessment based on the venous catheter complication data and the venous function impairment data to obtain central venous catheter risk data;

[0066] Step S44: construct a central venous catheter intelligent decision-making model based on the central venous catheter risk data to obtain a central venous catheter intelligent decision-making model.

[0067] The present invention can detect venous dysfunction caused by abnormal central venous catheter status at an early stage, thereby allowing timely intervention and reducing long-term impacts on the patient's health. This step helps to confirm whether the catheter has caused hemodynamic changes, thereby determining whether the catheter position needs to be adjusted. Complications are estimated by combining venous dysfunction data and catheter abnormal status data to identify complication types and their development trends, predict the occurrence of venous-related problems in advance, improve clinical early warning capabilities, and avoid potential serious complications. Risk assessment is further performed based on complication data to provide personalized risk assessments for different patients.

[0068] The present invention is that, through multi-source data fusion and intelligent analysis, it is possible to comprehensively monitor and evaluate the use status of central venous catheters and their related risks, thereby improving the safety and accuracy of clinical treatment. By acquiring central venous catheter case data and performing physical sign extraction and catheter structure feature analysis, it not only provides basic data for subsequent risk assessment, but also comprehensively grasps the patient's physiological characteristics and the structural characteristics of the catheter, laying a good foundation for accurate analysis. Through the collection and analysis of venous ultrasound image data, the catheter specifications can be estimated in combination with the patient's physical sign data, and the specifications and position changes of the central venous catheter can be accurately grasped. This process helps to deeply understand the position and stability of the catheter in the blood vessel, and provides a reliable data basis for the subsequent calculation of the probability of catheter displacement. Based on the catheter displacement probability data, the mis-touch area data is further collected, the area contacted by the catheter is analyzed, and the abnormal state is identified. This analysis can help identify potential risks caused by abnormal catheter position, such as vascular damage, blood flow obstruction and other problems. Risk assessment based on abnormal state data helps to comprehensively evaluate various problems encountered by the catheter during use, such as thrombosis, infection, vascular damage and other potential complications, and enhance the safety of the treatment process. Therefore, the present invention optimizes a traditional method for constructing a central venous catheter intelligent decision-making model based on multi-source data, solves the problem of inaccurate estimation of central venous catheter specifications and inaccurate central venous catheter risk analysis in the traditional method for constructing a central venous catheter intelligent decision-making model based on multi-source data, and improves the accuracy of central venous catheter specification estimation and central venous catheter risk analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 A schematic diagram of the steps of a method for constructing a central venous catheter intelligent decision-making model based on multi-source data;

[0070] Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart;

[0071] Figure 3 for Figure 1 Detailed implementation steps of step S4 in FIG.

[0072] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0073] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0074] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0075] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0076] To achieve this, please refer to Figures 1 to 3 A method for constructing a central venous catheter intelligent decision-making model based on multi-source data includes the following steps:

[0077] Step S1: Acquire central venous catheter case data; extract case patient vital signs based on the central venous catheter case data to obtain case patient vital signs data; analyze central venous catheter structure characteristics based on the central venous catheter case data to obtain central venous catheter structure characteristic data;

[0078] Step S2: collecting venous ultrasound images of case patients according to the case patient's physical sign data to obtain venous ultrasound image data of case patients; estimating the specifications of central venous catheters based on the structural characteristic data of central venous catheters according to the case patient's venous ultrasound image data to obtain central venous catheter specification data; calculating the probability of catheter position displacement according to the central venous catheter specification data and the case patient's venous ultrasound image data to obtain central venous catheter displacement probability data;

[0079] Step S3: collecting the central venous catheter mis-touch area according to the central venous catheter displacement probability data to obtain the central venous catheter mis-touch area data; analyzing the central venous catheter abnormal state according to the central venous catheter mis-touch area data and the venous ultrasound image data of the case patient to obtain the central venous catheter abnormal state data;

[0080] Step S4: Perform a central venous catheter risk assessment based on the central venous catheter abnormal status data to obtain central venous catheter risk data; construct a central venous catheter intelligent decision-making model based on the central venous catheter risk data to obtain a central venous catheter intelligent decision-making model.

[0081] In the embodiment of the present invention, reference Figure 1 FIG. 1 is a flow chart of steps of a method for constructing a central venous catheter intelligent decision model based on multi-source data of the present invention. In this example, the method for constructing a central venous catheter intelligent decision model based on multi-source data includes the following steps:

[0082] Step S1: Acquire central venous catheter case data; extract case patient vital signs based on the central venous catheter case data to obtain case patient vital signs data; analyze central venous catheter structure characteristics based on the central venous catheter case data to obtain central venous catheter structure characteristic data;

[0083] In an embodiment of the present invention, the case data of the central venous catheter is obtained, and the relevant data is extracted from the real clinical patient cases through the case database of the medical institution or the dedicated data acquisition equipment. The acquired data should include the basic information of the patient (such as age, gender, medical history, etc.), the venous conditions in the case, and the relevant medical imaging data. These data include the placement of the venous catheter, ultrasound image data, etc. After the data is integrated, the data is desensitized to remove all personal sensitive information. Through data cleaning and preprocessing, effective and qualified case data are screened out, and physical signs are extracted from these case data. Medical imaging analysis tools (such as image processing software) are used to extract the physical sign data of the vein. The extracted data include the size, position, morphology of the vein, and the physiological characteristics of the patient (such as blood pressure, heart rate, etc.). At the same time, the imaging analysis technology is used to analyze the structural characteristics of the venous catheter in detail to obtain the size, morphology, direction of the central venous catheter and the structure of the surrounding tissues. These data constitute the structural characteristic data of the central venous catheter.

[0084] Step S2: collecting venous ultrasound images of case patients according to the case patient's physical sign data to obtain venous ultrasound image data of case patients; estimating the specifications of central venous catheters based on the structural characteristic data of central venous catheters according to the case patient's venous ultrasound image data to obtain central venous catheter specification data; calculating the probability of catheter position displacement according to the central venous catheter specification data and the case patient's venous ultrasound image data to obtain central venous catheter displacement probability data;

[0085] In the embodiment of the present invention, after extracting the patient's vital sign data, the venous ultrasound image is collected. The patient's veins are scanned in real time using a high-resolution ultrasound device to obtain high-definition venous image data. During the image acquisition process, the position, morphology, surrounding tissues, and relative position of the venous catheter are focused on. Through image processing and analysis software, combined with image registration technology, the collected venous image is matched with the existing venous catheter structure feature data. Then, based on these image data and the existing structural feature data, the specifications of the central venous catheter are estimated through geometric modeling and machine learning methods to obtain accurate catheter size data. This process uses image segmentation technology based on morphological analysis, and through feature extraction and pattern recognition algorithms, the specific specifications and morphological changes of the venous catheter are calculated. At the same time, the displacement probability of the catheter position is calculated using the venous image data, and the probability calculation model and image registration technology are applied to analyze the stability of the catheter at different positions, and the probability data of the displacement of the central venous catheter are obtained.

