Mechanical thrombectomy risk prediction system and method for elderly patients
By constructing a Nomogram risk prediction model and combining CT radiomics, collateral omics, and clinical data of elderly patients, risk factors were screened out, solving the problem of risk assessment for mechanical thrombectomy in elderly patients and realizing individualized and precise clinical diagnosis and treatment.
Patent Information
- Application Number
- CN202311230055.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-09-21
- Publication Date
- 2026-04-17
AI Technical Summary
The lack of an effective risk assessment system in current technology makes it difficult to predict the risk of intracranial hemorrhage and poor prognosis after mechanical thrombectomy in elderly patients, leading to a dilemma for clinicians in preoperative decision-making.
A Nomogram risk prediction model was constructed. By analyzing CT radiomics data, collateral omics data, and clinical data of elderly patients, risk factors affecting poor prognosis of mechanical thrombectomy were screened, and a risk assessment system for mechanical thrombectomy surgery for elderly patients was established.
It provides individualized and precise clinical treatment plans, helping clinicians reduce the risks of mechanical thrombectomy in elderly patients and improve treatment outcomes during preoperative decision-making.
Smart Images
Figure CN121885168A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of risk prediction technology for stroke surgery, specifically to a system and method for predicting the risk of mechanical thrombectomy surgery for elderly patients. Background Technology
[0002] Stroke is a group of cerebrovascular diseases characterized by acute focal neurological deficits, and its incidence increases with age. The main treatment methods for stroke are intravenous thrombolysis and endovascular mechanical thrombectomy.
[0003] With the advancements in imaging techniques and the introduction of the tissue window concept in recent years, numerous large-scale randomized controlled trials (RCTs) have confirmed that patients with ischemic penumbra selected by imaging and treated with mechanical thrombectomy have a significantly better prognosis than those treated with intravenous thrombolysis alone. Although mechanical thrombectomy is effective, it carries a risk of symptomatic intracranial hemorrhage (SICH) and even death for patients over 80 years of age. Compared to younger patients, even with complete recanalization, elderly patients are twice as likely to experience SICH and poor outcomes post-surgery. Therefore, clinicians often face a dilemma when dealing with elderly patients due to concerns about surgical risks. How to effectively predict the risks of mechanical thrombectomy in elderly patients to aid preoperative decision-making and individualized treatment plans, and to minimize overtreatment or undertreatment to maximize their benefits during treatment, has become an urgent clinical challenge.
[0004] In recent years, researchers have introduced various risk scoring systems to predict the risk of mechanical thrombectomy in the overall stroke population. Among these, the Totaled Health Risks in Vascular Events (THRIVE) score, the Houston Intra-Arterial Therapy (HIAT) score, and the Preadmission comorbidities, level of consciousness, age, and neurological deficit (PLAN) score are commonly used in clinical practice. However, for the specific group of elderly patients, there is still no effective risk assessment system to help clinicians predict their risk of developing SICH and adverse outcomes post-operatively. Summary of the Invention
[0005] The purpose of this invention is to provide a system and method for predicting the risks of mechanical thrombectomy in elderly patients.
[0006] 1. To achieve the above objectives, in a first aspect, embodiments of the present invention provide a risk prediction system for mechanical thrombectomy surgery in elderly patients, comprising:
[0007] The database creation module is used to create a clinical database of elderly AIS-aLVO patients; the clinical database stores CT radiomics data and collateral omics data of elderly AIS-aLVO patients.
[0008] The risk factor screening module is used to screen risk factors for poor prognosis of acute mechanical thrombectomy in elderly patients based on the CT radiomics data and collateral omics data, combined with clinical data.
[0009] The model building module is used to build a Nomogram risk prediction model based on the risk factors.
[0010] The thrombectomy risk prediction module is used to predict the risk of mechanical thrombectomy in elderly AIS-aLVO patients using the Nomogram risk prediction model.
[0011] As one specific implementation of this application, the database creation module is specifically used for:
[0012] The study subjects were selected and grouped to obtain the first result; the study subjects were elderly patients who were confirmed by multimodal CT examination to have acute anterior circulation large vessel occlusive ischemic stroke, underwent mechanical thrombectomy, and were regularly followed up after the operation; the first result includes the imaging data and grouping of the study subjects;
[0013] Clinical information of all enrolled patients was collected and evaluated to obtain a second result;
[0014] Multimodal CT radiomics features of all enrolled patients were extracted and analyzed to obtain a third result; the third result includes conventional radiomics information analysis results and collateral radiomics information analysis results.
[0015] A clinical database of elderly AIS-aLVO patients was constructed based on the first result, the second result, and the third structure.
