{"title":"利用基于人工智能的颈动脉斑块形态学对颈动脉狭窄患者的 30 天死亡率和中风风险进行评分。","authors":"","doi":"10.1016/j.avsg.2024.05.016","DOIUrl":null,"url":null,"abstract":"<div><h3>Background</h3><p>The gold standard for determining carotid artery stenosis intervention is based on a combination of percent stenosis and symptomatic status. Few studies have assessed plaque morphology as an additive tool for stroke prediction. Our goal was to create a predictive model and risk score for 30-day stroke and death inclusive of plaque morphology.</p></div><div><h3>Methods</h3><p>Patients with a computed tomographic angiography head/neck between 2010 and 2021 at a single institution and a diagnosis of carotid artery stenosis were included in our analysis. Each computed tomography was used to create a three-dimensional image of carotid plaque based off image recognition software. A stepwise backward regression was used to select variables for inclusion in our prediction models. Model discrimination was assessed with area under the receiver operating characteristic curves (AUCs). Additionally, calibration was performed and the model with the least Akaike Information Criterion (AIC) was selected. The risk score was modeled from the Framingham Study. Primary outcome was mortality/stroke.</p></div><div><h3>Results</h3><p>We created 3 models to predict mortality/stroke from 366 patients: model A using only clinical variables, model B using only plaque morphology and model C using both clinical and plaque morphology variables. Model A used age, sex, peripheral arterial disease, hyperlipidemia, body mass index (BMI), chronic obstructive pulmonary disease (COPD), and history of transient ischemia attack (TIA)/stroke and had an AUC of 0.737 and AIC of 285.4. Model B used perivascular adipose tissue (PVAT) volume, lumen area, calcified volume, and target lesion length and had an AUC of 0.644 and AIC of 304.8. Finally, model C combined both clinical and software variables of age, sex, matrix volume, history of TIA/stroke, BMI, PVAT, lipid rich necrotic core, COPD and hyperlipidemia and had an AUC of 0.759 and an AIC of 277.6. Model C was the most predictive because it had the highest AUC and lowest AIC.</p></div><div><h3>Conclusions</h3><p>Our study demonstrates that combining both clinical factors and plaque morphology creates the best predication of a patient's risk for all-cause mortality or stroke from carotid artery stenosis. Additionally, we found that for patients with even 3 points in our risk score model has a 20% chance of stroke/death. Further prospective studies are needed to validate our findings.</p></div>","PeriodicalId":8061,"journal":{"name":"Annals of vascular surgery","volume":null,"pages":null},"PeriodicalIF":1.4000,"publicationDate":"2024-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"30-Day Risk Score for Mortality and Stroke in Patients with Carotid Artery Stenosis Using Artificial Intelligence Based Carotid Plaque Morphology\",\"authors\":\"\",\"doi\":\"10.1016/j.avsg.2024.05.016\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><h3>Background</h3><p>The gold standard for determining carotid artery stenosis intervention is based on a combination of percent stenosis and symptomatic status. Few studies have assessed plaque morphology as an additive tool for stroke prediction. Our goal was to create a predictive model and risk score for 30-day stroke and death inclusive of plaque morphology.</p></div><div><h3>Methods</h3><p>Patients with a computed tomographic angiography head/neck between 2010 and 2021 at a single institution and a diagnosis of carotid artery stenosis were included in our analysis. Each computed tomography was used to create a three-dimensional image of carotid plaque based off image recognition software. A stepwise backward regression was used to select variables for inclusion in our prediction models. Model discrimination was assessed with area under the receiver operating characteristic curves (AUCs). Additionally, calibration was performed and the model with the least Akaike Information Criterion (AIC) was selected. The risk score was modeled from the Framingham Study. Primary outcome was mortality/stroke.