A method, device, medium, and program product for cTCD-based cryptogenic stroke risk assessment

Abstract

The embodiment of the specification provides a cTCD-based cryptogenic stroke risk assessment method, device, medium and program product, and relates to the field of intelligent medical treatment. The method comprises the following steps: acquiring clinical data of a detection object, including right-to-left shunt grading data, systolic pressure value, plasma D-dimer level, neutrophil count, hyperlipidemia prevalence and drinking history; and calculating a cryptogenic stroke occurrence risk probability according to the clinical data. The application first organically fuses quantified cTCD-RLS grading with conventional blood biomarkers and clinical characteristics, and constructs a precise risk stratification tool for the RoPE score "gray zone" population.

Description

Technical Field

[0001] This invention relates to the field of intelligent healthcare, and more specifically, to a method, device, medium, and procedure for assessing the risk of cryptogenic stroke based on cTCD. Background Technology

[0002] Cryptogenic stroke (CS) refers to ischemic stroke whose cause remains unclear after standardized evaluation. It accounts for 30%–40% of all ischemic strokes and is more common in young patients. Current research confirms that right-to-left shunt (RLS) caused by patent foramen ovale (PFO) is a significant pathogenic mechanism of CS. Currently, commonly used clinical methods for detecting PFO-related RLS include contrast-enhanced transthoracic echocardiography (c-TTE), contrast-enhanced transesophageal echocardiography (c-TEE), and contrast-enhanced transcranial Doppler ultrasound (c-TCD). Among these, c-TCD has been widely used for RLS screening and diagnosis due to its high sensitivity, non-invasiveness, high repeatability, and ability to accurately quantify RLS grading.

[0003] However, PFO is a common cardiac abnormality with a prevalence of 14%–35%, but only a small percentage of PFO-RLS leads to CS. Clinically, there is a dilemma of both "overtreatment" and "undertreatment": on the one hand, some RLS patients without pathogenic risk may suffer surgical complications due to indiscriminate closure; on the other hand, some truly high-risk individuals miss intervention opportunities due to a lack of accurate identification tools. Therefore, there is an urgent need for a risk stratification tool that can accurately identify pathogenic RLS to guide individualized clinical decision-making.

[0004] Regarding the mechanism by which RLS leads to CS, the academic community generally believes that it involves the synergistic effect of multiple pathophysiological factors, and is not simply a matter of "the higher the RLS grade, the higher the probability of developing the disease." Existing comprehensive assessment tools (such as the Paradoxical Embolism Risk (RoPE) score and the Probability of Causal Cause of PFO-Related Stroke (PASCAL) classification system) have shortcomings. Although the RoPE score can identify low-risk and high-risk populations, its ability to differentiate between the intermediate-risk population (RoPE score 4-6), which accounts for 38% to 62%, is limited, leading to difficulties in clinical decision-making for this "grey area" population. In addition, the RoPE score lacks evaluation of the hemodynamic characteristics of PFO (such as RLS grade). Although the PASCAL classification system combines the anatomical and hemodynamic characteristics of PFO to achieve more refined stratification, it is highly dependent on invasive and expensive c-TEE examination, has many contraindications, poor patient comfort, and is difficult to popularize in primary hospitals, resulting in low accessibility.

[0005] Therefore, there is an urgent need for an automated, non-invasive risk assessment system that integrates functional RLS classification with readily available blood / clinical indicators, especially a novel predictive model that can compensate for the RoPE score's insufficient ability to distinguish between intermediate-risk populations and overcome the shortcomings of the PASCAL classification system, such as its reliance on invasive procedures. Summary of the Invention

[0006] This invention aims to address at least one of the technical problems existing in the prior art. To this end, this invention provides a method, device, medium, and procedure for cryptogenic stroke risk assessment based on cTCD. This invention combines six core factors: RLS classification, systolic blood pressure, plasma D-dimer level, neutrophil count, alcohol consumption history, and hyperlipidemia. For the first time, it organically integrates quantitative cTCD-RLS classification with routine blood biomarkers and clinical characteristics, constructing a precise risk stratification tool for individuals in the "grey area" of the RoPE score.

