Method, apparatus, medium and program product for predicting catheter-related thrombus risk based on tumor site

Abstract

The invention provides a method, equipment, medium and program product for predicting catheter-related thrombus risk based on a tumor site, and relates to the field of intelligent medical treatment. The method comprises the following steps: acquiring a tested tumor site; outputting an auxiliary prediction result of the CRT probability risk of the test according to different tumor parts; and if the tumor part is in the thoracic cavity, outputting an auxiliary prediction result of the CRT high risk of the subject, if the tumor part is in the abdominal cavity, outputting an auxiliary prediction result of the CRT risk of the subject, and if the tumor part is in the pelvic cavity, outputting an auxiliary prediction result of the CRT low risk of the subject. By collecting catheter insertion data of a tumor patient, importance analysis is carried out on a tumor part.

Description

Technical Field

[0001] The present invention relates to the field of intelligent medicine, and more specifically, to a method, device, medium, and program product for predicting the risk of catheter-related thrombosis based on the tumor location. Background Art

[0002] Cancer patients face various treatment needs and must have a stable venous access. Venous catheters are widely used due to their unique advantages, such as enabling safe multi-drug chemotherapy and intensive supportive treatment. Severe complications of venous catheters pose a considerable risk to clinical work, and one of the most common risks is catheter-related thrombosis (CRT). Considering the loss of venous access, the risk of pulmonary embolism, and additional costs, CRT is a major clinical problem.

[0003] Clinically, the determination method of CRT is mainly detected by means of clinical manifestations and signs, imaging examinations, etc. Considering the possible adverse consequences of CRT, how to identify and screen high-risk populations at an early stage has always been the focus of attention. Although the existing Khorana risk score is a commonly used risk score assessment for venous thromboembolism (VTE), it cannot accurately predict catheter-related thrombosis. Other judgment criteria require collecting more relevant factors of patients, while this solution aims to achieve the effect of earlier identification and screening of high-risk populations with fewer factors. Summary of the Invention

[0004] The present invention aims to at least solve one of the technical problems existing in the prior art. To this end, the present invention provides a method, device, medium, and program product for predicting the risk of catheter-related thrombosis based on the tumor location; by collecting catheter insertion data of tumor patients, importance analysis is performed on the tumor location.

[0005] The first aspect of this application discloses a method for predicting the risk of catheter-related thrombosis based on the tumor location, and the method includes:

[0006] S101, obtaining the tumor location of the subject;

[0007] S102, outputting an auxiliary prediction result of the probability risk of the subject having CRT according to different tumor locations; if the tumor location is in the chest cavity, output an auxiliary prediction result of high risk of the subject having CRT; if the tumor location is in the abdominal cavity, output an auxiliary prediction result of medium risk of the subject having CRT; if the tumor location is in the pelvic cavity, output an auxiliary prediction result of low risk of the subject having CRT.

[0008] In some embodiments, the method further includes: calculating the spatial distance from the tumor site to the superior vena cava where the catheter is located; outputting an auxiliary prediction result of the probability of CRT occurrence in the subject according to the spatial distance; if the spatial distance is less than a first threshold, outputting an auxiliary prediction result with a high probability of CRT occurrence in the subject; if the spatial distance is greater than the first threshold, outputting an auxiliary prediction result with a low probability of CRT occurrence in the subject.

[0009] In some embodiments, the order of CRT risks of the organs in the thoracic cavity from high to low is as follows: mediastinum, esophagus, lung / pleura, lymphoma.

[0010] In some embodiments, the order of CRT risks of the organs in the abdominal cavity from high to low is as follows: biliary tract, liver, pancreas, kidney / adrenal gland, stomach, peritoneum, intestine, rectum;

[0011] Optionally, the order of CRT risks in the pelvis from high to low is as follows: uterus / vagina, testis, bladder / ureter, ovary / fallopian tube, prostate.

[0012] In some embodiments, the tumor site further includes any one or more of the following: breast, head / neck, extremities, body wall.

[0013] In some embodiments, the order of CRT risks of the tumor site from high to low is as follows: mediastinum, esophagus, biliary tract, liver, breast, lung / pleura, pancreas, head / neck, kidney / adrenal gland, lymphoma, uterus / vagina, stomach, peritoneum, extremities, intestine, rectum, testis, bladder / ureter, ovary / fallopian tube, body wall, prostate.

[0014] In some embodiments, the method for determining the order of CRT risks of the tumor site from high to low includes:

[0015] Obtaining a training set sample including any one or more of the data of sample basic information, catheter type, systemic drug treatment, tumor site and the corresponding result information of the sample;

[0016] Processing the data of systemic drug treatment and tumor site by using the factor folding method to obtain the processed data features of systemic drug treatment and tumor site; converting the basic information and catheter type data into numerical features;

[0017] Performing sensitivity analysis by using the "model_profile" function to evaluate the change of the survival function value derived from the model with the change of the feature value; calculating the area and direction between the highest and lowest time survival function curves, determining the importance of the feature in predicting CRT according to the area and direction, and determining the sorting result.

[0018] The second aspect of the present application discloses a computer device, which includes: a memory and a processor; the memory is used to store a computer program; the processor executes the computer program to implement the steps of the above method.

[0019] The third aspect of the present application discloses a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the above method.

[0020] The fourth aspect of the present application discloses a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the steps of the above method.

[0021] The present application has the following beneficial effects:

[0022] 1. The present application innovatively discloses a method for predicting the CRT risk at the tumor site. By collecting multi-center large-scale catheter data of tumor patients and recording variables such as patient general information, catheter information, tumor status, systemic treatment, and blood markers as comprehensively as possible, an importance analysis of the tumor site is carried out, and it is found that the CRT risk can be predicted according to the tumor site, thus realizing the function of earlier identification and screening of high-risk populations.