[0086] Step S3: collecting the central venous catheter mis-touch area according to the central venous catheter displacement probability data to obtain the central venous catheter mis-touch area data; analyzing the central venous catheter abnormal state according to the central venous catheter mis-touch area data and the venous ultrasound image data of the case patient to obtain the central venous catheter abnormal state data;

[0087] In an embodiment of the present invention, the probability data of central venous catheter displacement is obtained, and the catheter mis-touch area is collected, the standard of the catheter mis-touch area is defined, and the venous catheter mis-touch area is marked by an image processing tool. Then, based on the patient's venous ultrasound image data and combined with the precise calculation results of the catheter position, the area of ​​catheter mis-touch is analyzed. This process involves image region segmentation technology, which is compared with case data to find the adjacent tissues touched by the catheter and calibrate them. Next, based on the data of the mis-touch area, the abnormal state of the central venous catheter is analyzed. This analysis process uses a multi-dimensional algorithm to evaluate whether the central venous catheter is in an abnormal state through a comprehensive analysis of the venous image data and the mis-touch area data, including problems such as blood flow obstruction and damage caused by the catheter, and obtain abnormal state data.

[0088] Step S4: Perform a central venous catheter risk assessment based on the central venous catheter abnormal status data to obtain central venous catheter risk data; construct a central venous catheter intelligent decision-making model based on the central venous catheter risk data to obtain a central venous catheter intelligent decision-making model.

[0089] In an embodiment of the present invention, after the abnormal state data of the central venous catheter is obtained, a risk assessment of the central venous catheter is performed. Data analysis tools and algorithm models are used to comprehensively evaluate the abnormal state of the catheter and analyze the risk of complications caused by the abnormal state. This process relies on a machine learning model to evaluate the combination of different risk factors by inputting abnormal state data of the catheter, patient vital signs data and imaging data. The algorithm quantifies the risks faced by the catheter by comparing the risk situations in different cases and combining standard clinical processes. Finally, based on the risk assessment results, a central venous catheter intelligent decision-making model is constructed. Based on the evaluation data, the model uses advanced decision support systems (such as rule-based systems or deep learning models) to optimize catheter management and provide personalized risk warnings.

[0090] Preferably, step S1 comprises the following steps:

[0091] Step S11: Acquire central venous catheter case data;

[0092] Step S12: performing case desensitization processing according to the central venous catheter case data to obtain central venous catheter case desensitization data;

[0093] Step S13: extracting the case patient's vital signs based on the central venous catheter case desensitization data to obtain the case patient's vital signs data;

[0094] Step S14: performing a central venous catheter structural feature analysis based on the central venous catheter case desensitized data to obtain central venous catheter structural feature data.

[0095] In an embodiment of the present invention, case data related to central venous catheters are extracted from a hospital database, an electronic medical record system (EMR) or a medical imaging system (PACS). These data generally include basic information of the patient (such as age, gender, length of stay, surgical history), catheter type, usage, catheter placement, surgical records, treatment plans, follow-up records, and ultrasound imaging data. Using data interface technology (such as the HL7 protocol or the FHIR standard), these data are extracted from different systems and formatted uniformly. Ensure the integrity and accuracy of the data, remove irrelevant data and mark them. The desensitization of central venous catheter case data uses standard de-identification technology. In this step, personal identity information is encrypted or replaced to ensure data privacy. Common desensitization methods include masking of data fields or replacing sensitive information such as name, ID number, contact information, etc. through encryption algorithms. In specific operations, a hash algorithm is used to process sensitive fields (such as patient ID, name, hospitalization number, etc.) so that the true identity cannot be reversely derived. At the same time, non-sensitive data (such as diagnosis results, treatment plans, etc.) are still retained. This operation complies with data protection regulations (such as GDPR or HIPAA) and ensures the availability of data for subsequent analysis. After the desensitization process is completed, the desensitized data of the central venous catheter case is obtained. Based on the central venous catheter case data that has been obtained and desensitized, the patient's physical signs are extracted from the case, and the patient's physiological signs, including blood pressure, heart rate, blood oxygen saturation and other conventional signs information, are extracted from the case using medical image processing tools and algorithms (such as image processing toolkits based on OpenCV or Matlab). For image data, the specific location and morphological characteristics of the venous catheter in the ultrasound image are extracted by image segmentation technology. Then, the diameter, length, shape, lumen state of the vein and the structural characteristics of the surrounding tissue are extracted by pattern recognition algorithms. In the process of physical sign extraction, the image is automatically analyzed by an automated algorithm to reduce human intervention, improve efficiency and accuracy, and obtain the physical sign data of the case patient. When performing structural feature analysis on the central venous catheter case data, imaging technology is used to analyze the morphology and structural characteristics of the catheter. This analysis is done by enhancing and magnifying the ultrasound images, using filtering technology to clearly display the outline of the catheter and surrounding tissues, and using geometric analysis methods to measure the dimensions of the catheter, such as diameter, curvature, length, etc., while evaluating the relationship between the catheter and surrounding tissues, such as whether there are signs of compression or damage. By analyzing multiple image frames, the three-dimensional structural data of the catheter is accurately obtained. In addition, combined with the patient's vital signs data, statistical analysis methods are used to explore the correlation between the catheter's structural characteristics and the patient's health status, and the structural characteristic data of the central venous catheter is obtained.

[0096] Preferably, step S2 comprises the following steps:

[0097] Step S21: collecting venous ultrasound images of the case patient according to the case patient's physical sign data to obtain venous ultrasound image data of the case patient;

[0098] Step S22: performing vein feature analysis based on the venous ultrasound image data of the case patient to obtain ultrasound image vein feature data;

[0099] Step S23: estimating the specifications of the central venous catheter based on the ultrasound image vein feature data and the central venous catheter structure feature data to obtain the central venous catheter specification data;

[0100] Step S24: Calculate the probability of catheter position displacement according to the central venous catheter specification data and the ultrasound image vein feature data to obtain central venous catheter displacement probability data.

[0101] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0102] Step S21: collecting venous ultrasound images of the case patient according to the case patient's physical sign data to obtain venous ultrasound image data of the case patient;

[0103] In an embodiment of the present invention, an ultrasonic diagnostic device is used to scan the patient. The ultrasonic device uses a high-frequency linear array probe that can provide high-resolution real-time images, which is particularly suitable for the detection of blood vessels and veins. According to the patient's specific physical signs, such as body shape, body position and health status, select a suitable probe frequency and imaging method (such as B-type ultrasonic imaging) to ensure that the details of the veins and their surrounding tissues can be accurately displayed. When collecting ultrasonic images, professional technicians operate the ultrasonic equipment to scan the patient's neck, chest or upper limbs, where common venous catheters are implanted, and ensure that the scanning angle and position are correct in order to obtain a clear venous image. During the image acquisition process, special attention is paid to the blood flow of the veins, the position of the catheter, and the structure of the surrounding tissues.

[0104] Step S22: performing vein feature analysis based on the venous ultrasound image data of the case patient to obtain ultrasound image vein feature data;

[0105] In the embodiment of the present invention, an image processing algorithm is used to pre-process the ultrasound image, including operations such as denoising, contrast enhancement and edge sharpening, so as to improve the image quality. The image segmentation technology is used to extract the vein area, and the boundary of the venous blood vessel is separated by threshold segmentation, edge detection or regional growth algorithm. The vein morphology is further described by morphological analysis method, and the geometric characteristics of the vein are extracted, including the diameter, curvature, length, wall thickness, etc. of the vein. For blood flow information, the blood flow velocity and direction are measured by Doppler ultrasound, and the hemodynamic characteristics are analyzed by time-velocity spectrogram. In addition, the spatial relationship between the vein and the surrounding structure is also extracted, such as the smoothness of the vein wall, whether the vascular wall is deformed, etc. This process uses image processing platforms such as OpenCV and Matlab, and extracts the key feature data of the vein through an automated method and stores it as structured data.