[0007] The first aspect of this application discloses a method for assessing the risk of cryptogenic stroke based on cTCD, including: The data collected for testing included right-to-left shunt classification, systolic blood pressure, plasma D-dimer levels, neutrophil count, prevalence of hyperlipidemia, and history of alcohol consumption. The probability of cryptogenic stroke was calculated based on the clinical data.

[0008] In some embodiments, the method further includes: The probability of occurrence of the cryptogenic stroke is compared with a preset risk threshold; If the probability of cryptogenic stroke is greater than or equal to a preset risk threshold, the subject is determined to be a high-risk group for cryptogenic stroke, and an assessment report recommending foramen ovale closure is generated. If the probability of cryptogenic stroke is less than a preset risk threshold, the subject is determined to be a low-risk group for cryptogenic stroke, and an assessment report recommending conservative treatment is generated.

[0009] The second aspect of this application discloses a cTCD-based system for assessing the risk of cryptogenic stroke, comprising: The data acquisition module is used to acquire clinical data of the test subjects, including right-to-left shunt classification data, systolic blood pressure value, plasma D-dimer level, neutrophil count, hyperlipidemia status and alcohol consumption history; The data processing module is used to calculate the probability of cryptogenic stroke based on the clinical data. The result output module is used to output the probability of the occurrence of the cryptogenic stroke.

[0010] A third aspect of this application discloses a computer device, comprising: 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-described method.

[0011] The fourth aspect of this application discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0012] The fifth aspect of this application discloses a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method.

[0013] This application has the following beneficial effects: 1. This application establishes for the first time a combination of six core predictive factors, consisting of RLS classification, systolic blood pressure, plasma D-dimer level, neutrophil count, prevalence of hyperlipidemia, and history of alcohol consumption. Experimental data show that the predictive efficacy of any single factor is low (AUC between 0.610 and 0.657), and cannot independently support clinical decision-making. However, this approach, by organically combining these six factors, improves the AUC of the constructed assessment method to 0.843, with a sensitivity of 90.5%. This performance leap is not a simple linear additive effect, but reflects the synergistic effect of hemodynamic disorders (RLS), coagulation activation (D-dimer), inflammatory response (neutrophils), and vascular risk factors (blood pressure, blood lipids, alcohol consumption) under multiple pathological mechanisms, solving the technical pain point of inaccurate prediction by a single indicator.

[0014] 2. Compared to the PASCAL system, which relies on c-TEE, this approach is entirely based on non-invasive cTCD examination, avoiding the pain of intubation, anesthesia risks, and contraindications, thus improving patient acceptance. The low cost, ease of operation, and high repeatability of cTCD equipment enable this assessment method to be implemented in primary hospitals and community screening scenarios, solving the problem of limited access to high-end assessment tools and achieving risk stratification for large populations.

[0015] 3. This solution is specifically optimized for the "intermediate-risk population" that cannot be effectively distinguished by the RoPE score. By outputting specific risk probabilities and comparing them with preset thresholds, the previously ambiguous "gray area" population is further subdivided into high-risk (recommended for closure) and low-risk (recommended for conservative treatment). The quantified risk stratification results directly assist doctors in formulating surgical criteria, not only effectively identifying high-risk patients suitable for closure surgery, but also reducing unnecessary PFO closure procedures, lowering medical costs and the risk of complications, and avoiding missed diagnoses of high-risk patients. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the method flow provided in the first aspect of the present invention; Figure 2 This is a schematic diagram of a cTCD-based cryptogenic stroke risk assessment system provided in the second aspect of the present invention. Figure 3 This is a comparison of the predictive power of the model provided in this application with that of any single factor (ROC analysis); Figure 4 This is the dynamic column graph interface of the Shiny webpage provided in this embodiment of the invention; Figure 5 This is a flowchart of the operation method for user selection of inspection mode provided in an embodiment of the present invention. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0019] In some of the processes described in the specification, claims, and accompanying drawings of this invention, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Figure 1This is a schematic diagram of a cTCD-based method for assessing the risk of cryptogenic stroke, provided by an embodiment of the present invention. Specifically, the method includes the following steps: S101: Acquire clinical data including right-to-left shunt classification, systolic blood pressure, plasma D-dimer level, neutrophil count, prevalence of hyperlipidemia, and history of alcohol consumption; Among them, the right-to-left diversion classification data belongs to functional indicators. Plasma D-dimer levels and neutrophil counts are hematological indicators. Systolic blood pressure, prevalence of hyperlipidemia, and history of alcohol consumption are basic clinical indicators; In some embodiments, the terms “subject” or “test subject” or “sample” as used herein refer to any animal (e.g., a mammal), including but not limited to humans, non-human primates, rodents, etc., which will become the recipient of a particular treatment. Generally, the terms “subject” and “patient” are used interchangeably herein when referring to human subjects. Preferably, the subject is a human.