[0023] 2. The present application innovatively uses multi-center large-scale catheter data to develop a machine learning model (SM-CRT) for predicting the general risk and risk time distribution of CRT. SM-CRT has 4 application scenarios, which are applied to guide the selection of catheter types, stratify the thrombosis risk of CRT prevention, assist in making decisions on catheter placement timing (PDF is used to determine the high-risk period and low-risk period of CRT at the patient level; on this basis, PDF can guide the selection of catheter removal time to avoid the most dangerous period), and optimize the ultrasound follow-up period, ultimately improving the patient prognosis. The performance and application of SM-CRT have been well verified in both the training and test data sets. Description of the Drawings

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those skilled in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0025] Figure 1 It is a schematic flowchart of the method provided in the first aspect of the embodiments of the present invention;

[0026] Figure 2 It is a schematic diagram of the computer device provided in the embodiments of the present invention;

[0027] Figure 3 It is a schematic diagram of the architecture of an exemplary computing device provided by an embodiment of the present invention;

[0028] Figure 4 It is a schematic diagram of a storage medium provided by an embodiment of the present invention;

[0029] Figure 5 It is a model interpretation of SM-CRT provided by an embodiment of the present invention. The direction-marked area between the highest and lowest time survival function curves in the sensitivity analysis is represented by a histogram. The larger the number, the higher the importance of predicting CRT. These functions are arranged by category and color-coded;

[0030] Figure 6 It is a functional chart showing the application of SM-CRT provided by an embodiment of the present invention. Figure 6 a, A histogram showing the selection of catheter types by comparing the risk predictions of different catheter types. Figure 6 b, The Kaplan-Meier curve shows the risk identification based on the risk stratification risk prediction of SM-CRT. Figure 6 c, The probability density function (PDF) determined by SM-CRT, guiding the catheter time by providing the high CRT risk period (between t1 and t2). Figure 6 d, The cumulative distribution function (CDF) determined by SM-CRT to optimize the ultrasound follow-up period, and the recommended follow-up days are marked as red dots;

[0031] Figure 7 It is provided by an embodiment of the present invention that the risk period stratification based on Distr prediction demonstrates the ability of SM-CRT to optimize the extubation time decision. Figure 7 a, Schematic diagram of patient risk period stratification based on distribution prediction. Patient characteristics are input into the model to predict the instantaneous CRT probability over time. Then, a smoothed density estimate is performed on the Gaussian kernel and the distribution with a bandwidth of 3. The instantaneous probability greater than the highest 50% is the high-risk period, and the first high-risk day (t1), the day with the highest risk, and the first day after the high-risk period (t2) are marked with red dots. The actual catheter placement days are marked with a colored solid line. If the actual extubation day of the patient occurs before t1, the patient is assigned to the low-risk group; if the extubation occurs on or after t1 but before t2, the patient is assigned to the high-risk group; if the extubation occurs at or after t2, the patient is assigned to the long-term group. Figure 7 b, Comparing the number of CRT events per day between different groups predicted by SMCRT based on distribution through Fisher's exact test on the training and test data sets. The data is represented as a histogram. The P-values for multiple comparisons between groups are adjusted by the Benjamini-Hochberg (BH) procedure to control the false discovery rate (FDR). These groups are color-coded. Figure 7c, Comparing the number of daily CRT events between different SM-CRT risks and groups predicted based on distr by Fisher's exact test on training and test datasets. Data are represented as histograms. The groups are color-coded;

[0032] Figure 8 is the computational performance of the trained catheter-related thrombosis (CRT) survival model provided by an embodiment of the present invention. Figure 8 a, Concordance index (c-index) of the model for external ten-fold nested cross-validation (left bar chart) and based on three external test datasets (right heat map). The data on the left are represented as box plots, where the median line is the median, the left and right hinges correspond to the first and third quartiles respectively, and the whiskers correspond to the minimum or maximum value within 1.5 × IQR of the hinge (where IQR is the interquartile range). The median is also color-coded. Any data beyond the whiskers are considered outliers. The data on the right are represented as heat maps, where the c-index is color-coded and labeled. Figure 8 b, Time-dependent c-index of SM-CRT based on three external test datasets. Advantage = prospective test cohort; Ext1 = external test 1; Ext2 = external test 2;

[0033] Figure 9 is a forest plot showing the univariate Cox analysis hazard ratios for CRT-free survival provided by an embodiment of the present invention. HR = hazard ratio; CI = confidence interval; CRT = catheter-related thrombosis; BMI = body mass index; KPS = Karnofsky performance score; ALB = albumin; Hb = hemoglobin; PLT = platelets; WBC = white blood cells; CVC = central venous catheter; FICC = femoral inserted central catheter; PICC = peripherally inserted central venous catheter; PORT = implanted venous port; Anti.HER.mAb = human epidermal growth factor receptor monoclonal antibody; Anti.CD20.mAb = anti-CD20 monoclonal antibody. Detailed implementation manners

[0034] To enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0035] In some processes described in the specification, claims, and above-mentioned drawings of the present invention, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations can be performed not in the order in which they appear herein or in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations can be performed sequentially or in parallel. It should be noted that the descriptions such as "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit that "first" and "second" are of different types.

[0036] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work belong to the scope of protection of the present invention.

[0037] Figure 1 It is a schematic flowchart of a method for predicting the risk of catheter-related thrombosis based on the tumor location provided by an embodiment of the present invention. Specifically, the method includes the following steps:

[0038] S101, obtaining the tumor location of the subject;

[0039] S102, outputting an auxiliary prediction result of the probability risk of the subject having CRT according to different tumor locations; if the tumor location is in the chest cavity, outputting an auxiliary prediction result of the subject having a high risk of CRT; if the tumor location is in the abdominal cavity, outputting an auxiliary prediction result of the subject having a medium risk of CRT; if the tumor location is in the pelvic cavity, outputting an auxiliary prediction result of the subject having a low risk of CRT.

[0040] In some embodiments, the method further includes: calculating the spatial distance from the tumor location to the superior vena cava where the catheter is located; outputting an auxiliary prediction result of the probability of the subject having CRT according to the spatial distance; if the spatial distance is less than the first threshold, outputting an auxiliary prediction result of a high probability of the subject having CRT; if the spatial distance is greater than the first threshold, outputting an auxiliary prediction result of a low probability of the subject having CRT.

[0041] In some embodiments, the order of the risk of CRT in the organs in the chest cavity from high to low is as follows: mediastinum, esophagus, lung / pleura, lymphoma. The organs in the chest cavity generally include: heart, lungs, trachea, bronchi (bronchi are branches of the trachea responsible for delivering air to various parts of the lungs), thymus, esophagus, large blood vessels (such as aorta, pulmonary artery, pulmonary vein, and superior vena cava), lymph nodes.