[0106] Step S23: estimating the specifications of the central venous catheter based on the ultrasound image vein feature data and the central venous catheter structure feature data to obtain the central venous catheter specification data;

[0107] In an embodiment of the present invention, when estimating the specifications of a central venous catheter based on the characteristic data of a venous ultrasound image, geometric modeling is performed on the venous structure identified in the ultrasound image. This process measures the diameter, depth, tortuosity and other data of the vein, and combines it with the patient's vital signs data to infer the size and specifications of the catheter. To this end, a geometry-based algorithm (such as least squares fitting) is used to fit and estimate the morphological characteristics of the catheter. In addition, taking into account the individual differences of different patients, by comparing the known standard size data of the central venous catheter (such as the inner diameter, outer diameter, length, etc. of the catheter), combined with the venous contour of the ultrasound image, a regression analysis method is used to further refine the estimation of the specifications of the central venous catheter.

[0108] Step S24: Calculate the probability of catheter position displacement according to the central venous catheter specification data and the ultrasound image vein feature data to obtain central venous catheter displacement probability data.

[0109] In an embodiment of the present invention, when calculating the probability of catheter position displacement based on the specification data of the central venous catheter and the vein characteristic data, the vein position in the venous ultrasound image and the positioning information of the catheter relative to the vein are used to calculate the relative position change of the catheter using a geometric analysis method. By measuring the two-dimensional coordinates and three-dimensional spatial position of the catheter in the image, combined with the patient's vital sign data, the deviation between the initial position and the current position of the catheter is determined. In order to estimate the probability of catheter displacement, a statistical model based on the Monte Carlo method is used to simulate the catheter position offset under different conditions. Through multiple simulations and probability statistics, the probability and degree of catheter displacement are obtained, and the risk level of catheter displacement is further accurately calculated. This process uses programming environments such as Matlab and Python, combined with numerical calculations and Monte Carlo methods, to complete the quantitative analysis of the probability of catheter displacement.

[0110] Preferably, step S22 includes the following steps:

[0111] Step S221: collecting image texture features according to the venous ultrasound image data of the case patient to obtain ultrasound image texture feature data;

[0112] Step S222: performing venous blood vessel edge feature analysis based on the ultrasound image texture feature data and the venous ultrasound image data of the case patient to obtain venous blood vessel edge feature data;

[0113] Step S223: performing contour feature analysis according to the venous blood vessel edge feature data and the ultrasonic image texture feature data to obtain venous blood vessel contour data;

[0114] Step S224: Performing vein feature analysis based on the vein contour data and the vein edge feature data to obtain ultrasound image vein feature data.

[0115] In an embodiment of the present invention, the acquired ultrasound image is preprocessed by an image processing tool, including operations such as noise removal, contrast enhancement and edge sharpening, so as to improve the image quality. Texture features in ultrasound images are extracted using texture analysis methods, such as gray level co-occurrence matrix (GLCM) or local binary pattern (LBP) and other technologies. Specifically, the gray level co-occurrence matrix obtains characteristic parameters describing textures, such as contrast, energy, entropy and correlation, by analyzing the gray level change pattern of pixel pairs in the image; the local binary pattern describes the local texture structure of the image through the pixel relationship in the local area. In ultrasound images, texture features help to identify the uniformity of the vascular wall, the pattern of blood flow in the vascular cavity and its contrast with the surrounding tissue. Using programming tools such as Matlab or Python, combined with image processing libraries (such as OpenCV, scikit-image, etc.), texture feature data is automatically extracted, and when performing venous vascular edge feature analysis based on ultrasound image texture feature data and case patient venous ultrasound image data, venous vascular edge detection is performed. Using the Canny edge detection algorithm or the Sobel operator, the contour information of the venous vascular is obtained by identifying the changes in the venous edge in the ultrasound image. At this point, the image needs to be further smoothed to reduce noise interference and ensure the accuracy of edge detection. After edge detection, the image contour tracking algorithm (such as the Ramer-Douglas-Peucker algorithm) is used to finely extract the blood vessel edge to obtain more accurate blood vessel edge data. By analyzing the boundary between the blood vessel and the surrounding tissue in the image, the edge features of the venous blood vessels, such as the smoothness and sharpness of the edge and the contrast between the blood vessel and the surrounding tissue, are obtained. The edge feature analysis is performed using the image processing library in Matlab or Python (such as OpenCV, scikit-image, etc.), and the results are converted into numerical data. When the contour feature analysis is performed based on the edge feature data of the venous blood vessels and the texture feature data of the ultrasound image, the overall contour of the venous blood vessels is further analyzed by combining the edge feature data and the texture data. Through the morphological transformation technology, the blood vessel contour is expanded, eroded, opened, etc. to eliminate small area noise and defects and extract a clearer blood vessel contour. On this basis, the blood vessel contour is modeled using polynomial fitting or spline interpolation algorithms (such as B-spline curves or Bezier curves) to obtain a continuous description of the blood vessel contour. By calculating the smoothness and curvature of the vascular contour and the changes in the inner diameter of the vascular lumen, the regularity of the vascular morphology and its impact on catheter placement are evaluated. The contour feature data obtained includes the precise size, curvature and boundaries of the surrounding tissue of the blood vessel, which are synthesized by combining the texture feature data with data analysis software (such as Matlab) to obtain the contour data of the venous blood vessel.When performing vein feature analysis based on venous contour data and venous edge feature data, the contour and edge feature data are integrated to analyze the overall structure of the venous vessels. The geometric features of the venous vessels are extracted by analyzing the shape, diameter, wall thickness and relationship with the surrounding tissues of the venous vessels. This process can use morphological analysis tools (such as segmentation technology based on Voronoi diagrams) to identify the opening and branching locations of the venous vessels, and further analyze the tributaries, bends and changes in the surrounding tissues of the vessels. In order to accurately extract the characteristics of the venous vessels, curve fitting or morphological contour analysis methods are used to extract the geometric properties of the vessels, such as the size, shape, continuity, etc. of the vessels, from the edges and contours of the veins. Combined with the texture features of the ultrasound image, the uniformity of the vascular wall and the distribution of blood flow are evaluated. Image processing tools and algorithms in the analysis process, such as Matlab's Image Processing Toolbox and the scikit-image library in Python, can realize the extraction and analysis of such vein features.

[0116] Preferably, step S222 includes the following steps:

[0117] Calculate the grayscale value of the ultrasound image according to the venous ultrasound image data of the case patient, so as to obtain the grayscale value data of the venous ultrasound image;

[0118] Calculate the global grayscale value standard deviation of the ultrasound image according to the grayscale value data of the venous ultrasound image to obtain the global grayscale value standard deviation data of the ultrasound image;

[0119] Performing ultrasound image grayscale contrast analysis based on ultrasound image global grayscale value standard deviation data to obtain ultrasound image grayscale contrast data;

[0120] Calculate the gradient change of texture feature direction according to the texture feature data of the ultrasonic image to obtain the gradient change data of the texture feature direction of the image;

[0121] The venous blood vessel edge feature data is obtained by performing venous blood vessel edge feature analysis based on the image texture feature directional gradient change data and the ultrasound image gray value contrast data.