[0016] The selection and grouping of research subjects are as follows:
[0017] Study subjects were selected based on inclusion and exclusion criteria. Inclusion criteria included: ① Age ≥ 80 years; ② All patients arrived at the hospital within 24 hours of symptom onset or the last known normality, and underwent a one-stop multimodal CT scan; ③ CTA showed occlusion of the internal carotid artery and / or the M1 segment of the middle cerebral artery; ④ Mechanical thrombectomy was performed within 24 hours of symptom onset or the last known normality; ⑤ Informed consent was obtained from the patient or their guardian. Exclusion criteria included: ① Intracranial hemorrhage on cranial imaging upon admission; ② Transient ischemic attack (TIA); ③ Intravenous thrombolysis prior to mechanical thrombectomy; ④ A history of cancer or discovery of cancer during hospitalization, hematologic disorders, or autoimmune diseases; ⑤ Incomplete clinical diagnosis and treatment history and medical records; ⑥ Imaging quality not meeting diagnostic requirements.
[0018] The study subjects were divided into a SICH group and a non-SICH group based on whether intracranial hemorrhage occurred after mechanical thrombectomy; and into a good prognosis group and a poor prognosis group based on whether adverse outcomes occurred 3 months after mechanical thrombectomy.
[0019] For each group of patients, they were randomly divided into two subgroups using computer random numbers: 70% were assigned to the training group and 30% to the validation group.
[0020] As one specific implementation of this application, the clinical information of all enrolled patients is collected and evaluated, specifically as follows:
[0021] All enrolled patients had relevant clinical data collected, including age, gender, medical history, medication history, smoking history, pre-admission mRS score, blood pressure, blood glucose, blood lipids, laboratory tests, NIHSS score, and cerebral infarction type.
[0022] Record the details of the mechanical thrombectomy procedure, including the time from the onset of the patient's illness to the imaging examination, the time from the onset of the illness to the start of the procedure, the time from the imaging examination to the start of the procedure, the time from the start of the procedure to the recanalization of the blood vessel, and the procedure method.
[0023] A modified thrombolysis grading system for cerebral infarction was used to assess vascular recanalization, with 0-2a representing poor recanalization and 2b-3 representing good recanalization.
[0024] As one specific implementation of this application, the multimodal CT radiomics features of all enrolled patients are extracted and analyzed, specifically as follows:
[0025] Acquire preoperative CT images of all patients and extract the imaging features of the CT images;
[0026] ASPECT scores were obtained by plain CT scans to evaluate early ischemic changes in the middle cerebral artery supply area, and CTA was used to evaluate the location of vascular occlusion and the degree of thrombus burden.
[0027] The preoperative establishment of leptomeningeal collaterals was evaluated using the rLMC score based on CTA, and the infarct core volume of CBV was obtained through CTP to assess the functional status of leptomeningeal collaterals.
[0028] As a specific implementation of this application, risk factors for poor prognosis of acute mechanical thrombectomy in elderly patients are screened out, specifically:
[0029] Univariate analysis was used to analyze relevant clinical information, routine imaging information, and collateral omics information to preliminarily screen risk factors associated with acute-phase mechanical thrombectomy (SICH) and poor prognosis in elderly patients.
[0030] As a specific implementation of this application, a Nomogram risk prediction model is constructed based on the aforementioned risk factors, specifically as follows:
[0031] The minimum absolute contraction and selection operator algorithm is used to incorporate meaningful variables from univariate analysis into a multivariate logistic regression model to construct a Nomogram risk prediction model.
[0032] As a preferred implementation of this application, the system further includes a model processing module, used for:
[0033] Model demonstration: The constructed Nomogram risk prediction model is visualized;
[0034] Model evaluation: The performance of the Nomogram risk prediction model was evaluated using receiver operating characteristics and area under the ROC curve; the calibration curve of the risk prediction model was plotted and the HL goodness-of-fit test was performed to evaluate the calibration of the Nomogram risk prediction model; the DCA curve was plotted to evaluate the clinical application value of the Nomogram risk prediction model.
[0035] Model Validation and Optimization: The Nomogram risk prediction model was validated and optimized using the validation group in the grouping scenario.
[0036] Secondly, embodiments of this application provide a method for predicting the risks of mechanical thrombectomy surgery for elderly patients, including:
[0037] A clinical database of elderly patients with AIS-aLVO was established; the clinical database contains CT radiomics data and collateral omics data of elderly patients with AIS-aLVO.
[0038] Based on the CT radiomics data and collateral omics data, combined with clinical data, risk factors for poor prognosis of mechanical thrombectomy in elderly patients during the acute phase were screened.
[0039] Construct a Nomogram risk prediction model based on the aforementioned risk factors;
[0040] The Nomogram risk prediction model was used to predict the risk of mechanical thrombectomy in elderly patients with AIS-aLVO.