[0042] In some embodiments, the order of CRT risks of the organs in the abdominal cavity from high to low is as follows: biliary tract, liver, pancreas, kidney / adrenal gland, stomach, peritoneum, intestine, rectum. The organs in the abdominal cavity generally include: stomach (the main organ for digesting food), small intestine (including duodenum, jejunum, and ileum, which is the main site for digestion and absorption), large intestine (cecum, appendix, colon, rectum, and anus, mainly responsible for water absorption and waste excretion), liver (the largest digestive gland, participating in metabolism, detoxification, and storage of nutrients), gallbladder (storing and concentrating bile), pancreas (secreting digestive enzymes and hormones such as insulin), spleen (participating in blood filtration and immune response), kidney (located at the rear of the abdominal cavity, responsible for filtering blood and forming urine), adrenal gland (located above the kidney, secreting hormones such as adrenaline and cortisol).

[0043] Optionally, the order of CRT risks of the pelvic cavity from high to low is as follows: uterus / vagina, testis, bladder / ureter, ovary / fallopian tube, prostate. The organs in the pelvic cavity (the pelvic cavity is located below the abdominal cavity) generally include: bladder (storing urine), rectum (a part of the large intestine, located in the pelvic cavity, responsible for storing and excreting feces), reproductive organs (female: ovary, fallopian tube, uterus, and vagina; male: testis, epididymis, vas deferens, prostate, and a part of the penis), pelvic bones (including hip bone, sacrum, and coccyx).

[0044] In some embodiments, the tumor site further includes any one or more of the following: breast, head / neck, limbs, body wall. The body wall does not belong to one of the body cavities in the thoracic cavity, abdominal cavity, or pelvic cavity, but constitutes the boundary of these body cavities. The body wall is a structure composed of bones, muscles, connective tissues, and skin, which surrounds and protects the organs in the thoracic cavity, abdominal cavity, and pelvic cavity. The body wall includes the following parts: Chest wall: surrounding and protecting the organs in the thoracic cavity, such as the heart and lungs. The chest wall is mainly composed of the sternum, ribs, and the thoracic part of the spine in the back, as well as the muscles and skin covering them. Abdominal wall: surrounding and protecting the organs in the abdominal cavity, such as the stomach, liver, spleen, kidneys, etc. The abdominal wall is composed of the muscles and skin of the abdomen, as well as the lumbar part of the spine. Pelvic wall: constituting the bottom and sides of the pelvic cavity, protecting the organs in the pelvic cavity, such as the bladder, rectum, and reproductive organs. The pelvic wall is composed of the hip bone, sacrum, and coccyx, as well as the related muscles and skin.

[0045] In some embodiments, the order of CRT risks of the tumor site from high to low is as follows: mediastinum, esophagus, biliary tract, liver, breast, lung / pleura, pancreas, head / neck, kidney / adrenal gland, lymphoma, uterus / vagina, stomach, peritoneum, limbs, intestine, rectum, testis, bladder / ureter, ovary / fallopian tube, body wall, prostate.

[0046] In some embodiments, the method for determining the sorting result of the risk of CRT occurring at the tumor site from high to low includes: obtaining a training set sample including any one or several of the sample basic information, catheter type, systemic drug treatment, and tumor site data and the corresponding result information of the sample; processing the systemic drug treatment and tumor site data by the factor folding method to obtain the processed systemic drug treatment and tumor site data features; converting the basic information and catheter type data into numerical features; performing sensitivity analysis by using the "model_profile" function to evaluate the change of the survival function value derived from the model with the change of the feature value; calculating the area and direction between the highest and lowest time survival function curves, and determining the importance of the feature predicting CRT according to the area and direction to determine the sorting result.

[0047] In some embodiments, the result that the spatial distance from the tumor site to the superior vena cava where the catheter is located is an important factor for the probability of CRT occurrence is determined by the SM-CRT model: The SM-CRT model is a survival model that provides a risk score (crank, used to evaluate the general risk over time and calculate the c-index) and predicts the survival distribution (distr, predicts the survival distribution, used to guide the catheter and ultrasound follow-up time). The construction method of the SM-CRT model includes:

[0048] obtaining a training set sample including any one or several of the sample basic information, catheter type, systemic drug treatment, and tumor site data and the corresponding result information of the sample; processing the systemic drug treatment and tumor site data by the factor folding method to obtain the processed systemic drug treatment and tumor site data features (converting the basic information and catheter type data into numerical features; using a variety of feature engineering methods to build features here, applying the factor folding method to reduce the number of sparse features; performing one-hot encoding on the categorical variables as the input of the model); inputting the data features, numerical features, and result information into a machine learning model for training to obtain the SM-CRT model.

[0049] In some more specific embodiments, the result information includes: CRT event, event time, catheter insertion duration; wherein, the CRT event is defined as venous thrombosis along the catheter; the time from catheter insertion to the occurrence of CRT is considered the CRT-free survival time (CFS), the catheter removal without CRT and the catheter placement without CRT until the last ultrasound follow-up are considered the censored time points;

[0050] In some more specific embodiments, the sample basic information includes any one or several of the following: age, gender, body mass index, KPS, smoking status, drinking status, previous or concurrent diseases;

[0051] In some more specific embodiments, the previous or concurrent diseases include any one or more of the following: hypertension, diabetes, coronary heart disease, cerebral infarction, hyperlipidemia, history of venous thrombosis;

[0052] In some more specific embodiments, the catheter types include any one or more of the following: FICC, PICC, CVC, PORT;

[0053] In some more specific embodiments, the systemic drug treatment is recorded during CRT occurrence or catheter removal, mainly including any one or more of the following: chemotherapy (platinum-based, microtubule inhibitors, antimetabolites, antitumor antibiotics, nitrogen mustard, DNA topoisomerase inhibitors), targeted therapy (immunotherapy, anti-angiogenic therapy, Anti.HER.mAb, Anti.CD20.mAb);

[0054] In some more specific embodiments, the tumor sites include any one or more of the following (rare tumor sites are grouped with the nearest site, and tumors are divided into 21 groups): lung / pleura, esophagus, mediastinum, stomach, intestine, pancreas, liver, biliary tract, kidney / adrenal gland, peritoneum, rectum, bladder / ureter, uterus / vagina, prostate, ovary / fallopian tube, testis, breast, head / neck, trunk wall, extremities, and lymphoma;

[0055] Optionally, the data further includes experimental test data (blood markers), tumor stage; the experimental test data (blood markers) includes any one or more of the following: albumin, hemoglobin, platelets, white blood cells, D-dimer; the tumor stage includes I, II, III, IV; optionally, the data further includes whether the catheter tip position is appropriate.