[0122] In an embodiment of the present invention, after obtaining the ultrasonic image data, the noise and clutter are removed by the image preprocessing step to improve the image quality. The color ultrasonic image is processed by the grayscale conversion method and converted into a single-channel grayscale image. During the conversion process, the RGB image is grayed using the weighted average method (for example, the commonly used weighting coefficients 0.299, 0.587, and 0.114) to obtain the grayscale value of each pixel, and the grayscale distribution data of the image is obtained by calculating the grayscale value of each pixel of the image. Using image processing tools (such as the rgb2gray function of Matlab or the cv2.cvtColor method in the OpenCV library of Python), the ultrasonic image is efficiently converted into a grayscale image, and the grayscale value information of each pixel is exported. After obtaining the grayscale value data of the venous ultrasonic image, the grayscale value of the image is statistically analyzed. The standard deviation of the grayscale value is calculated using the grayscale histogram of the image, and the standard deviation is used as an indicator to measure the degree of change in the grayscale distribution of the image. The calculation process includes: counting the frequency of each gray value in the image, and then calculating the mean of the image gray value based on the frequency of each gray value; then, calculating the square of the difference between each gray value and the mean, and weighted averaging to obtain the global gray value standard deviation. The standard deviation reflects the gray fluctuation in the image. A higher standard deviation indicates a larger brightness change in the image, and a lower standard deviation indicates a more uniform image. This operation is calculated using the std() function in Matlab or the numpy.std() method in Python, and the result is a single value, which represents the global gray standard deviation of the image. The gray contrast of the image is further analyzed using the global gray value standard deviation data of the ultrasound image. Gray contrast is a measure of the brightness difference of the image, usually defined as the difference between the maximum gray value and the minimum gray value in the image. When calculating, the maximum and minimum gray values ​​are found through the gray value matrix of the image. The contrast calculation formula is: contrast = maximum gray value - minimum gray value, or some weighted contrast indicators such as root mean square error (RMS) are used to measure the contrast change in the image. Based on the grayscale standard deviation of the image, a higher standard deviation value will produce a larger grayscale contrast. This operation is calculated using the range() function of Matlab or the numpy.ptp() method in Python to obtain the grayscale contrast data of the venous ultrasound image. When analyzing the directional gradient change of texture features of ultrasound images, texture features are extracted from the image. The commonly used methods are gray-level co-occurrence matrix (GLCM) or local binary pattern (LBP). The gradient change of texture features is calculated, and the texture changes in different directions in the image are particularly concerned. The gradient change is calculated by applying the Sobel operator or gradient convolution to the texture feature image to obtain the gradient change of the image in the horizontal, vertical and diagonal directions. For the gradient in each direction, the change amplitude of the pixel grayscale value is calculated to analyze the directional change characteristics of the image texture.By calculating the directional gradient change data of texture features, the roughness of the blood vessel wall, the direction of blood flow, and the boundary changes with the surrounding tissues are further revealed. This process is implemented by the gradient() function in Matlab or numpy.gradient() in Python to obtain the data of the directional gradient change of texture features. By combining the directional gradient change data of texture features with the gray value contrast data of ultrasound images, the edge features of venous blood vessels are further analyzed, and the edge area of ​​blood vessels is extracted based on the directional gradient change data of texture features. For the blood vessel part in the ultrasound image, the texture features usually show obvious directional differences, while the grayscale changes of the edge part are more drastic. By fusing these two sets of data, edge detection algorithms (such as Canny or Sobel algorithms) are used to accurately locate the edge position of the blood vessel. Subsequently, the sharpness, curve changes, and smoothness of the blood vessel edge are analyzed by combining the edge position with the grayscale contrast. The edge is optimized by morphological operations (such as dilation and erosion) to remove small-range noise and retain the real blood vessel edge. This operation is performed by the edge() function of Matlab or cv2.Canny() of Python to obtain the edge feature data of the venous blood vessel.

[0123] Preferably, step S23 includes the following steps:

[0124] Step S231: performing image vein fine-grained analysis according to the ultrasound image vein feature data to obtain ultrasound image vein fine-grained data;

[0125] Step S232: performing image vein wall thickness detection based on the ultrasound image vein fine-grained data to obtain ultrasound image vein wall thickness data;

[0126] Step S233: performing image vein position measurement on the ultrasound image vein feature data, thereby obtaining ultrasound image vein position data;

[0127] Step S234: performing image vein depth detection according to the ultrasound image vein position data to obtain ultrasound image vein depth data;

[0128] Step S235: estimating the specifications of the central venous catheter based on the central venous catheter structural feature data based on the ultrasound image vein depth data to obtain the central venous catheter specification data.

[0129] In an embodiment of the present invention, when fine-grained analysis is performed on the vein features in the ultrasound image, the vein region in the image is extracted, and the information of each pixel point of the image is further refined. A high-precision segmentation algorithm, such as a convolutional neural network (CNN) model based on deep learning or a classic edge detection method (such as Canny edge detection) is used to identify the edge and internal structure of the vein region. By performing small-scale pixel-by-pixel processing on the image, it is possible to analyze slight changes in the vein wall, such as morphological fluctuations or small lesions in the venous lumen. In image processing, a local feature extraction method (such as a local binary pattern LBP) is used to perform texture analysis on the neighborhood around each pixel, thereby obtaining fine-grained vein structure data. These fine-grained features will provide accurate regional information for subsequent steps such as vein wall thickness measurement and depth detection. This process can be implemented through image processing tools such as the regionprops() function in Matlab or the skimage.measure.label() function in Python, to perform fine segmentation and analysis to obtain fine-grained data of the vein. When using the obtained fine-grained vein data for vein wall thickness detection, it is necessary to define the measurement area of ​​the vein wall. By performing high-precision segmentation on the image, the boundary between the venous lumen and the surrounding tissue is identified, and a gradient algorithm based on grayscale value changes (such as the Sobel operator) is applied in this area to detect the thickness of the venous wall. The venous wall in the image usually shows obvious grayscale changes, and the boundary position of the venous wall is accurately determined by calculating the rate of change of the grayscale value. The method for calculating the thickness of the venous wall is usually to convert the pixel unit into an actual length unit, such as millimeters or centimeters, based on the physical scale of the image. The thickness of the venous wall in the ultrasound image is obtained by measuring the wall thickness of the vein at different positions multiple times. This step uses the imgradient() function in Matlab or the cv2.Sobel() function in Python to complete the image gradient calculation, and combines the distance transformation method to obtain the accurate thickness value of the venous wall. According to the vein feature data in the ultrasound image, when measuring the vein position, the position coordinates of the vein should be extracted from the regional segmentation result of the vein. Through the edge detection of the vein, the starting point and the end point of the vein are accurately obtained, and the position of the central axis of the vein is calculated. Commonly used algorithms include the Canny operator based on edge detection, or the method based on the active contour model (SnakeModel) to further refine the position of the vein. The measurement of vein position needs to be combined with the resolution of the image to convert the image coordinate system into actual physical coordinates. By measuring the depth of the vein, the position distribution of the vein in the image is determined. This step is combined with morphological operations (such as corrosion and expansion) to optimize the detection results and remove the misidentified areas.Use the bwboundaries() function in Matlab or skimage.measure.regionprops() in Python to extract the boundary data of the vein and obtain the specific position coordinates of the vein in the image. When performing vein depth detection based on the vein position data in the ultrasound image, the position data of the vein in the image is used to determine the depth of the vein relative to the probe. Depth detection usually relies on the two-dimensional grayscale distribution of the ultrasound image, and the brightness value in the image is inversely proportional to the depth of the object. In the depth measurement process, it is necessary to obtain the actual depth of the vein by calibrating the depth scale of the image based on the propagation speed and reflection principle of the ultrasound. In the specific operation, the vein depth in the image is obtained by calculating the vertical distance from the vein position to the ultrasound probe. This step uses image calibration technology to convert the pixel values ​​in the image into actual depth values. Common methods include calibration through reference objects of known depth. The depth information of the image is used to further infer the insertion depth and position of the venous catheter. In Matlab, the imdistline() tool is used to assist in measuring the depth, and in Python, scipy.ndimage.measurements.center_of_mass() can be used to calculate the depth. After obtaining the vein depth data, the specifications of the central venous catheter are estimated. According to the depth information of the vein, combined with the diameter, shape and position of the vein, the specifications of the central venous catheter are inferred. In this process, it is necessary to obtain the size data of the vein, such as the inner diameter and outer diameter of the blood vessel. Then, based on the depth information of the blood vessel, the specifications of the catheter suitable for insertion at different depths are inferred. The estimation of the catheter specifications is usually calculated based on empirical formulas or standard values ​​in the medical database, and the matching catheter specifications are given in combination with the depth and morphological information of the vein. This process involves complex geometric models and biomechanical principles, and factors such as the diameter, curvature and material properties of the catheter need to be comprehensively considered. By combining the image depth and size information, interpolation or regression analysis methods are used to obtain the specifications of the central venous catheter.