[0041] This invention, implemented in this embodiment, targets elderly patients with AIS-aLVO. Based on one-stop multimodal CT information, it analyzes relevant collateral omics information, conventional radiomics information, and clinical data to screen risk factors affecting poor prognosis of mechanical thrombectomy in the acute phase in elderly patients. A Nomogram-based risk prediction model is constructed, and the generalization ability and clinical feasibility of the model are further verified through a prospective cohort study. Ultimately, a risk assessment system for mechanical thrombectomy in elderly patients with acute anterior circulation large vessel occlusion is established, providing more evidence-based basis for clinicians to make preoperative decisions and enabling individualized and precise diagnosis and treatment for elderly patients with ischemic stroke. Attached Figure Description
[0042] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.
[0043] Figure 1 This is a structural diagram of the mechanical thrombectomy surgery risk prediction system for elderly patients provided in the first embodiment of the present invention;
[0044] Figure 2 This is a schematic diagram of the CBS thrombosis burden score (AJNR Am J Neuroradiol, 2009, 30:525-31);
[0045] Figure 3 This is a schematic diagram of the rLMC score (AJNR Am J Neuroradiol, 2011, 32:1640-5);
[0046] Figure 4 This is an example of a Nomogram prediction model (Front Neurol, 2021, 12:710144);
[0047] Figure 5 This is a flowchart of a method for predicting the risk of mechanical thrombectomy surgery for elderly patients, provided in the first embodiment of the present invention. Detailed Implementation
[0048] 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, not all, of the embodiments of the present invention. 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.
[0049] The Nomogram model is a computational tool that transforms the results of risk prediction models into visual graphs. It does not approximate the prediction model; instead, it directly assigns a score to each level of each predictor based on the contribution of each predictor to the outcome variable (the magnitude of the regression coefficient). The scores of all predictors are then summed to obtain a total score. Finally, by using a function relating the total score to the probability of the outcome event, the predicted probability of an individual outcome event is quantitatively calculated.
[0050] As a predictive risk measurement tool that provides clinicians with intuitive and visual results, the Nomogram model has received widespread attention in clinical practice in recent years. Researchers have established Nomogram models for patients with minor ischemic stroke and transient ischemic attack, which have effectively predicted adverse prognoses and the risk of recurrence. However, there are currently no reported studies using Nomogram models to predict the risk of mechanical thrombectomy in elderly stroke patients.
[0051] Therefore, this study will use elderly patients with AIS-aLVO as the research subjects to explore the risk factors for SICH and poor prognosis after mechanical thrombectomy, and on this basis, construct a Nomogram risk prediction model, in order to play an important role in the preoperative decision-making of elderly patients and make clinical diagnosis and treatment more individualized and precise.
[0052] Please refer to Figure 1 This invention provides a risk prediction system for mechanical thrombectomy surgery in elderly patients, comprising:
[0053] The database creation module is used to create a clinical database of elderly AIS-aLVO patients; the clinical database stores CT radiomics data and collateral omics data of elderly AIS-aLVO patients.
[0054] The risk factor screening module is used to screen risk factors for poor prognosis of acute mechanical thrombectomy in elderly patients based on the CT radiomics data and collateral omics data, combined with clinical data.
[0055] The model building module is used to build a Nomogram risk prediction model based on the risk factors.
[0056] The thrombectomy risk prediction module is used to predict the risk of mechanical thrombectomy in elderly AIS-aLVO patients using the Nomogram risk prediction model.
[0057] Specifically, in this embodiment, the database establishment module is specifically used for:
[0058] The study subjects were selected and grouped to obtain the first result; the study subjects were elderly patients who were confirmed by multimodal CT examination to have acute anterior circulation large vessel occlusive ischemic stroke, underwent mechanical thrombectomy, and were regularly followed up after the operation; the first result includes the imaging data and grouping of the study subjects;
[0059] Clinical information of all enrolled patients was collected and evaluated to obtain a second result;
[0060] Multimodal CT radiomics features of all enrolled patients were extracted and analyzed to obtain a third result; the third result includes conventional radiomics information analysis results and collateral radiomics information analysis results.
[0061] A clinical database of elderly AIS-aLVO patients was constructed based on the first result, the second result, and the third structure.
[0062] Furthermore, the mechanical thrombectomy surgery risk prediction system in this embodiment also includes a model processing module, used for:
[0063] Model demonstration: The constructed Nomogram risk prediction model is visualized;
[0064] Model evaluation: The performance of the Nomogram risk prediction model was evaluated using receiver operating characteristics and area under the ROC curve; the calibration curve of the risk prediction model was plotted and the HL goodness-of-fit test was performed to evaluate the calibration of the Nomogram risk prediction model; the DCA curve was plotted to evaluate the clinical application value of the Nomogram risk prediction model.
[0065] Model Validation and Optimization: The Nomogram risk prediction model was validated and optimized using the validation group in the grouping scenario.
[0066] For example, constructing a clinical database of elderly patients with acute anterior circulation large vessel occlusion ischemic stroke includes:
[0067] (1) Screening and grouping of research subjects
[0068] 1) Selection and Exclusion Criteria
[0069] This study aims to include 200 elderly patients with acute anterior circulation large vessel occlusion ischemic stroke treated at a hospital from January 1, 2011 to December 31, 2021, to establish a clinical database.