[0056] In some embodiments, the method further includes: calculating the importance of data features and numerical features in S102; the calculation method of the importance includes: performing sensitivity analysis by using the "model_profile" function to evaluate the change of the survival function value derived from the model with the change of the feature value; calculating the area and direction between the highest and lowest time survival function curves, and determining the importance of the feature in predicting CRT according to the area and direction;

[0057] In some embodiments, the methods used by the machine learning model include any one or more of the following: rfsrc, ranger, cforest, coxph, xgboost, penalized, coxboost; preferably coxph.

[0058] In some embodiments, the SM-CRT model has the following 4 application scenarios:

[0059] Application scenario 1 is to identify the risk of CRT based on the SM-CRT model. Specifically, the identification method includes the following steps:

[0060] Obtain data of the subject including basic information, catheter type used, systemic drug treatment, and tumor location; input the data into the SM-CRT model described in the first aspect of the present application to calculate the risk value; output an auxiliary prediction result of the high or low risk of CRT for the subject according to the risk value; specifically, when the risk value is greater than the first threshold, output an auxiliary prediction result of high risk of CRT for the subject, and when the risk value is less than the first threshold, output an auxiliary prediction result of low risk of CRT for the subject;

[0061] Optionally, the method for determining the first threshold includes, but is not limited to, selecting the median of the CRT model prediction risk ranking scores of each data set; customizing the cut-off value for a specific actual application scenario.

[0062] Application scenario 2 is to guide the selection of catheter type based on the SM-CRT model. Specifically, the guiding method includes the following steps:

[0063] Obtain data of the subject's basic information, systemic drug treatment, tumor location, and the expected catheter type to be used; input the data of the basic information, systemic drug treatment, tumor location, and the expected catheter type into the SM-CRT model described in the first aspect of the present application, and output the predicted risk values of CRT for using different catheter types; determine the finally used catheter type according to the predicted risk values; Experimental verification: Compared with CVC and PORT, FICC and PICC have a higher risk of CRT.

[0064] Application scenario 3 is to guide the catheter removal time based on the SM-CRT model. Specifically, the guiding method includes the following steps:

[0065] Obtain data of the subject including basic information, catheter type used, systemic drug treatment, and tumor location; input the data into the SM-CRT model described in the first aspect of the present application to predict the high CRT risk period, and predict the catheter removal time of the subject according to the high CRT risk period;

[0066] Optionally, the catheter removal time can be a specific time point or a time interval;

[0067] Optionally, the high CRT risk period is between the first high-risk day and the second high-risk day;

[0068] Optionally, the first high-risk day and the second high-risk day are determined by calculating the instantaneous probability. The determination method includes: calculating the time distribution of the probability density function (PDF) of the training set samples; calculating the kernel density estimation of the time distribution with a Gaussian kernel and a bandwidth of N; calculating the instantaneous probability varying with time based on the kernel density estimation, and calculating the highest instantaneous probability in the previous M time (6 months); recording the days with the instantaneous probability greater than the first percentage (50%) of the highest instantaneous probability as the high-risk period; recording the first day of the determined high-risk period as the first high-risk day t1, and recording the first day after the determined high-risk period as the second high-risk day t2;

[0069] Optionally, the high-risk period further includes the day with the highest risk, and the time corresponding to the highest risk is the day with the highest risk;

[0070] Furthermore, the method further includes: calculating the risk value of the subject based on the SM-CRT model, and predicting the extubation time of the subject according to the risk value and the high CRT risk period; specifically, further grouping the patients by combining risk ranking prediction and distribution prediction, and by selecting the median risk ranking value based on the data set as the cut-off value, dividing the patients into six groups according to different combinations of the risk ranking score (high risk ranking value and low risk ranking value) and the distribution period (high risk, low risk, and long term) (high risk ranking + high risk, high risk ranking + long term, high risk ranking + low risk, low risk ranking + high risk, low risk ranking + long term, low risk ranking + low risk; low risk ranking + high risk and low risk ranking + low risk are similar and both are lower than high risk ranking + high risk).

[0071] Application scenario 4 is to guide the ultrasound follow-up cycle based on the SM-CRT model. Specifically, the guiding method includes the following steps:

[0072] Obtain the data of the subject including basic information, catheter type used, systemic drug treatment, tumor location, and the first time; input the data into the SM-CRT model described in the first aspect of the present application to calculate the cumulative distribution function (CDF) at different times; if the CDF at the second time increases by the second threshold compared to the first time, then the second time is the predicted ultrasound follow-up time adjacent to the first time; the ultrasound follow-up time can be a specific time point; the first time is the time of the last ultrasound examination;

[0073] In some embodiments, the term "subject" or "test subject" or "test sample" used herein refers to any animal (e.g., mammal), including but not limited to humans, non-human primates, rodents, etc., which will be the recipient of a specific treatment. Generally, the terms "subject" and "patient" can be used interchangeably herein when referring to human subjects. Preferably, the subject is a human. In some embodiments, the test sample is a patient clinically used for prognostic evaluation.

[0074] In some embodiments, the auxiliary prediction results include, but are not limited to, paper or electronic reports. These results are only obtained by the intelligent machine through analyzing relevant data of the subjects and are only for reference by medical staff, not the final diagnosis results of the subjects.

[0075] In some embodiments, the first threshold or the second threshold is obtained by training with training set samples or according to prior knowledge. It can be a specific threshold or an interval range, and the specific form is not specifically limited in this embodiment. Letters such as N and M that appear in this embodiment are all rational numbers.

[0076] Figure 2 is a schematic diagram of a computer device provided by an embodiment of the present invention, as Figure 2 shown, the device 2000 may include: one or more processors 2010, and one or more memories 2020; wherein, computer-readable code is stored in the memory, and when the computer-readable code is run by the one or more processors, the method described above can be executed.

[0077] The processor in this embodiment may be an integrated circuit chip with signal processing capabilities. The above-mentioned processor may be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, operations and logic block diagrams disclosed in the embodiments of the present disclosure. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc., and may be of the X86 architecture or the ARM architecture.

[0078] Generally speaking, the various example embodiments of the present disclosure can be implemented in hardware or dedicated circuits, software, firmware, logic, or any combination thereof. Some aspects can be implemented in hardware, while other aspects can be implemented in firmware or software that can be executed by a controller, a microprocessor or other computing devices. When aspects of the embodiments of the present disclosure are illustrated or described as block diagrams, flowcharts or using some other graphical representation, it will be understood that the blocks, devices, systems, technologies or methods described herein can be implemented as non-limiting examples in hardware, software, firmware, dedicated circuits or logic, general hardware or controllers or other computing devices, or some combination thereof.