[0130] Preferably, step S3 comprises the following steps:

[0131] Step S31: collecting the central venous catheter mis-touch area according to the central venous catheter displacement probability data to obtain the central venous catheter mis-touch area data;

[0132] Step S32: estimating the probability of local thrombosis according to the central venous catheter mis-touch area data and the venous ultrasound image data of the case patient, and obtaining ultrasound image local thrombosis probability data;

[0133] Step S33: performing central venous catheter abnormal state analysis based on the local thrombosis probability data of the ultrasound image and the central venous catheter mis-touch area data to obtain central venous catheter abnormal state data.

[0134] In an embodiment of the present invention, after obtaining the probability data of central venous catheter displacement, it is necessary to further analyze the area where the mis-touch occurs for the displaced catheter. The mis-touch area refers to the position where the catheter contacts the surrounding vascular wall, heart wall or anatomical structure after displacement, thereby causing tissue damage. Based on the probability data of catheter displacement, the central venous catheter position is accurately aligned with the venous anatomical structure using image registration technology. The patient's venous ultrasound image is processed using image segmentation and registration algorithms, and the actual position of the catheter is overlapped with the image data of the vascular wall. By calculating the offset and relative position of the catheter relative to the vascular wall, the potential mis-touch area is identified. This process can use an image processing method based on threshold segmentation, combined with morphological operations to highlight the mis-touch area. In Matlab, the imoverlay() function superimposes the mis-touch area with the image, and the skimage.measure.label() in Python can also be used to calibrate the position of the mis-touch area to obtain the mis-touch area data. After obtaining the contact area data of the central venous catheter, it is necessary to further evaluate whether these areas have the risk of local thrombosis. Thrombosis usually occurs where the catheter contacts the venous wall or where the blood flow is slower. Therefore, by combining the location information of the catheter contact area and the hemodynamic characteristics in the ultrasound image, the risk of thrombosis can be estimated. The blood flow velocity analysis algorithm is used to evaluate the blood flow pattern in the venous ultrasound image, and the location of the thrombosis is inferred by combining the diameter of the vascular lumen and the pressure point of the catheter contact. This analysis can be based on a fluid dynamics simulation method to evaluate the probability of thrombosis by calculating the area where the catheter contacts the vascular wall and combining the local blood flow deceleration area. Commonly used image processing methods include gray-level co-occurrence matrix (GLCM) analysis and texture analysis to extract local features of the vascular wall. In addition, physiological parameters (such as blood viscosity, blood flow rate, etc.) are combined with the blood flow pattern of the ultrasound image to quantify the risk of thrombosis. Use Matlab's regionprops() function or Python's scipy.ndimage library to evaluate local thrombosis and obtain local thrombosis probability data. Based on the local thrombosis probability data obtained and the central venous catheter mis-touch area data obtained in step S31, the abnormal state of the central venous catheter is analyzed, and the two data are fused. By weighted average method or logistic regression analysis, the mis-touch area and thrombosis probability are combined to comprehensively evaluate the abnormal risk of the catheter. Specifically, if the catheter is in the mis-touch area and the probability of local thrombosis is high, it will cause catheter dysfunction or complications. This analysis can use statistical methods, such as Bayesian networks or multivariate regression analysis, to quantify the probability of abnormal conditions. Image data and hemodynamic data can be fused and analyzed through a specific algorithm to obtain the abnormal state of the catheter under different conditions.The evaluation was further enhanced using flow velocity analysis in ultrasound images, changes in venous wall morphology, and the relative position of the catheter to the vessel wall. The fitlm() function in Matlab or the sklearn.linear_model.LogisticRegression model in Python was used to implement regression analysis and derive catheter abnormality data.

[0135] Preferably, step S32 includes the following steps:

[0136] Step S321: collecting ultrasound image vein morphology change data based on the central venous catheter mis-touch area data and the case patient's venous ultrasound image data to obtain ultrasound image vein morphology change data;

[0137] Step S322: performing venous puncture detection according to the ultrasound image venous morphology change data to obtain ultrasound image venous puncture data;

[0138] Step S323: performing venous vessel injury detection according to the ultrasonic image venous vessel puncture data to obtain ultrasonic image venous vessel injury data;

[0139] Step S324: performing vasospasm analysis based on the ultrasound image venous vascular injury data to obtain ultrasound image vasospasm probability data;

[0140] Step S325: estimating venous blood flow obstruction according to the central venous catheter mis-touch area data and the ultrasound image venous vascular injury data to obtain venous blood flow obstruction data;

[0141] Step S326: estimating the probability of local thrombosis based on the ultrasound image vascular spasm probability data and the venous blood flow obstruction data, and obtaining the ultrasound image local thrombosis probability data.