[0070] Inclusion criteria: ① Age ≥ 80 years; ② All patients arrived at the hospital within 24 hours of symptom onset or the last time they returned to normal and underwent a one-stop multimodal CT examination; ③ CTA showed occlusion of the internal carotid artery and / or the M1 segment of the middle cerebral artery; ④ Mechanical thrombectomy was performed within 24 hours of symptom onset or the last time they returned to normal; ⑤ Informed consent was signed by the patient or their guardian.
[0071] Exclusion criteria: ① Intracranial hemorrhage on cranial imaging upon admission; ② Transient ischemic attack; ③ Intravenous thrombolysis prior to mechanical thrombectomy; ④ History of cancer or discovery of cancer during hospitalization, hematologic disorders, or autoimmune diseases; ⑤ Incomplete clinical diagnosis and treatment history and clinical medical records; ⑥ Imaging image quality that does not meet diagnostic requirements.
[0072] 2) Grouping of research subjects
[0073] Patients were divided into a SICH group and a non-SICH group based on whether intracranial hemorrhage occurred after mechanical thrombectomy; and into a good prognosis group (mRS score 0–2) and a poor prognosis group (mRS score 3–6) based on whether adverse outcomes occurred 3 months after mechanical thrombectomy. Patients in each group were further randomly assigned to two subgroups using computer-generated random numbers: 70% to the training group and 30% to the validation group.
[0074] ①SICH diagnostic criteria: The definition of symptomatic intracranial hemorrhage was adopted from the European-Australasian Acute Stroke Study II (ECASS II) criteria, which is secondary hemorrhage at the ischemic site after acute ischemic stroke, resulting in an increase of ≥4 points in the NIHSS score. In addition, we also adopted the NINDS (National Institute of Neurological Disorders and Stroke trial) definition of symptomatic intracranial hemorrhage, which is defined as follow-up imaging indicating the presence of parenchymal hemorrhage, with the hemorrhage site consistent with the patient's clinical symptoms, resulting in an increase of ≥1 point in the patient's NIHSS score.
[0075] ② Prognostic evaluation: All patients were followed up by telephone or outpatient visits. Clinical functional prognosis was evaluated using the 3-month mRS score, and the scoring criteria are shown in Table 1. An mRS score > 2 was considered a poor prognosis.
[0076] Table 1. Modified Rankin Scale (mRS)
[0077]
[0078]
[0079] (2) Clinical information collection and evaluation
[0080] Medical records were collected from all enrolled patients, including: age, gender, past medical history (obesity, hypertension, diabetes, hyperlipidemia, atrial fibrillation, coronary heart disease, heart failure), medication history (antiplatelet aggregation or anticoagulants), smoking history, pre-admission mRS score, blood pressure, blood glucose, blood lipids, and laboratory tests (complete blood count, liver function, kidney function, coagulation function, C-reactive protein, procalcitonin, etc.) at admission.
[0081] Stroke severity was assessed using the NIHSS score, administered upon admission and 24 hours after the interventional procedure. Cerebral infarction was classified according to the TOAST (Trial of Org 10172 in Acute Ischemic Stroke Treatment) criteria, including large artery atherosclerosis, cardioembolic, small artery occlusion, and other identifiable / unknown causes. All scores were independently assessed by a neurologist. The presence of atrial fibrillation, hypertension, diabetes, dyslipidemia, etc., was diagnosed by the physician according to clinical guidelines.
[0082] In addition, the mechanical thrombectomy procedure was recorded, including the time from patient symptom onset to imaging examination, the time from symptom onset to the start of surgery, the time from imaging examination to the start of surgery, the time from the start of surgery to recanalization, and the surgical method. The modified thrombolysis in cerebral infarction score (mTICI) was used to assess recanalization (0-2a represents poor recanalization, 2b-3 represents good recanalization).
[0083] Grade 0, no blood perfusion, no forward blood perfusion at the distal end of the occlusion;
[0084] Grade 1: There is a small amount of forward blood flow through the occlusion site, but only a little or slow distal blood flow perfusion.
[0085] Grade 2a, occluded vessel supply area with less than 50% forward perfusion blood flow;
[0086] Grade 2b, occluded vessel supply area with forward perfusion flow exceeding 50%;
[0087] Grade 3, the forward perfusion blood flow rapidly covers the entire blood supply area of the occluded vessel.
[0088] Successful recanalization refers to a recanalization status of grade 2b or 3 after surgery.
[0089] (3) Multimodal CT radiomics feature extraction and analysis
[0090] 1) Imaging examination
[0091] All patients suspected of having acute ischemic stroke underwent multimodal one-stop CT examination (head CT and head and neck CTA, CTP) immediately upon admission to obtain CTA and CTP images. The CBV infarct core volume was obtained using iStroke ischemic stroke quantitative analysis software. The infarct core was defined as the area with a relative cerebral blood flow less than 25% of the contralateral side.