[0079] For example, the method or device according to the embodiments of the present disclosure can also be implemented by means of Figure 3 the architecture of the computing device 3000 shown. As Figure 3As shown, the computing device 3000 may include a bus 3010, one or more CPUs 3020, a read-only memory (ROM) 3030, a random access memory (RAM) 3040, a communication port 3050 connected to a network, an input / output component 3060, a hard disk 3070, etc. Storage devices in the computing device 3000, such as the ROM 3030 or the hard disk 3070, may store various data or files used for the processing and / or communication of the methods provided by this disclosure, as well as program instructions executed by the CPU. The computing device 3000 may also include a user interface 3080. Of course, Figure 3 the architecture shown is merely exemplary, and when implementing different devices, one or more components in the computing device shown may be omitted according to actual needs. Figure 3

[0080] An embodiment of the present invention also provides a computer-readable storage medium, such as Figure 4 shown, which is a schematic diagram of the storage medium 4000 provided by an embodiment of the present invention. Computer-readable instructions 4010 are stored on the computer storage medium 4020. When the computer-readable instructions 4010 are run by a processor, the methods according to the embodiments of this disclosure described with reference to the above figures may be executed. The computer-readable storage medium in the embodiments of this disclosure may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. The non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct rambus random access memory (DR RAM). It should be noted that the memories for the methods described herein are intended to include but not be limited to these and any other suitable types of memories. It should be noted that the memories for the methods described herein are intended to include but not be limited to these and any other suitable types of memories.

[0081] An embodiment of the present disclosure also provides a computer program product or system, including a computer program, which implements the steps of the above method when executed by a processor.

[0082] In some embodiments, the present embodiment also discloses a system for predicting the risk of CRT based on the tumor location. The system includes: an acquisition module, configured to acquire the tumor location of the subject; a result output module, configured to output an auxiliary prediction result of the probability risk of the subject having CRT according to different tumor locations; if the tumor location is in the chest cavity, output an auxiliary prediction result of the high risk of the subject having CRT; if the tumor location is in the abdominal cavity, output an auxiliary prediction result of the medium risk of the subject having CRT; if the tumor location is in the pelvic cavity, output an auxiliary prediction result of the low risk of the subject having CRT. Specific embodiment method: Research design and population: Tumor patients who underwent catheterization, met the inclusion criteria and were treated at four centers were recruited. Data were collected prospectively and analyzed retrospectively. The study included 1 training group and 3 test groups. Patients recruited from Center 1 and Center 2 were combined for multicenter training and testing. Specifically, patients whose data were collected before June 30, 2023 were included in the training cohort, and patients whose data were collected after July 1, 2023 were included in the prospective test cohort. Patients recruited from Center 3 and Center 4 were included in a separate external test set.

[0085] Inclusion and exclusion criteria: Eligible patients were pathologically diagnosed with malignant tumors; successful catheterization was performed; voluntarily participated in this study and reported their data; age ≥ 18 years old; extubation or non-extubation under ultrasound but at least one vascular ultrasound examination was performed after catheterization. Key exclusion criteria included incomplete patient information, unknown primary tumor location, and no follow-up (patients who did not record ultrasound data or did not attend follow-up appointments were excluded). It should be noted that if a patient had multiple catheters, each case was treated differently in our study.

[0086] Catheter insertion method: The modified Seldinger technique under ultrasound guidance was used for the placement of central venous catheters (CVCs), femoral artery inserted central catheters (FICCs), peripherally inserted central catheters (PICCs), and implanted venous ports (PORTs). All CVCs and PORTs were non-tunneled catheters inserted through the subclavian vein or jugular vein on the healthy side (without tumors such as breast cancer in the vicinity), all PICCs were inserted into the superior vena cava through the basilic vein on the healthy side, and all FICCs were inserted into the superior vena cava through the basilic vein on the healthy side. Based on the CVC, it was inserted into the inferior vena cava through the femoral vein on the uninjured side. FICC was mainly applied to 46 patients with superior vena cava syndrome (SVCS), which could cause edema in the head, neck, and upper extremities, resulting in hemodynamic stasis and thrombosis 47. The materials of CVC, FICC, and PORT were polyurethane, and the material of PICC was silicone rubber. The length of CVC, PORT, and FICC in the vein was about 18 cm, and the length of PICC in the vein was about 40 cm. The outer diameter (OD) of the catheter was as follows: 4Fr for Center 1 and 2; 4Fr for the PICC at Center 3; 6Fr for the PORT going to Center 3; 6Fr at Center 4. After non-femoral artery catheterization, chest X-ray examination (including the upper extremities and neck on the same side of the catheter) was performed to confirm the direction of the catheter and the position of the catheter tip. All X-ray films were evaluated by doctors specializing in venous catheter insertion, and at least one radiologist was responsible for the chest X-ray report. For non-femoral artery catheters, the catheter tip was usually located in the lower third of the superior vena cava, the cavoatrial junction, or the upper third of the right atrium. If the vertebra was used as a reference, the 6th - 8th thoracic vertebrae (T6 - T8) were considered the appropriate position; otherwise, the position was considered inappropriate. For FICC, the catheter tip was placed at the origin of the inferior vena cava near the common iliac vein, regardless of the inappropriate position.

[0087] Data collection: General patient information, past or concomitant diseases, tumor status, catheter-related information, and baseline laboratory data were recorded when the catheter was inserted. When CRT occurred or during catheter removal, the drugs used during catheter insertion and outcome information were recorded, including CRT events, event time, and catheter insertion duration. CRT events were defined as venous thrombosis along the catheter. Variable details:

[0088] Baseline variables recorded at the first catheter insertion included 5 categories and 47 features, and these variables included the following: 1) General information: age, gender, body mass index (BMI), Karnofsky Performance Status (KPS), smoking status, and alcohol consumption status. Previous or co-existing diseases: hypertension, diabetes, coronary heart disease, cerebral infarction, hyperlipidemia, and history of venous thrombosis. 2) Tumor status: tumor type and stage (I, II, III, and IV). For tumor type, tumors were grouped by location. In particular, rare tumor sites were grouped with the nearest sites. Tumors were divided into 21 groups: lung / pleura, esophagus, mediastinum, stomach, intestine, pancreas, liver, biliary tract, kidney / adrenal gland, peritoneum, rectum, bladder / ureter, uterus / vagina, prostate, ovary / fallopian tube, testis, breast, head / neck, body wall, extremities, and lymphoma. 3) Catheter-related information: catheter type (PICC, CVC, PORT, or FICC) and malposition of the catheter tip. 4) Baseline laboratory data: albumin (ALB), hemoglobin (Hb), platelets (PLT), white blood cells (WBC), and D-dimer.