[0142] In an embodiment of the present invention, after obtaining the data of the central venous catheter mis-touch area, it is necessary to further evaluate the impact of the catheter displacement on the vascular morphology. The change in vein morphology is judged by the changes in the vascular contour, lumen morphology and vascular wall in the ultrasound image. In this step, the patient's venous ultrasound image is processed to identify the boundary of the blood vessel and compare it with the normal venous morphology. Image processing includes edge detection and morphological operations (such as expansion, corrosion, etc.), thereby highlighting the morphological characteristics of the blood vessel. In order to obtain the change in vein morphology, by comparing the venous images before and after the catheter is placed, the image difference analysis method is used to determine the change in the blood vessel after the catheter is mis-touched. The specific operation uses the edge detection algorithm (such as Canny()) and findContours() in Python's OpenCV library to extract vein morphology data, and performs morphological transformation at the same time to obtain a data set of vascular morphology changes. Based on the ultrasound image vein morphology change data, further detect whether the venous blood vessel has been punctured. In this step, puncture detection is mainly performed by judging whether there are signs of damage or rupture of the venous wall, and using the vascular edge information extracted from the vein morphology change data, combined with the intensity change characteristics in the ultrasound image, to detect whether there are abnormal vascular wall penetration or puncture marks. Blood vessel puncture is usually accompanied by local vascular wall rupture or penetration, and the image shows an area with obvious contrast difference from the surrounding vascular tissue. This detection is performed by using mutation analysis of image grayscale values ​​or changes in image texture features (such as grayscale co-occurrence matrix, local binary pattern). Through the analysis of these features, combined with the time series data of ultrasound images, it is judged whether the blood vessel is punctured. After the vascular puncture is detected, it is further analyzed whether there is vascular damage. Vascular damage is usually accompanied by vascular wall rupture, bleeding or abnormal blood flow. In this process, it is necessary to further analyze the vascular area in the ultrasound image in detail, and locate the specific location of venous damage through the regional data of vascular puncture. Then, the high-frequency detail information of the ultrasound image (such as tissue echo, grayscale value change, etc.) is used to analyze the tissue characteristics of the damaged area, identify the abnormal morphology of the vascular wall, bleeding, and signs of blood flow interruption. The specific morphology of the damaged area is detected by analyzing the local grayscale value difference, echo intensity change or high-frequency component of the image. Common processing methods include using the region growing algorithm in image processing, image segmentation technology, etc. The regionprops() function in Matlab combined with bwlabel() can be used for region identification and extraction of damaged regional data. Vascular damage is often accompanied by vascular spasm, especially during intravenous catheter operation, where changes in external pressure and blood flow lead to vascular contraction. By analyzing the morphological changes of blood vessels in ultrasound images, it is possible to detect whether the vessels show signs of spasm.Specifically, vasospasm can cause the lumen of the blood vessel to narrow, which is manifested as the inward contraction of the blood vessel wall in the image, resulting in poor blood flow. Based on the venous vascular injury data, the morphological characteristics of the veins in the image (such as blood vessel diameter, wall thickness, etc.) are further analyzed to identify whether contraction occurs. Spasm is identified by using methods such as image contrast analysis or changes in arterial and venous resolution. In Matlab or Python, the regionprops() function is used to quantitatively analyze the changes in blood vessel diameter, and blood flow velocity data (such as Doppler signals) are combined to further confirm whether vasospasm exists. By quantifying these changes, the probability data of vasospasm is obtained. Venous blood flow obstruction is usually caused by factors such as catheter mistouch, vascular injury or spasm. In this step, the central venous catheter mistouch area data is combined with the venous vascular injury data to evaluate the obstruction of venous blood flow. By analyzing the area where the catheter contacts the blood vessel wall and the site of vascular injury, it is inferred whether the blood flow is affected. Venous blood flow obstruction is usually accompanied by a decrease in local blood flow velocity or reverse flow. In order to estimate the severity of blood flow obstruction, a blood flow simulation method based on image processing is used. Combined with the geometric structure of the blood vessels, hemodynamic characteristics and the location of vascular wall damage, the resistance and flow change of blood flow are calculated using a fluid dynamics model. The degree of blood flow obstruction is quantified by the flow velocity information (such as Doppler images) and the change of blood vessel diameter in the ultrasound image. Numerical solution methods such as scipy.integrate.solve_ivp() in Python are used to simulate hemodynamics and obtain quantitative data of blood flow obstruction. After obtaining the probability data of vascular spasm and venous blood flow obstruction data, the probability of thrombosis is further predicted. Thrombosis is usually closely related to the slowing of blood flow and damage to the vascular lumen. Combined with the areas of vascular spasm and blood flow obstruction, the physiological model is used to infer the risk of local thrombosis. Based on image texture features, hemodynamic analysis and local pressure changes, combined with the morphological data of venous blood vessels, the probability of thrombosis is obtained. In this process, statistical methods such as regression analysis or Bayesian network are used to integrate different data sources to further improve the accuracy of predicting the risk of thrombosis. A comprehensive analysis is performed by combining the changes in blood flow velocity, vascular morphology, and blood rheology models in ultrasound images.

[0143] Preferably, step S33 includes the following steps:

[0144] Step S331: performing venous thrombosis enlargement analysis based on the local thrombosis probability data of the ultrasound image to obtain venous thrombosis enlargement data;

[0145] Step S332: performing venous vessel blockage detection based on the venous vessel thrombus enlargement data to obtain venous vessel blockage data;

[0146] Step S333: performing central venous catheter function loss detection according to the venous vessel blockage data and the venous vessel thrombosis enlargement data to obtain central venous catheter function loss data;

[0147] Step S334: estimating the probability of venous infection according to the local thrombosis probability data of the ultrasound image and the central venous catheter mis-touch area data to obtain venous infection probability data;

[0148] Step S335: performing central venous catheter abnormal state analysis based on the venous infection probability data and the central venous catheter function loss data to obtain central venous catheter abnormal state data.

[0149] In the embodiment of the present invention, the local thrombus formation probability data of the ultrasound image is used to analyze whether the thrombus has grown. The thrombus growth is usually accompanied by local blood vessel compression or poor blood flow, which leads to the expansion of the thrombus volume. This process analyzes the local thrombus formation probability data, and uses the thrombus morphology changes and local blood flow velocity change information in the image to infer whether the thrombus has expanded. The basis for judging the thrombus growth includes changes in the vascular cavity, the fuzzy or irregular morphology of the thrombus edge, and the gray value change of the thrombus area in the ultrasound image. Accurate thrombus growth data is obtained, and the thrombus area is extracted by image segmentation technology. The degree of thrombus growth is quantified by comparing image data of different periods. The consequence of thrombus growth leads to complete blockage of the blood vessel. Based on the venous thrombus growth data, vascular blockage detection is performed. Blockage is usually manifested as complete occlusion of the vascular lumen, resulting in blood flow cessation or severe slowing. The vascular lumen diameter information and hemodynamic data (such as Doppler ultrasound measurement data) in the image are used to identify whether there is a complete blockage phenomenon. The detection method includes calculating the stenosis and flow velocity changes of the blood vessel. If the vascular lumen is completely closed, it indicates that the blood vessel is blocked. In order to accurately extract the blocked area, the image segmentation technology is used to further analyze the morphological changes of the vascular lumen area and compare it with the morphology of normal blood vessels to determine whether there is blood flow interruption. The cv2.findContours() function in Matlab or Python can be used to extract the vascular lumen area, and further analyze whether there is blockage in combination with blood flow velocity information. The severity of vascular blockage is quantified by blood flow velocity, vascular lumen morphology and other data. The loss of central venous catheter function is usually caused by poor blood flow due to vascular blockage or thrombus enlargement. In this step, the degree of central venous catheter function loss is evaluated by venous vessel blockage data and thrombus enlargement data, and the degree of vascular blockage is analyzed. Combined with the catheter position and hemodynamic model, it is determined whether the catheter function is limited. Thrombus enlargement and blockage lead to blockage of the fluid in the catheter, thereby affecting its normal function. By comparing the blood flow data in the blood vessel (such as blood flow velocity, pressure distribution, etc.) with the expected function of the catheter, it is evaluated whether the catheter has lost function. This analysis combines changes in vascular resistance, blood flow velocity and the geometry of the catheter. Thrombosis and central venous catheter mis-touch areas both lead to venous infection. In this step, the probability of venous infection is estimated based on the local thrombosis probability data and catheter mis-touch area data in the ultrasound image, combined with the risk of bacterial infection in the vascular lumen. Venous infection usually occurs at the contact point between the catheter and the vascular wall, especially at the site of thrombosis. The catheter mis-touch area is analyzed in detail to assess the local infection risk. Secondly, based on the venous wall morphology, thrombosis and local flow data in the image, the probability of bacterial proliferation and infection in these areas is inferred. Quantitative analysis is performed by extracting image features and combining them with the physiological model of microbial infection.Python's scipy library can be used for data analysis. The probability of intravenous infection is calculated by combining Bayesian networks or regression models. The estimated value of infection is obtained based on the combined data of catheter position, thrombosis, and mis-touch area. The abnormal state of central venous catheters is usually manifested as infection, blood flow obstruction, loss of function, etc. In this step, a comprehensive analysis of the abnormal state of the catheter is performed by combining the intravenous infection probability data and the central venous catheter loss of function data. Intravenous infection and loss of function are usually manifestations of catheter abnormalities, affecting the overall function of the catheter. By integrating the intravenous infection and catheter loss of function data, it is analyzed whether the catheter is in an abnormal state and the severity of the abnormality is evaluated. By constructing a comprehensive risk assessment model, considering multiple factors such as infection, thrombosis, and catheter displacement, the comprehensive abnormal state of the catheter is obtained.