[0092] 2) Conventional radiomics information analysis
[0093] A head CT scan was performed to rule out hemorrhage, and the ASPECT score was assessed to evaluate early ischemic changes in the middle cerebral artery supply area. The location of vascular occlusion was evaluated by CTA, and the degree of thrombus burden was evaluated by CTA-thrombus burden score (CBS).
[0094] ①ASPECTS scoring: Ten regions are selected on CT images at the basal ganglia level: caudate nucleus head, lentiform nucleus, internal capsule, insula, precortical region, lateral insular cortex, and posterior insula; and three regions at the level above the basal ganglia, corresponding to the precortical region, lateral insular cortex, and posterior insula, respectively. Each region is worth 1 point, for a total of 10 points. The presence of early signs of ischemia or low-density shadows in any of the ten regions reduces the score by 1 point, with a minimum score of 0.
[0095] ②CBS Score: Unlike the ASPECT score, which focuses on the location of lesions in the brain parenchyma, the CBS score emphasizes the location of thrombi within the blood vessels. More advanced than the ASPECT score, it assigns different weights to different locations. The CBS score uses the extent of embolized vessels in the anterior circulation from CTA, based on a 10-point scoring system. A maximum score of 10 represents the absence of thrombi. When thrombi are present at a corresponding location but the vessel is not visualized on CTA, points are deducted accordingly. For example, if thrombi are present in the supraclinoid segment of the internal carotid artery, the proximal and distal M1 segments of the middle cerebral artery, and visualization is interrupted on CTA, 2 points are deducted for each abnormality. If thrombi are present in the subclinoid segment of the internal carotid artery, the two branches of the M2 segment of the middle cerebral artery, or the A1 segment of the anterior cerebral artery, 1 point is deducted for each abnormality. Figure 2 As shown.
[0096] 3) Lateral omics information analysis
[0097] The rLMC score based on CTA was used to evaluate the establishment of leptomeningeal collaterals before surgery; at the same time, the infarct core volume of CTP (defined as the area with a relative cerebral blood flow of less than 25% of the contralateral side) was obtained through post-processing software to assess the collateral function status.
[0098] The rLMC score evaluates collateral circulation in the anterior circulation supply area, including nine regions: the anterior cerebral artery (ACA) supply area, the lateral fissure area, the basal ganglia area, and the ASPECTS regions (M1-M6). A score of 0 indicates no collateral compensation; 1 indicates lower collateral blood flow perfusion on the affected side than the healthy side; and 2 indicates that collateral blood flow perfusion on the affected side is roughly the same as or higher than the healthy side. The lateral fissure area is scored using 0, 2, and 4 points. The total score is 20 points; lower scores indicate poorer collateral circulation. Figure 3 As shown.
[0099] It should be noted that all imaging feature assessments were conducted independently by a neurologist and a neuroradiologist, and any disputes over the results were discussed by both parties to reach a consensus.
[0100] Furthermore, this embodiment uses conventional radiomics and collateral omics data based on one-stop multimodal CT, combined with relevant clinical data, to screen for risk factors of poor prognosis in elderly patients undergoing acute mechanical thrombectomy, and establishes a risk assessment system based on the Nomogram model, mainly including the following steps:
[0101] (1) Risk factors for SICH and poor prognosis in elderly patients undergoing mechanical thrombectomy
[0102] Univariate logistic regression analysis was used to preliminarily screen risk factors associated with acute-phase mechanical thrombectomy (SICH) and poor prognosis in elderly patients. The significance level was set at α = 0.05.
[0103] (2) Construction of the regression model
[0104] A regression model predicting adverse prognoses was constructed using multivariate logistic regression analysis. Due to the correlation among variables in this study, the Least Absolute Contraction and Selection Operator (LASSO) algorithm was used to select variables for inclusion in the multivariate logistic regression model. This method considers collinearity among independent variables, and all resulting variables are significantly correlated with the dependent variable. Radiomic features and clinical indicators highly correlated with the risk of adverse prognoses from mechanical thrombectomy, identified through univariate analysis, were included in the LASSO regression model. The significance level α was used for the regression. 进 =0.1, α 出 =0.15.
[0105] (3) Establishment of the Nomogram risk scoring system
[0106] A Nomogram model was built to predict surgical risks in elderly patients based on the variables selected by LASSO. The constructed risk prediction model was then visualized. Figure 4As shown, each scale of the independent predictors in the Nomogram model corresponds to a score above the Nomogram image. All scores are then summed to obtain the total score, which corresponds to the predicted probability of a poor prognosis for the patient below the Nomogram image. Furthermore, a calibration curve is used to evaluate the consistency between the predictive power of the constructed risk model and its actual diagnostic power.