[0089] A class of 11 features and outcome information was recorded during CRT occurrence or catheter removal, and these variables included the following: 1) Drugs used during catheter insertion: chemotherapy (platinum-based, microtubule inhibitors, antimetabolites, antitumor antibiotics, nitrogen mustards, DNA topoisomerase inhibitors), targeted therapy (immunotherapy, anti-angiogenic therapy, Anti.HER.mAb, Anti.CD20.mAb), and nutrition obtained from medical records, whether provided through the catheter or not. 2) Outcome information: CRT event, event time, and catheter insertion duration. The CRT event was defined as venous thrombosis along the catheter.

[0090] Multiple feature engineering methods were used to prune features. The factor folding method was applied to tumor type and chemotherapy type to reduce the number of sparse features. For example, "lung" and "pleura" were combined into "lung / pleura". One-hot encoding was performed on categorical variables as input to the model. Since the sample size of our dataset was large and the number of features was relatively small (58 in total), collinearity removal and variable selection were not performed to avoid potential loss of model performance.

[0091] CRT evaluation: The patients were continuously examined until extubation as required by the doctor. The diagnosis of CRT was performed by vascular ultrasound Doppler and color imaging (GE LOGIQ T M E9). Ultrasonography was performed every 3 months after catheter insertion, during extubation, and when any clinical symptoms suggesting CRT were detected. The occurrence of CRT was regarded as the target event in the survival analysis, and the time from catheter insertion to the occurrence of CRT was considered the CRT-free survival (CFS) time. Extubation without CRT and catheter placement without CRT until the last ultrasound follow-up were considered censored time points.

[0092] Survival model construction and validation: The model was trained using the training set. To select a model with appropriate hyperparameters and evaluate the model performance, a nested ten-fold cross-validation strategy was applied. The inner ten-fold cross-validation was used for hyperparameter tuning, and the outer ten-fold cross-validation was used for internal validation. Twenty-three basic models and twenty-one ensemble models (Details: Twenty-three basic models, including'surv.akritas','surv.blackboost','surv.cforest','surv.coxboost','surv.coxph','surv.coxtime','surv.ctree','surv.cv_coxboost','surv.cv_glmnet','surv.deephit','surv.gbm','surv.glmboost','surv.glmnet','surv.kaplan','surv.loghaz','surv.nelson','surv.pchazard','surv.penalized','surv.ranger','surv.rfsrc','surv.rpart', 'Surv.xgboost.aft' and 'Surv.xgboost.cox' were developed. In addition, six ensemble models based on the top 2 to top 7 best-performing models (in the outer cross-validation), named surv.ens_topN (e.g., surv.ens_top2), and twenty-one ensemble models based on all models were developed. Possible pairwise combinations between the top 7 best-performing models (in the outer cross-validation), named model1_ens_model2 (e.g., ranger_ens_coxph).). These models were trained to predict the risk score (crank) and the predicted survival distribution (distr) prediction types. Crank is the risk score, used to evaluate the general risk over time and calculate the c-index, and distr is the predicted survival distribution, used to guide catheter and ultrasound follow-up times. The default weighted discrete distribution in the mlr3probaR package was applied to this model. The concordance index (c-index) was used as the evaluation metric to compare the performance of different models in the outer ten-fold cross-validation. The c-index ranges from 0 to 1, a value of 0.5 indicates no discrimination, equivalent to random chance, while a value of 1.0 indicates perfect discrimination. The Brier score was used for calibration comparison. The Brier score ranges from 0 to 1, and a lower value indicates higher prediction accuracy. Then the model was trained using all the training data and tested using the test set.Model Explanation: Although the Cox model can be easily interpreted based on the feature coefficients; however, due to the scale differences between features (e.g., platelet count is a continuous variable from 0 - 1000, while gender is a binary variable with values 0 or 1), the coefficients may not accurately reflect the feature importance and thus which features contribute to the model prediction results. Therefore, the importance of each feature is determined by calculating the area between the highest and lowest time survival function curves in the sensitivity analysis, and the direction of importance is considered. The "survex" package 48 in R is used for survival model interpretation. The "model_profile" function evaluates the change in the survival function values derived from the model as the feature values change, and is used for sensitivity analysis. Then the area between the highest and lowest time survival function curves is calculated to evaluate the importance of each feature. If the highest feature value is associated with the lowest curve, the feature is considered to have a risk effect (the area is marked as positive); otherwise, it is considered to have a protective effect (the area is marked as negative). The larger the number, the higher the importance of predicting CRT.

[0093] Statistical Analysis: Kaplan-Meier survival curves were generated to estimate the event time distribution. The "muhaz" package and the "muhaz" function were used to perform a smoothed estimation of the hazard function for right-censored data using global and local bandwidth selection algorithms and boundary kernel functions, with other parameters set to default values.

[0094] Univariate Cox proportional hazards regression analysis was used to evaluate the association between variables and CFS. The Wald test was used to evaluate the significance of each variable in the Cox regression model. The hazard ratio (HR) and 95% confidence interval (CI) for each variable were calculated. An HR value greater than 1 indicates an increased risk of CRT, while an HR value less than 1 indicates a protective effect. The log2HR of different tumor sites in different cavities was compared by the Wilcoxon rank sum test. Specifically, prostate tumor patients did not experience CRT events and HR could not be calculated, so we designated -2.5 as the log2HR value for prostate tumors in this analysis. The P-values for multiple comparisons between different cavities were adjusted by the Benjamini-Hochberg (BH) procedure to control the false discovery rate (FDR).

[0095] The c-indexes obtained from ten-fold cross-validation of different models were compared by the Wilcoxon rank test. The median of the SM-CRT prediction risk values for each dataset was selected as the cut-off value to divide patients into high-risk or low-risk groups so that each group contained approximately equal numbers. The Kaplan-Meier method was used to analyze the event time data to estimate the survival probability, and the log-rank test was used for between-group comparison. A Cox proportional hazards regression model estimating the hazard ratio (HR) was used to quantify the impact of risk-based stratification on CRT outcomes.