[0150] Preferably, step S4 comprises the following steps:

[0151] Step S41: Perform venous function impairment detection based on the abnormal state data of the central venous catheter to obtain venous function impairment data;

[0152] Step S42: Estimating venous catheter complications according to the venous function impairment data and the central venous catheter abnormal state data to obtain central venous catheter complication data;

[0153] Step S43: performing a central venous catheter risk assessment based on the venous catheter complication data and the venous function impairment data to obtain central venous catheter risk data;

[0154] Step S44: construct a central venous catheter intelligent decision-making model based on the central venous catheter risk data to obtain a central venous catheter intelligent decision-making model.

[0155] As an example of the present invention, refer to Figure 3 As shown, in this example, step S4 includes:

[0156] Step S41: Perform venous function impairment detection based on the abnormal state data of the central venous catheter to obtain venous function impairment data;

[0157] In an embodiment of the present invention, the detection of venous function impairment is based on the analysis of known abnormal state data of the central venous catheter, and is mainly achieved through the physiological model and hemodynamic model of venous function. In this step, the functional changes of the venous system are analyzed based on the abnormal state data of the central venous catheter, such as catheter position deviation, thrombus enlargement, local blood flow obstruction, etc. By comparing the normal venous function data and the abnormal state data, the changes in the elasticity of the blood vessels, blood flow velocity, vascular pressure, etc. are identified, and then it is determined whether the venous function is impaired. Venous function impairment is usually accompanied by hemodynamic changes, including decreased blood flow velocity, abnormal increase or decrease in pressure, etc. By using ultrasonic Doppler blood flow velocity measurement and pressure sensor data, combined with image feature analysis technology (such as vascular cavity analysis based on threshold segmentation), it is determined whether the vein has functional impairment.

[0158] Step S42: Estimating venous catheter complications according to the venous function impairment data and the central venous catheter abnormal state data to obtain central venous catheter complication data;

[0159] In an embodiment of the present invention, the estimation of complications of venous catheters is based on the combined analysis of venous function impairment data and catheter abnormal state data. In this step, by analyzing the type and degree of venous function impairment, combined with the state of the catheter itself, such as position displacement, infection, thrombosis, etc., the complications caused are identified. For example, thrombosis or catheter mistouch leads to venous puncture, thrombus displacement or local infection, which in turn leads to blood flow obstruction or loss of catheter function. By analyzing the dynamic changes of venous blood flow, combined with the morphology, elasticity, thickness and other information of the vascular wall, it is determined whether the catheter has potential complications. For venous function impairment data, it is usually necessary to track the use status and complications of the catheter in real time through a dynamic monitoring system (such as hemodynamic simulation, blood viscosity model, etc.). In addition, machine learning algorithms (such as support vector machines, decision trees) are used to quantitatively estimate complications, and the model input data include blood flow velocity, catheter position, thrombus size, infection risk, etc.

[0160] Step S43: performing a central venous catheter risk assessment based on the venous catheter complication data and the venous function impairment data to obtain central venous catheter risk data;

[0161] In an embodiment of the present invention, the risk assessment of the central venous catheter combines the analysis results of venous function damage and catheter complications, mainly relying on the risk assessment model, and calculating the comprehensive risk of catheter abnormality by evaluating the weights of different factors. In this step, the venous function damage data and complication data are weighted and scored. A weight is assigned to each indicator according to the severity of each type of damage (such as abnormal blood flow, catheter blockage, infection, etc.). Generally, these weights are derived from clinical studies and empirical data. Then, the overall risk of the central venous catheter is calculated by weighted scoring. The evaluation model is constructed by regression analysis, decision tree model or neural network. Common regression analysis methods include Logistic regression or Cox regression model, which obtains catheter risk data by modeling the influencing factors of venous function damage and complications. In addition, based on historical case data and medical records, deep learning models such as LSTM (long short-term memory network) are trained to model time series data to predict the risks faced by catheters in the future.

[0162] Step S44: construct a central venous catheter intelligent decision-making model based on the central venous catheter risk data to obtain a central venous catheter intelligent decision-making model.

[0163] In an embodiment of the present invention, this step involves inputting the central venous catheter risk data obtained in the previous step into the intelligent decision-making model for comprehensive judgment and formulation. The central venous catheter intelligent decision-making model is based on the input risk data, combined with multi-source data (such as patient signs, venous ultrasound images, hemodynamic data, etc.), and generates catheter use and intervention plans through multi-dimensional evaluation and decision-making algorithms. The construction of the intelligent decision-making model usually adopts machine learning or optimization algorithms, such as decision trees, support vector machines, random forests, genetic algorithms, reinforcement learning and other methods, and ensures the quality of input data through data preprocessing (such as standardization, missing value filling, etc.). Then, a decision tree model or a random forest model is constructed to predict catheter risks by training historical data, and intervention recommendations are given according to different risk levels. At the same time, the central venous catheter intelligent decision model is trained according to the training historical data and catheter risks, so that the central venous catheter intelligent decision model gives catheter specifications and models, and puncture site selection recommendations according to the patient's treatment plan.