[0107] (4) Evaluation of the Nomogram risk scoring system
[0108] 1) Discrimination: The discrimination of the risk prediction model is evaluated using receiver operating characteristic (ROC) and area under the ROC curve (AUC). The sensitivity, specificity, accuracy, positive predictive value, and negative predictive value are also given as indicators to comprehensively reflect the discrimination of the model and evaluate the performance of the prediction model.
[0109] 2) Calibration degree: Plot the calibration curve of the risk prediction model and perform HL goodness-of-fit test to evaluate the calibration degree of the model.
[0110] 3) Clinical suitability: Draw DCA curves to evaluate the clinical application value of the risk prediction model.
[0111] (5) Validation and optimization of the prediction model
[0112] The validation set dataset was fed into the best prediction model for internal validation and optimization.
[0113] The above studies aim to identify the risk factors for poor prognosis of mechanical thrombectomy in elderly patients during the acute phase, and to establish a Nomogram risk prediction model based on this.
[0114] Furthermore, this embodiment also verifies the generalization ability and clinical feasibility of the Nomogram risk scoring system through a prospective cohort study, mainly including: (1) selecting study subjects and grouping them; (2) collecting clinical, conventional radiomics, and collateral information; and (3) external validation of the Nomogram risk prediction model. Steps (1) and (2) are described above and will not be repeated here. Step (3) specifically involves:
[0115] The prospective validation set dataset was used to perform external validation, correction and optimization of the best Nomogram risk scoring system model to clarify the generalization ability and clinical feasibility of the prediction model.
[0116] The above research aims to verify the generalization ability and clinical feasibility of the Nomogram risk prediction model, and ultimately establish a risk assessment system for mechanical thrombectomy in elderly patients with acute anterior circulation large vessel occlusion, and promote the technology in Chongqing and Southwest China.
[0117] As described above, this embodiment, based on one-stop multimodal CT information, will analyze relevant collateral omics information, conventional radiomics information, and clinical data to screen risk factors affecting the poor prognosis of mechanical thrombectomy in the acute phase in elderly patients. A Nomogram-based risk prediction model will be constructed, and the generalization ability and clinical feasibility of the model will be verified through a prospective cohort study. Ultimately, a risk assessment system for mechanical thrombectomy in elderly patients with acute anterior circulation large vessel occlusion will be established, providing more evidence-based basis for clinicians to make preoperative decisions and enabling individualized and precise diagnosis and treatment for elderly patients with ischemic stroke.
[0118] Based on the same inventive concept, embodiments of the present invention also provide a method for predicting the risks of mechanical thrombectomy surgery for elderly patients. For example... Figure 5 As shown, the method includes:
[0119] S1. Establish a clinical database of elderly patients with AIS-aLVO.
[0120] The clinical database stores CT radiomics data and collateral omics data of elderly AIS-aLVO patients.
[0121] Step S1 specifically includes:
[0122] The study subjects were selected and grouped to obtain the first result; the study subjects were elderly patients who were confirmed by multimodal CT examination to have acute anterior circulation large vessel occlusive ischemic stroke, underwent mechanical thrombectomy, and were regularly followed up after the operation; the first result includes the imaging data and grouping of the study subjects;
[0123] Clinical information of all enrolled patients was collected and evaluated to obtain a second result;
[0124] Multimodal CT radiomics features of all enrolled patients were extracted and analyzed to obtain a third result; the third result includes conventional radiomics information analysis results and collateral radiomics information analysis results.
[0125] A clinical database of elderly AIS-aLVO patients was constructed based on the first result, the second result, and the third structure.
[0126] Furthermore, the research subjects were selected and grouped as follows:
[0127] Study subjects were selected based on inclusion and exclusion criteria. Inclusion criteria included: ① Age ≥ 80 years; ② All patients arrived at the hospital within 24 hours of symptom onset or the last known normality, and underwent a one-stop multimodal CT scan; ③ CTA showed occlusion of the internal carotid artery and / or the M1 segment of the middle cerebral artery; ④ Mechanical thrombectomy was performed within 24 hours of symptom onset or the last known normality; ⑤ Informed consent was obtained from the patient or their guardian. Exclusion criteria included: ① Intracranial hemorrhage on cranial imaging upon admission; ② Transient ischemic attack (TIA); ③ Intravenous thrombolysis prior to mechanical thrombectomy; ④ A history of cancer or discovery of cancer during hospitalization, hematologic disorders, or autoimmune diseases; ⑤ Incomplete clinical diagnosis and treatment history and medical records; ⑥ Imaging quality not meeting diagnostic requirements.
[0128] The study subjects were divided into a SICH group and a non-SICH group based on whether intracranial hemorrhage occurred after mechanical thrombectomy; and into a good prognosis group and a poor prognosis group based on whether adverse outcomes occurred 3 months after mechanical thrombectomy.