[0096] The Fisher's exact test was used to compare the differences between the risk-period stratified groups, and the p-values were adjusted by the BH procedure. To address the potential confounding problem caused by the imbalance of variables between groups in risk-period stratification, we adopted propensity score matching (PSM). PSM was performed by estimating the propensity scores using the risk prediction of the model. Matching was performed using nearest neighbor matching (1:1 ratio) with a caliper of 0.01 without replacement to minimize the imbalance. All tests were two-sided. Unless otherwise stated, a p-value < 0.05 was considered to indicate statistical significance. All statistical analyses were performed using version 4.3 of the R statistical software.

[0097] Results:

[0098] Patient characteristics and univariate analysis: From January 1, 2017 to June 30, 2023, we collected 25,899 patients in the training cohort. A total of 1,031 patients (4%) who received CRT were recorded. According to the Kaplan-Meier survival curve, over time, the probability of non-CRT survival decreased from 1 to 0.786. The survival curve indicated that the hazard rate increased sharply before the 11th day after catheterization and then decreased to near 0 two years after catheterization, highlighting the time-dependent risk distribution of CRT events.

[0099] A univariate proportional hazards regression model was used to examine the potential risk factors associated with CRT ( Figure 9 ). Notably, the mediastinum and esophagus are closer to the superior vena cava than other sites and have the highest HR values ( Figure 9 ). This prompted us to speculate whether the spatial distance from the tumor to the superior vena cava is related to CRT. We found that thoracic tumors had higher HR values compared to abdominal tumors (median 2.539 vs. 0.912, p = 0.017) and pelvic tumors (median 2.539 vs. 0.46, p = 0.024). In addition, abdominal tumors had higher HR values compared to pelvic tumors (median 0.912 vs. 0.46, p = 0.0023) (the testis was considered to be located in the pelvis). These data suggest that the tumor location and the distance from the tumor to the superior vena cava where the catheter is located are important predictors of CRT.

[0100] The survival model can accurately predict CRT events: A survival model integrating the above variables was established to predict CRT events (SM-CRT). The results showed that surv.rfsrc, surv.ranger, surv.cforest, surv.coxph, surv.xgboost.aft, surv.penalized, and surv.coxboost had the highest c-index values in the external nested ten-fold base model cross-validation ( Figure 8a), and these high-performance base models were selected for ensemble learning. Then, data from all patients in the training cohort were used to train the final model for testing with 3 test datasets. The surv.coxph model had the highest c-index among the base models, and xgboost.aft_ens_penalized had the highest c-index among the ensemble models with the prospective test dataset; however, xgboost.aft_ens_penalized was only slightly higher than surv.coxph (0.702 vs. 0.697). Considering the lower complexity and higher interpretability of surv.coxph, it was selected as the final SM-CRT model, and the c-indexes of this model with the two external test datasets were 0.703 and 0.7, respectively, confirming its good performance. The Brier score was used for calibration comparison, and the results showed that surv.coxph had a lower Brier score in all test datasets. Overall, SM-CRT showed high performance in CRT prediction. This made it possible to select the catheter type for a given patient by comparing alternative types of risk prediction.

[0101] To further evaluate the performance of the surv.coxph model over time, we calculated the time-dependent c-index for each model, and the results showed that SM-CRT could achieve generally high c-indexes in the early stage of catheterization, even reaching 0.8 in all three test datasets ( Figure 8 b). Then, the c-index gradually decreased over time until reaching the overall c-index level ( Figure 8 b), indicating that prediction became more difficult over time. Then, the median of the SM-CRT predicted risk scores for each dataset was selected as the cutoff value for classifying patients into high-risk or low-risk groups. In the prospective test (HR = 1.86, p = 0.0058) and external test 1 (HR = 31.39, p < 0.001), patients with higher SM-CRT risk scores had significantly higher risks of receiving CRT than those with lower SM-CRT risk scores. And in external test 2 with a small sample size, the HR was slightly higher (HR = 1.94, p = 0.051). These results indicated that SMCRT could accurately identify high-risk CRT cases and low-risk cases.

[0102] The distribution prediction of CRT can guide the extubation time: The risk prediction of the SM-CRT model can be used to predict the general risk of patients and guide the selection of catheter types. However, the risk prediction results cannot be used to determine the extubation time of patients, which is another actionable factor and the most effective method for preventing CRT. Here, we explored whether the SM-CRT model can be used to guide the extubation time of patients based on the "distr" prediction that provides information on the probability time distribution of CRT. The patients were stratified into risk periods (low risk, high risk, and long term) according to the SM-CRT distr prediction ( Figure 7 a). We first made comparisons in each dataset. The results showed that the number of CRT events per day in the low-risk group (0.09% vs. 0.19%, padj < 0.001) and the long-term group (0.09% vs. 0.19%, padj < 0.001) was significantly lower than that in the high-risk group based on the training set. Similar results were obtained using the prospective test dataset (0.10% vs. 0.21%, padj = 0.107; 0.04% vs. 0.21%, padj < 0.001). However, the cumulative number of days in the low-risk group was extremely low (less than 1,000 days), and no CRT events were recorded in the two external test datasets. Therefore, we combined all three test datasets for comparison. The results showed that according to the test datasets, the number of CRT events per day in the low-risk group and the long-term group was significantly lower than that in the high-risk group (0.08% vs. 0.22%, padj = 0.0254; 0.09%) vs. 0.22%, padj < 0.001) ( Figure 7 b). Due to the potential bias introduced by patient variables, we subsequently performed propensity score matching (PSM) to balance the patient variables according to the risk scores predicted by SM-CRT. The results also showed that the number of CRT events per day in the low-risk group and the long-term group was significantly less than that in the high-risk group. These results indicate that SM-CRT can be used to effectively identify high-risk CRT periods, thus optimizing the extubation time of catheters in clinical practice.

[0103] Some patients may have a low-risk prediction. In this case, even during the high-risk period defined by the distr prediction, the CRT risk is not high; therefore, we further grouped the patients by combining risk and distribution predictions. By selecting the median risk value based on all datasets as the cutoff, we divided the patients into six groups according to different combinations of risk scores (high-risk value vs. low-risk value) and distribution periods (high risk, low risk, and long term). The results confirmed that in the two trainings, the number of CRT events per day in the high-risk period of patients with low-risk scores was similar to that in the low-risk period of patients with high-risk scores (0.08% vs. 0.11%, p = 0.159) and in the test (0.15% vs. 0.14%, p = 1) datasets ( Figure 7c). And both were significantly lower than the high-risk periods of patients with high risk scores in the training and test datasets ( Figure 7 c). Therefore, SM-CRT can provide more detailed risk stratification for complex clinical practice by combining risk and distribution prediction.