[0022] In some embodiments, the sample to be tested is a patient undergoing prognostic assessment in a clinical setting.

[0023] In some embodiments, the right-to-left shunt grading data is quantitative grading data obtained by contrast-enhanced transcranial Doppler ultrasound. The prevalence of hyperlipidemia includes a binary variable indicating whether or not one has hyperlipidemia; The drinking history includes whether or not one has a history of drinking alcohol.

[0024] In some embodiments, the above six clinical data were obtained through the following screening method: 757 hospitalized patients were retrospectively collected (all with positive cTCD results; according to the inclusion and exclusion criteria, 340 patients were finally included, of whom 133 had CS and 207 did not), and 37 candidate variables were initially identified. After univariate screening, Lasso dimensionality reduction, and multivariate regression analysis, confounding, meaningless, and severely collinear indicators were eliminated layer by layer, and finally six core factors were selected: RLS grade, systolic blood pressure, plasma D-dimer level, neutrophil count, history of alcohol consumption, and hyperlipidemia.

[0025] S102: Calculate the probability of cryptogenic stroke based on the clinical data; In some embodiments, the clinical data are processed using a regression model or a machine learning model to obtain the probability of cryptogenic stroke.

[0026] In some embodiments, the regression model is a regression model with a penalty term, including LASSO regression model, ridge regression model, or elastic network regression model. Alternatively, the machine learning model may be an ensemble learning model, which may include a random forest model or a gradient boosting tree model.

[0027] In some embodiments, the regression model is a logistic regression model, and the calculation of the probability of cryptogenic stroke based on the clinical data includes: Logit(P)=β0+β1×RLS+β2×SBP+β3×Hyperlipidemia+β4×Alcohol+β5×Neutrophil +β6×D-Dimer, P = 1 / (1+e) Logit(P) , Wherein, P is the probability of the occurrence of cryptogenic stroke, β0 is the intercept term, β1-β6 are the regression coefficients of each predictor; RLS is the quantitative value corresponding to the right-to-left shunt grade, SBP is the systolic blood pressure value, Hyperlipidemia is the binarized variable of hyperlipidemia status, Alcohol is the binarized variable of alcohol consumption history, Neutrophil is the neutrophil count value, and D-Dimer is the plasma D-dimer level value.

[0028] In some more specific embodiments, the specific formula for calculating Logit(P) is as follows: Logit(P)=-8.756+0.562×RLS+0.037×SBP+1.952×Hyperlipidemia + 0.925×Alcohol + 0.220×Neutrophil + 2.093×D-Dimer, However, please note that the values ​​of the regression coefficients and intercept terms mentioned above are not fixed and will change depending on the data.

[0029] In some embodiments, the regression model described above can also be constructed using a LASSO (Least Absolute Shrinkage and Selection Operator) regression model. LASSO regression, by introducing an L1 regularization penalty term into the loss function, can automatically compress the regression coefficients of some insignificant variables to zero during model training, thereby achieving simultaneous completion of feature selection and model construction.

[0030] In this specific application, data from 37 candidate variables of 340 hospitalized patients with positive cTCD were retrospectively collected, and LASSO regression was used for dimensionality reduction. The optimal penalty parameter λ was determined through cross-validation. The model automatically selected six core predictive factors from the 37 candidate variables: RLS grade, systolic blood pressure, plasma D-dimer level, neutrophil count, hyperlipidemia, and history of alcohol consumption, and assigned non-zero regression coefficients to each factor. Based on the selected six factors and their coefficients, a LASSO regression scoring formula was constructed to calculate the probability of cryptogenic stroke.

[0031] Alternatively, the regression model can be constructed using the Elastic Net regression model. Elastic Net regression combines the L1 regularization term of LASSO regression and the L2 regularization term of Ridge regression, enabling both feature selection and group selection of collinear variables.