[0129] For each group of patients, they were randomly divided into two subgroups using computer random numbers: 70% were assigned to the training group and 30% to the validation group.
[0130] Furthermore, clinical information of all enrolled patients was collected and evaluated, specifically as follows:
[0131] All enrolled patients had relevant clinical data collected, including age, gender, medical history, medication history, smoking history, pre-admission mRS score, blood pressure, blood glucose, blood lipids, laboratory tests, NIHSS score, and cerebral infarction type.
[0132] Record the details of the mechanical thrombectomy procedure, including the time from the onset of the patient's illness to the imaging examination, the time from the onset of the illness to the start of the procedure, the time from the imaging examination to the start of the procedure, the time from the start of the procedure to the recanalization of the blood vessel, and the procedure method.
[0133] A modified thrombolysis grading system for cerebral infarction was used to assess vascular recanalization, with 0-2a representing poor recanalization and 2b-3 representing good recanalization.
[0134] Furthermore, multimodal CT radiomics features of all enrolled patients were extracted and analyzed, specifically:
[0135] Acquire preoperative CT images of all patients and extract the imaging features of the CT images;
[0136] ASPECT scores were obtained by plain CT scans to evaluate early ischemic changes in the middle cerebral artery supply area, and CTA was used to evaluate the location of vascular occlusion and the degree of thrombus burden.
[0137] The preoperative establishment of leptomeningeal collaterals was evaluated using the rLMC score based on CTA, and the infarct core volume of CBV was obtained through CTP to assess the functional status of leptomeningeal collaterals.
[0138] S2. Based on the CT radiomics data and collateral omics data, combined with clinical data, risk factors for poor prognosis of mechanical thrombectomy in elderly patients during the acute phase are screened.
[0139] Step S2 specifically includes:
[0140] Univariate analysis was used to analyze relevant clinical information, routine imaging information, and collateral omics information to preliminarily screen risk factors associated with acute-phase mechanical thrombectomy (SICH) and poor prognosis in elderly patients.
[0141] S3, Construct a Nomogram risk prediction model based on the risk factors.
[0142] In this embodiment, the minimum absolute contraction and selection operator algorithm is used to incorporate meaningful variables from the univariate analysis into the multivariate logistic regression model to construct a Nomogram risk prediction model.
[0143] S4 visualizes the constructed Nomogram risk prediction model.
[0144] S5 evaluates, validates, and optimizes the Nomogram risk prediction model.
[0145] In this embodiment, model evaluation includes: assessing the performance of the Nomogram risk prediction model using receiver operating characteristics and area under the ROC curve; plotting the calibration curve of the risk prediction model and performing the HL goodness-of-fit test to evaluate the calibration degree of the Nomogram risk prediction model; and plotting the DCA curve to evaluate the clinical application value of the Nomogram risk prediction model.
[0146] Model Validation and Optimization: The Nomogram risk prediction model was validated and optimized using the validation group in the grouping scenario.
[0147] S6. The Nomogram risk prediction model was used to predict the risk of mechanical thrombectomy in elderly patients with AIS-aLVO.
[0148] It should be noted that for more detailed procedures of each step in the method embodiment, please refer to the aforementioned system embodiment section, which will not be repeated here.
[0149] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A risk prediction system for mechanical thrombectomy surgery in elderly patients, characterized in that, include: The database creation module is used to create a clinical database of elderly AIS-aLVO patients; the clinical database stores CT radiomics data and collateral omics data of elderly AIS-aLVO patients. The risk factor screening module is used to screen risk factors for poor prognosis of acute mechanical thrombectomy in elderly patients based on the CT radiomics data and collateral omics data, combined with clinical data. The model building module is used to build a Nomogram risk prediction model based on the risk factors. The thrombectomy risk prediction module is used to predict the risk of mechanical thrombectomy in elderly AIS-aLVO patients using the Nomogram risk prediction model.
2. The system as described in claim 1, characterized in that, The database creation module is specifically used for: The study subjects were selected and grouped to obtain the first result; the study subjects were elderly patients who were confirmed by multimodal CT examination to have acute anterior circulation large vessel occlusive ischemic stroke, underwent mechanical thrombectomy, and were regularly followed up after the operation; the first result includes the imaging data and grouping of the study subjects; Clinical information of all enrolled patients was collected and evaluated to obtain a second result; Multimodal CT radiomics features of all enrolled patients were extracted and analyzed to obtain a third result; the third result includes conventional radiomics information analysis results and collateral radiomics information analysis results. A clinical database of elderly AIS-aLVO patients was constructed based on the first result, the second result, and the third structure.