[0104] The model interpretation systematically revealed the key factors related to CRT: Our goal was to interpret SM-CRT based on all test datasets. Most of the results were similar to those of the univariate analysis ( Figure 5 ). For different tumor sites, the importance order for predicting CRT was as follows: mediastinum, esophagus, biliary tract, liver, breast, lung / pleura, pancreas, head / neck, kidney / adrenal gland, lymphoma, uterus / vagina, stomach, peritoneum, limb, intestine, rectum, testis, bladder / ureter, ovary / fallopian tube, trunk wall, and prostate ( Figure 5 ). Thoracic tumors were also confirmed to have a higher risk than abdominal tumors, although this difference was not significant (median area 111.14 vs. 44.07, p = 0.067), while pelvic tumors (median area 111.14 vs. -16.56, p = 0.024). In addition, the area of abdominal tumors was higher than that of pelvic tumors (median 44.07 vs. -16.56, p = 0.036) (the testis was considered to be located in the pelvis). Regarding chemotherapy, the importance order was as follows: nitrogen mustard, platinum, antimetabolite, DNA topoisomerase inhibitor, microtubule inhibitor, and antitumor antibiotic ( Figure 5 ). Tumor stage and blood markers were not very important for CRT prediction ( Figure 5 ). Overall, catheter type, tumor site, type of systemic treatment, and general patient characteristics were important factors related to CRT.

[0105] Application of SM-CRT in a web-based Shiny application: The results of SM-CRT during application are shown as follows: Function one is to guide the selection of catheter type by comparing the risk prediction of optional catheter types of SM-CRT ( Figure 6 a). Note that the systemic treatment in this tool was not set to be comparable because the main concern in clinical practice is the antitumor effect rather than CRT prevention. Function two is to identify high-risk patients based on the risk prediction of SM-CRT ( Figure 6 b). The cut-off value was set to the median risk in the training dataset; however, for specific practical application scenarios, the cut-off value needs to be customized. Function three is to guide catheter timing by providing the high-CRT risk period through the PDF determined according to the distr prediction of SM-CRT ( Figure 6c). In clinical practice, it is recommended to perform more customized risk and time classification by combining risk and distribution prediction to guide the catheter insertion time and the application of antithrombotic drugs. Function four is to optimize the ultrasound follow-up cycle according to the cumulative distribution function (CDF) determined by the distr prediction of SM-CRT ( Figure 6 d). If the CDF increases by 5% in the cumulative CRT probability compared to the previous ultrasound examination (which needs to be further adjusted according to specific clinical situations), a new ultrasound examination is recommended.

[0106] The exemplary embodiments of the present disclosure described in detail above are merely illustrative and not restrictive. Those skilled in the art should understand that various modifications and combinations can be made to these embodiments or their features without departing from the principles and spirit of the present disclosure, and such modifications should fall within the scope of the present disclosure.

Claims

1. A method for predicting the risk of catheter-related thrombosis based on the tumor location, characterized in that The method includes: S101. Obtain the tumor location of the subject; S102. Output an auxiliary prediction result of the probability risk of CRT occurrence for the subject according to different tumor locations; if the tumor location is in the chest cavity, output an auxiliary prediction result of a high risk of CRT occurrence for the subject; if the tumor location is in the abdominal cavity, output an auxiliary prediction result of a medium risk of CRT occurrence for the subject; if the tumor location is in the pelvic cavity, output an auxiliary prediction result of a low risk of CRT occurrence for the subject.

2. The method for predicting the risk of catheter-related thrombosis based on the tumor location according to claim 1, wherein The method further includes: calculating the spatial distance from the tumor location to the superior vena cava where the catheter is located; outputting an auxiliary prediction result of the probability of CRT occurrence for the subject according to the spatial distance; if the spatial distance is less than the first threshold, output an auxiliary prediction result of a large probability of CRT occurrence for the subject; if the spatial distance is greater than the first threshold, output an auxiliary prediction result of a small probability of CRT occurrence for the subject.

3. The method for predicting the risk of catheter-related thrombosis based on the tumor location according to claim 1, wherein The order of the risk of CRT occurrence in the organs in the chest cavity from high to low is as follows: mediastinum, esophagus, lung / pleura, lymphoma.

4. The method for predicting the risk of catheter-related thrombosis based on the tumor location according to claim 1, wherein, The order of the risk of CRT occurrence in the organs in the abdominal cavity from high to low is as follows: biliary tract, liver, pancreas, kidney / adrenal gland, stomach, peritoneum, intestine, rectum; Optionally, the order of the risk of CRT occurrence in the pelvic cavity from high to low is as follows: uterus / vagina, testis, bladder / ureter, ovary / fallopian tube, prostate.

5. The method for predicting the risk of catheter-related thrombosis based on the tumor location according to claim 1, wherein The tumor location further includes any one or more of the following: breast, head / neck, limbs, body wall.

6. The method for predicting the risk of catheter-related thrombosis based on the tumor location according to claim 1, wherein The order of the risk of CRT occurrence in the tumor location from high to low is as follows: mediastinum, esophagus, biliary tract, liver, breast, lung / pleura, pancreas, head / neck, kidney / adrenal gland, lymphoma, uterus / vagina, stomach, peritoneum, limbs, intestine, rectum, testis, bladder / ureter, ovary / fallopian tube, body wall, prostate.

7. The method for predicting the risk of catheter-related thrombosis based on the tumor location according to claim 1, wherein The method for determining the order of the risk of CRT occurrence in the tumor location from high to low includes: Obtain a training set sample including any one or more of the data of sample basic information, catheter type, systemic drug treatment, tumor location and the corresponding result information of the sample; Process the data of systemic drug treatment and tumor location by using the factor folding method to obtain the processed data features of systemic drug treatment and tumor location; convert the basic information and catheter type data into numerical features; Perform sensitivity analysis by using the change of the survival function value with the change of the feature value; calculate the area and direction between the highest and lowest time survival function curves, determine the importance of the feature in predicting CRT according to the area and direction, and determine the sorting result.

8. A computer device, characterized in that, The device includes: a memory and a processor; the memory is used for storing a computer program; the processor executes the computer program to implement the steps of the method according to any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, A computer program is stored thereon, and when the computer program is executed by the processor, the steps of the method according to any one of claims 1-7 are implemented.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1-7 are implemented.

Images