[0032] In the specific application of this embodiment, when there is a high correlation among candidate variables (e.g., collinearity may exist between multiple inflammatory markers or different blood pressure measurements), the elastic network can retain or remove related variables as a group simultaneously, rather than randomly selecting one like LASSO. By adjusting the ratio of L1 and L2 regularization parameters, the model achieves a balance between feature sparsity and variable group selection. The final six selected factors and their weight coefficients are used to calculate the probability of cryptogenic stroke.

[0033] Alternatively, when using a random forest model, the random forest can capture the non-linear relationship between factors and outcomes, as well as the interaction effects between factors, by constructing multiple decision trees and integrating their predictions.

[0034] In this specific application, RLS classification, systolic blood pressure, plasma D-dimer level, neutrophil count, hyperlipidemia, and history of alcohol consumption are used as input features, and the occurrence of cryptogenic stroke is used as a binary outcome variable to train a random forest model. During model training, each decision tree is built on a randomly sampled subset of samples and a randomly selected subset of features, and finally outputs the probability of cryptogenic stroke occurrence through majority voting or average probability.

[0035] Alternatively, when using a gradient boosting tree model, the gradient boosting tree iteratively builds decision trees, with each new tree focusing on correcting the prediction residuals of the preceding tree set, gradually improving the overall model's prediction accuracy.

[0036] In this specific application, six factors—RLS classification, systolic blood pressure, plasma D-dimer level, neutrophil count, hyperlipidemia, and history of alcohol consumption—are used as input features, with the occurrence of cryptogenic stroke as the outcome variable. A gradient boosting algorithm (such as XGBoost, LightGBM, or CatBoost) is used to train the prediction model. By setting appropriate hyperparameters such as learning rate, tree depth, and number of iterations, the model progressively optimizes the loss function on the training set, ultimately outputting the calibrated probability of cryptogenic stroke occurrence.

[0037] In some embodiments, the method further includes: The probability of occurrence of the cryptogenic stroke is compared with a preset risk threshold; If the probability of cryptogenic stroke is greater than or equal to a preset risk threshold, the subject is determined to be a high-risk group for cryptogenic stroke, and an assessment report recommending foramen ovale closure is generated. If the probability of cryptogenic stroke is less than a preset risk threshold, the subject is determined to be a low-risk group for cryptogenic stroke, and an assessment report recommending conservative treatment is generated.

[0038] In some embodiments, the aforementioned assessment report includes, but is not limited to, paper or electronic versions. The results are obtained by intelligent machines based on the relevant data of the subjects and are intended only as a reference for medical personnel, not as the final diagnostic result of the subjects.

[0039] In some embodiments, the relevant thresholds mentioned above are obtained through training with training set samples. They can be specific thresholds or ranges, and the specific form is not specifically limited in this embodiment.

[0040] After internal and external independent cohort validation, the RASCAL model demonstrated good discrimination, calibration, and clinical applicability. Compared with the RoPE score, this model overcomes its limited ability to differentiate between intermediate-risk populations. In the same external validation cohort, the RASCAL model achieved an AUC of 0.843, with a sensitivity of 90.5% and a specificity of 65.6%; while the RoPE score, validated using a 7-cutoff value, showed an AUC of 0.704, a sensitivity of 69.4%, and a specificity of 62.5%. These results clearly demonstrate that by combining RLS triage with clinical history and blood biomarkers, this model significantly improves the accuracy of CS risk stratification.

[0041] Furthermore, the predictive power of this invention (AUC=0.843) is significantly superior to the predictive power of any single factor (based on the same dataset analyzed in this study): using only D-dimer to predict CS, the AUC was 0.657, systolic blood pressure 0.643, hyperlipidemia 0.610, alcohol consumption history 0.633, neutrophils 0.617, and RLS grade 0.615. No single factor mentioned above could achieve the predictive level of this model. This significant improvement in predictive power compared to individual factors demonstrates that the overall effect is clearly "greater than the sum of its parts" (e.g., Figure 3 (As shown).