[0164] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for constructing a central venous catheter intelligent decision-making model based on multi-source data, characterized in that: The following steps are involved: Step S1: Acquire central venous catheter case data; extract case patient vital signs based on the central venous catheter case data to obtain case patient vital signs data; analyze central venous catheter structure characteristics based on the central venous catheter case data to obtain central venous catheter structure characteristic data; Step S2: collecting venous ultrasound images of case patients according to the case patient's physical sign data to obtain venous ultrasound image data of case patients; estimating the specifications of central venous catheters based on the structural characteristic data of central venous catheters according to the case patient's venous ultrasound image data to obtain central venous catheter specifications data; calculating the probability of catheter position displacement according to the central venous catheter specifications data and the case patient's venous ultrasound image data to obtain central venous catheter displacement probability data; Step S3: collecting the central venous catheter mis-touch area according to the central venous catheter displacement probability data to obtain the central venous catheter mis-touch area data; analyzing the central venous catheter abnormal state according to the central venous catheter mis-touch area data and the venous ultrasound image data of the case patient to obtain the central venous catheter abnormal state data; Step S4: Perform a central venous catheter risk assessment based on the central venous catheter abnormal status data to obtain central venous catheter risk data; construct a central venous catheter intelligent decision-making model based on the central venous catheter risk data to obtain a central venous catheter intelligent decision-making model.

2. The method for constructing a central venous catheter intelligent decision-making model based on multi-source data according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: Acquire central venous catheter case data; Step S12: performing case desensitization processing according to the central venous catheter case data to obtain central venous catheter case desensitization data; Step S13: extracting the case patient's vital signs based on the central venous catheter case desensitization data to obtain the case patient's vital signs data; Step S14: performing a central venous catheter structural feature analysis based on the central venous catheter case desensitized data to obtain central venous catheter structural feature data.

3. The method for constructing a central venous catheter intelligent decision-making model based on multi-source data according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: collecting venous ultrasound images of the case patient according to the case patient's physical sign data to obtain venous ultrasound image data of the case patient; Step S22: performing vein feature analysis based on the venous ultrasound image data of the case patient to obtain ultrasound image vein feature data; Step S23: estimating the specifications of the central venous catheter based on the ultrasound image vein feature data and the central venous catheter structure feature data to obtain the central venous catheter specification data; Step S24: Calculate the probability of catheter position displacement according to the central venous catheter specification data and the ultrasound image vein feature data to obtain central venous catheter displacement probability data.

4. The method for constructing a central venous catheter intelligent decision-making model based on multi-source data according to claim 3 is characterized in that: Step S22 includes the following steps: Step S221: collecting image texture features according to the venous ultrasound image data of the case patient to obtain ultrasound image texture feature data; Step S222: performing venous blood vessel edge feature analysis based on the ultrasound image texture feature data and the venous ultrasound image data of the case patient to obtain venous blood vessel edge feature data; Step S223: performing contour feature analysis according to the venous blood vessel edge feature data and the ultrasonic image texture feature data to obtain venous blood vessel contour data; Step S224: Performing vein feature analysis based on the vein contour data and the vein edge feature data to obtain ultrasound image vein feature data.

5. The method for constructing a central venous catheter intelligent decision-making model based on multi-source data according to claim 4 is characterized in that: Step S222 includes the following steps: Calculate the grayscale value of the ultrasound image according to the venous ultrasound image data of the case patient, so as to obtain the grayscale value data of the venous ultrasound image; Calculate the global grayscale value standard deviation of the ultrasound image according to the grayscale value data of the venous ultrasound image to obtain the global grayscale value standard deviation data of the ultrasound image; Performing ultrasound image grayscale contrast analysis based on ultrasound image global grayscale value standard deviation data to obtain ultrasound image grayscale contrast data; Calculate the gradient change of texture feature direction according to the texture feature data of the ultrasonic image to obtain the gradient change data of the texture feature direction of the image; The venous blood vessel edge feature data is obtained by performing venous blood vessel edge feature analysis based on the image texture feature directional gradient change data and the ultrasound image gray value contrast data.

6. The method for constructing a central venous catheter intelligent decision-making model based on multi-source data according to claim 3, characterized in that: Step S23 includes the following steps: Step S231: performing image vein fine-grained analysis according to the ultrasound image vein feature data to obtain ultrasound image vein fine-grained data; Step S232: performing image vein wall thickness detection based on the ultrasound image vein fine-grained data to obtain ultrasound image vein wall thickness data; Step S233: performing image vein position measurement on the ultrasound image vein feature data, thereby obtaining ultrasound image vein position data; Step S234: performing image vein depth detection according to the ultrasound image vein position data to obtain ultrasound image vein depth data; Step S235: estimating the specifications of the central venous catheter based on the central venous catheter structural feature data based on the ultrasound image vein depth data to obtain the central venous catheter specification data.

7. The method for constructing a central venous catheter intelligent decision-making model based on multi-source data according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: collecting the central venous catheter mis-touch area according to the central venous catheter displacement probability data to obtain the central venous catheter mis-touch area data; Step S32: estimating the probability of local thrombosis according to the central venous catheter mis-touch area data and the venous ultrasound image data of the case patient, and obtaining ultrasound image local thrombosis probability data; Step S33: performing central venous catheter abnormal state analysis based on the local thrombosis probability data of the ultrasound image and the central venous catheter mis-touch area data to obtain central venous catheter abnormal state data.

8. The method for constructing a central venous catheter intelligent decision-making model based on multi-source data according to claim 7, characterized in that: Step S32 The following steps are involved: Step S321: collecting ultrasound image vein morphology change data based on the central venous catheter mis-touch area data and the case patient's venous ultrasound image data to obtain ultrasound image vein morphology change data; Step S322: performing venous puncture detection according to the ultrasound image venous morphology change data to obtain ultrasound image venous puncture data; Step S323: performing venous vessel injury detection according to the ultrasonic image venous vessel puncture data to obtain ultrasonic image venous vessel injury data; Step S324: performing vasospasm analysis based on the ultrasound image venous vascular injury data to obtain ultrasound image vasospasm probability data; Step S325: estimating venous blood flow obstruction according to the central venous catheter mis-touch area data and the ultrasound image venous vascular injury data to obtain venous blood flow obstruction data; Step S326: estimating the probability of local thrombosis based on the ultrasound image vascular spasm probability data and the venous blood flow obstruction data, and obtaining the ultrasound image local thrombosis probability data.

9. The method for constructing a central venous catheter intelligent decision-making model based on multi-source data according to claim 7, characterized in that: Step S33 includes the following steps: Step S331: performing venous thrombosis enlargement analysis based on the local thrombosis probability data of the ultrasound image to obtain venous thrombosis enlargement data; Step S332: performing venous vessel blockage detection based on the venous vessel thrombus enlargement data to obtain venous vessel blockage data; Step S333: performing central venous catheter function loss detection according to the venous vessel blockage data and the venous vessel thrombosis enlargement data to obtain central venous catheter function loss data; Step S334: estimating the probability of venous infection according to the local thrombosis probability data of the ultrasound image and the central venous catheter mis-touch area data to obtain venous infection probability data; Step S335: performing central venous catheter abnormal state analysis based on the venous infection probability data and the central venous catheter function loss data to obtain central venous catheter abnormal state data.

10. The method for constructing a central venous catheter intelligent decision-making model based on multi-source data according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: Perform venous function impairment detection based on the abnormal state data of the central venous catheter to obtain venous function impairment data; Step S42: Estimating venous catheter complications according to the venous function impairment data and the central venous catheter abnormal state data to obtain central venous catheter complication data; Step S43: performing a central venous catheter risk assessment based on the venous catheter complication data and the venous function impairment data to obtain central venous catheter risk data; Step S44: construct a central venous catheter intelligent decision-making model based on the central venous catheter risk data to obtain a central venous catheter intelligent decision-making model.

Images