3. The system as described in claim 2, characterized in that, The research subjects were selected and grouped as follows: Study subjects were selected based on inclusion and exclusion criteria. Inclusion criteria included: ① Age ≥ 80 years; ② All patients arrived at the hospital within 24 hours of symptom onset or the last known normality, and underwent a one-stop multimodal CT scan; ③ CTA showed occlusion of the internal carotid artery and / or the M1 segment of the middle cerebral artery; ④ Mechanical thrombectomy was performed within 24 hours of symptom onset or the last known normality; ⑤ Informed consent was obtained from the patient or their guardian. Exclusion criteria included: ① Intracranial hemorrhage on cranial imaging upon admission; ② Transient ischemic attack (TIA); ③ Intravenous thrombolysis prior to mechanical thrombectomy; ④ A history of cancer or discovery of cancer during hospitalization, hematologic disorders, or autoimmune diseases; ⑤ Incomplete clinical diagnosis and treatment history and medical records; ⑥ Imaging quality not meeting diagnostic requirements. The study subjects were divided into a SICH group and a non-SICH group based on whether intracranial hemorrhage occurred after mechanical thrombectomy; and into a good prognosis group and a poor prognosis group based on whether adverse outcomes occurred 3 months after mechanical thrombectomy. For each group of patients, they were randomly divided into two subgroups using computer random numbers: 70% were assigned to the training group and 30% to the validation group.
4. The system as described in claim 2, characterized in that, Clinical information of all enrolled patients was collected and evaluated, specifically as follows: All enrolled patients had relevant clinical data collected, including age, gender, medical history, medication history, smoking history, pre-admission mRS score, blood pressure, blood glucose, blood lipids, laboratory tests, NIHSS score, and cerebral infarction type. Record the details of the mechanical thrombectomy procedure, including the time from the onset of the patient's illness to the imaging examination, the time from the onset of the illness to the start of the procedure, the time from the imaging examination to the start of the procedure, the time from the start of the procedure to the recanalization of the blood vessel, and the procedure method. A modified thrombolysis grading system for cerebral infarction was used to assess vascular recanalization, with 0-2a representing poor recanalization and 2b-3 representing good recanalization.
5. The system as described in claim 2, characterized in that, Multimodal CT radiomics features were extracted and analyzed from all enrolled patients, specifically: Acquire preoperative CT images of all patients and extract the imaging features of the CT images; ASPECT scores were obtained by plain CT scans to evaluate early ischemic changes in the middle cerebral artery supply area, and CTA was used to evaluate the location of vascular occlusion and the degree of thrombus burden. The preoperative establishment of leptomeningeal collaterals was evaluated using the rLMC score based on CTA, and the infarct core volume of CBV was obtained through CTP to assess the functional status of leptomeningeal collaterals.
6. The system as described in claim 2, characterized in that, Risk factors for poor prognosis in elderly patients undergoing acute mechanical thrombectomy were identified, specifically: Univariate analysis was used to analyze relevant clinical information, routine imaging information, and collateral omics information to preliminarily screen risk factors associated with acute-phase mechanical thrombectomy (SICH) and poor prognosis in elderly patients.
7. The system as described in claim 6, characterized in that, Based on the aforementioned risk factors, a Nomogram risk prediction model is constructed as follows: The minimum absolute contraction and selection operator algorithm is used to incorporate meaningful variables from univariate analysis into a multivariate logistic regression model to construct a Nomogram risk prediction model.
8. The system as described in claim 7, characterized in that, The system also includes a model processing module for: Model demonstration: The constructed Nomogram risk prediction model is visualized; Model evaluation: The performance of the Nomogram risk prediction model was evaluated using receiver operating characteristics and area under the ROC curve; the calibration curve of the risk prediction model was plotted and the HL goodness-of-fit test was performed to evaluate the calibration of the Nomogram risk prediction model; the DCA curve was plotted to evaluate the clinical application value of the Nomogram risk prediction model. Model Validation and Optimization: The Nomogram risk prediction model was validated and optimized using the validation group in the grouping scenario.
9. A method for predicting the risk of mechanical thrombectomy surgery in elderly patients, characterized in that, include: A clinical database of elderly patients with AIS-aLVO was established; the clinical database contains CT radiomics data and collateral omics data of elderly patients with AIS-aLVO. Based on the CT radiomics data and collateral omics data, combined with clinical data, risk factors for poor prognosis of mechanical thrombectomy in elderly patients during the acute phase were screened. Construct a Nomogram risk prediction model based on the aforementioned risk factors; The Nomogram risk prediction model was used to predict the risk of mechanical thrombectomy in elderly patients with AIS-aLVO.
10. The method as described in claim 9, characterized in that, After constructing the Nomogram risk prediction model based on the risk factors, the method further includes: The constructed Nomogram risk prediction model is visualized. The performance of the Nomogram risk prediction model was evaluated using receiver operating characteristics and area under the ROC curve; the calibration curve of the risk prediction model was plotted and the HL goodness-of-fit test was performed to evaluate the calibration of the Nomogram risk prediction model; and the DCA curve was plotted to assess the clinical application value of the Nomogram risk prediction model. The Nomogram risk prediction model was validated and optimized using the validation group in the grouping scenario.