[0042] In some embodiments, this embodiment also discloses a cTCD-based cryptogenic stroke risk assessment system, such as... Figure 2 As shown, it includes: The data acquisition module 201 is used to acquire clinical data of the test subjects, including right-to-left shunt classification data, systolic blood pressure value, plasma D-dimer level, neutrophil count, hyperlipidemia status and alcohol consumption history; Data processing module 202 is used to calculate the probability of cryptogenic stroke based on the clinical data; The result output module 203 is used to output the probability of occurrence of the hidden-source stroke.

[0043] In some embodiments, the results output module is also used to output a visual chord chart and generate a risk stratification report; In some embodiments, the system further includes a storage module for storing the constructed mathematical model and evaluation logs.

[0044] In some embodiments, the above mathematical model can be transformed into a dynamic nomogram and deployed in an interactive web application system based on the Shiny framework (interface as shown in the image). Figure 4 (As shown).

[0045] This application integrates cTCD grading with blood biomarkers to output continuous risk probability values, which are then visually displayed in a dynamic nomogram format, helping clinicians quantify and assess individualized risk of cryptogenic stroke. More specifically, as an example, a detailed description of the dynamic nomogram risk assessment process based on the Shiny webpage is provided, including the following steps (specific operational steps are as follows...). Figure 5 (as shown) Step 1: Data Acquisition The system receives six parameters from the user through an interactive user interface (built on the Shiny framework): RLS classification: Determined by cTCD detection, it is divided into Grade I (1~10); Grade II (11~25); Grade III (>25, non-rain curtain); Grade IV (>25, rain curtain / uncountable) based on the number of microvesicles in the middle cerebral artery within 25 seconds; Input format is drop-down menu.

[0046] Systolic blood pressure (SBP): Input box for numerical values, unit mmHg.

[0047] Hyperlipidemia: Select the radio button (YES / NO), based on clinical diagnosis.

[0048] Alcohol consumption history: Select via radio button (YES / NO), based on clinical history taking.

[0049] Neutrophil count: Numeric input box, unit × 10 9 / L.

[0050] Plasma D-dimer: Numerical input box, unit ug / ml.

[0051] Step 2: Data Preprocessing The system automatically checks whether the input values ​​are within the preset reasonable range (e.g., SBP 0~184mmHg, D-dimer 0~4ug / ml). No missing values ​​are handled (all input fields are required).

[0052] Step 3: Calculate the risk probability using the built-in mathematical model. The model is based on a formula constructed using multivariate logistic regression: Logit(P)=β0+β1X1+β2X2+β3X3+β4X4+β5X5+β6X6 β0= 8.756 (intercept, determined based on the training queue). β1 = 0.562 (RLS grading coefficient). β2 = 0.037 (systolic pressure coefficient) β3 = 1.952 (hyperlipidemia coefficient, YES for yes, NO for no, where YES is assigned a value of 1 and NO is assigned a value of 0). β4 = 0.925 (drinking history coefficient, YES for yes and NO for no, where YES is assigned a value of 1 and NO is assigned a value of 0). β5 = 0.220 (neutrophil count coefficient). β6 = 2.093 (D-dimer system number).

[0053] Calculate the probability of risk: P = 1 / (1+e) Logit(P) , Example: A patient has an RLS grade of 4, SBP of 120, hyperlipidemia, alcohol consumption history, neutrophils of 4.9, D-dimer of 0.2, and the probability of occurrence (P) is approximately 90.1%.

[0054] Step 4: Output Module By inputting the actual values ​​of the six features required by the model, the application can instantly predict the risk probability of a patient developing CS.

[0055] A third aspect of this application discloses a computer device, which may include: one or more processors and one or more memories; wherein the memories store computer-readable code that, when run by the one or more processors, can perform the methods described above.

[0056] The processor in this embodiment can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, operations, and logic block diagrams disclosed in this embodiment. The general-purpose processor can be a microprocessor or any conventional processor, and can be based on an x86 or ARM architecture.

[0057] In general, the various exemplary embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, firmware, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. When aspects of embodiments of this disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it will be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as non-limiting examples in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0058] For example, the methods or apparatus according to embodiments of this disclosure can also be implemented using the architecture of a computing device. A computing device may include a bus, one or more CPUs, read-only memory (ROM), random access memory (RAM), a communication port connected to a network, input / output components, a hard disk, etc. Storage devices in the computing device, such as ROM or hard disk, may store various data or files used for processing and / or communication of the methods provided in this disclosure, as well as program instructions executed by the CPU. The computing device may also include a user interface. Of course, the architecture described above is merely exemplary, and one or more components of the computing device described above may be omitted as needed when implementing different devices.

[0059] This invention also provides a computer-readable storage medium storing computer-readable instructions. When the computer-readable instructions are executed by a processor, the methods disclosed in this embodiment can be performed. The computer-readable storage medium in this embodiment can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be 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 (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous interconnected dynamic random access memory (SLDRAM), and direct memory bus random access memory (DRRAM). It should be noted that the memory of the methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0060] This disclosure also provides a computer program product or system, including a computer program that, when executed by a processor, implements the steps of the above-described method.

[0061] The exemplary embodiments of this disclosure described in detail above are merely illustrative and not restrictive. Those skilled in the art will understand that various modifications and combinations can be made to these embodiments or their features without departing from the principles and spirit of this disclosure, and such modifications should fall within the scope of this disclosure.

Claims

1. A method for assessing the risk of cryptogenic stroke based on cTCD, characterized in that, include: The data collected for testing included right-to-left shunt classification, systolic blood pressure, plasma D-dimer levels, neutrophil count, prevalence of hyperlipidemia, and history of alcohol consumption. The probability of cryptogenic stroke was calculated based on the clinical data.

2. The method for assessing the risk of cryptogenic stroke based on cTCD according to claim 1, characterized in that, The clinical data is processed using regression or machine learning models to obtain the probability of cryptogenic stroke.

3. The method for assessing the risk of cryptogenic stroke based on cTCD according to claim 2, characterized in that, The regression model is a regression model with a penalty term, which includes LASSO regression model, ridge regression model or elastic network regression model; Alternatively, the machine learning model may be an ensemble learning model, which may include a random forest model or a gradient boosting tree model.

4. The method for assessing the risk of cryptogenic stroke based on cTCD according to claim 2, characterized in that, The regression model is a logistic regression model, and the calculation of the risk probability of cryptogenic stroke based on the clinical data includes: Logit(P)=β0+β1×RLS+β2×SBP+β3×Hyperlipidemia+β4×Alcohol+β5×Neutrophil+β6×D-Dimer, P=1 / 1+e Logit(P) , Wherein, P is the probability of the occurrence of cryptogenic stroke, β0 is the intercept term, β1-β6 are the regression coefficients of each predictor; RLS is the quantitative value corresponding to the right-to-left shunt grade, SBP is the systolic blood pressure value, Hyperlipidemia is the binarized variable of hyperlipidemia status, Alcohol is the binarized variable of alcohol consumption history, Neutrophil is the neutrophil count value, and D-Dimer is the plasma D-dimer level value.

5. The method for assessing the risk of cryptogenic stroke based on cTCD according to claim 1, characterized in that, The right-to-left shunt grading data is quantitative grading data obtained through contrast-enhanced transcranial Doppler ultrasound. The prevalence of hyperlipidemia includes a binary variable indicating whether or not one has hyperlipidemia; The drinking history includes whether or not one has a history of drinking alcohol.

6. The method for assessing the risk of cryptogenic stroke based on cTCD according to claim 1, characterized in that, The method further includes: The probability of occurrence of the cryptogenic stroke is compared with a preset risk threshold; If the probability of cryptogenic stroke is greater than or equal to a preset risk threshold, the subject is determined to be a high-risk group for cryptogenic stroke, and an assessment report recommending foramen ovale closure is generated. If the probability of cryptogenic stroke is less than a preset risk threshold, the subject is determined to be a low-risk group for cryptogenic stroke, and an assessment report recommending conservative treatment is generated.

7. A cTCD-based system for assessing the risk of cryptogenic stroke, characterized in that, include: The data acquisition module is used to acquire clinical data of the test subjects, including right-to-left shunt classification data, systolic blood pressure value, plasma D-dimer level, neutrophil count, hyperlipidemia status and alcohol consumption history; The data processing module is used to calculate the probability of cryptogenic stroke based on the clinical data. The result output module is used to output the probability of the occurrence of the cryptogenic stroke.

8. A computer device, characterized in that, The device 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 method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the steps of the method as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the method described in any one of claims 1-6.

Images