</p></div><div><h3>Results</h3><p>We created 3 models to predict mortality/stroke from 366 patients: model A using only clinical variables, model B using only plaque morphology and model C using both clinical and plaque morphology variables. Model A used age, sex, peripheral arterial disease, hyperlipidemia, body mass index (BMI), chronic obstructive pulmonary disease (COPD), and history of transient ischemia attack (TIA)/stroke and had an AUC of 0.737 and AIC of 285.4. Model B used perivascular adipose tissue (PVAT) volume, lumen area, calcified volume, and target lesion length and had an AUC of 0.644 and AIC of 304.8. Finally, model C combined both clinical and software variables of age, sex, matrix volume, history of TIA/stroke, BMI, PVAT, lipid rich necrotic core, COPD and hyperlipidemia and had an AUC of 0.759 and an AIC of 277.6. Model C was the most predictive because it had the highest AUC and lowest AIC.</p></div><div><h3>Conclusions</h3><p>Our study demonstrates that combining both clinical factors and plaque morphology creates the best predication of a patient's risk for all-cause mortality or stroke from carotid artery stenosis. Additionally, we found that for patients with even 3 points in our risk score model has a 20% chance of stroke/death. Further prospective studies are needed to validate our findings.</p></div>\",\"PeriodicalId\":8061,\"journal\":{\"name\":\"Annals of vascular surgery\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":1.4000,\"publicationDate\":\"2024-07-14\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Annals of vascular surgery\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S0890509624003972\",\"RegionNum\":4,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"PERIPHERAL VASCULAR DISEASE\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Annals of vascular surgery","FirstCategoryId":"3","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0890509624003972","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"PERIPHERAL VASCULAR DISEASE","Score":null,"Total":0}
引用次数: 0
摘要
目的:确定颈动脉狭窄干预的金标准是基于狭窄百分比和症状状态的组合。很少有研究将斑块形态作为中风预测的附加工具进行评估。我们的目标是建立一个包含斑块形态的 30 天中风和死亡预测模型和风险评分:我们的分析包括 2010-2021 年间在一家机构接受头颈部 CT 血管造影检查并确诊为颈动脉狭窄的患者。利用图像识别软件生成颈动脉斑块的三维图像。我们采用逐步回归法选择变量纳入预测模型。通过接收者操作特征曲线(AUC)评估模型的识别率。此外,还进行了校准,并选择了阿凯克信息准则(AIC)最小的模型。风险评分模型来自弗雷明汉研究。主要结果是死亡率/中风:我们建立了三个模型来预测 366 名患者的死亡率/中风情况:模型 A 仅使用临床变量,模型 B 仅使用斑块形态学变量,模型 C 同时使用临床变量和斑块形态学变量。模型 A 使用了年龄、性别、PAD、高脂血症、体重指数、慢性阻塞性肺病和 TIA/ 中风史,AUC 为 0.737,AIC 为 285.4。模型 B 使用了血管周围脂肪组织体积、管腔面积、钙化体积和靶病变长度,其 AUC 为 0.644,AIC 为 304.8。最后,模型 C 结合了年龄、性别、基质体积、TIA/中风史、体重指数、血管周围脂肪组织、富脂坏死核心、慢性阻塞性肺病和高脂血症等临床和软件变量,其 AUC 为 0.759,AIC 为 277.6。模型 C 的 AUC 最高,AIC 最低,因此最具预测性:我们的研究表明,将临床因素和斑块形态结合在一起,能最好地预测颈动脉狭窄患者的全因死亡或中风风险。此外,我们还发现,在我们的风险评分模型中,哪怕只有 3 分的患者也有 20% 的中风/死亡几率。还需要进一步的前瞻性研究来验证我们的发现。
30-Day Risk Score for Mortality and Stroke in Patients with Carotid Artery Stenosis Using Artificial Intelligence Based Carotid Plaque Morphology
Background
The gold standard for determining carotid artery stenosis intervention is based on a combination of percent stenosis and symptomatic status. Few studies have assessed plaque morphology as an additive tool for stroke prediction. Our goal was to create a predictive model and risk score for 30-day stroke and death inclusive of plaque morphology.
Methods
Patients with a computed tomographic angiography head/neck between 2010 and 2021 at a single institution and a diagnosis of carotid artery stenosis were included in our analysis. Each computed tomography was used to create a three-dimensional image of carotid plaque based off image recognition software. A stepwise backward regression was used to select variables for inclusion in our prediction models. Model discrimination was assessed with area under the receiver operating characteristic curves (AUCs). Additionally, calibration was performed and the model with the least Akaike Information Criterion (AIC) was selected. The risk score was modeled from the Framingham Study. Primary outcome was mortality/stroke.
Results
We created 3 models to predict mortality/stroke from 366 patients: model A using only clinical variables, model B using only plaque morphology and model C using both clinical and plaque morphology variables. Model A used age, sex, peripheral arterial disease, hyperlipidemia, body mass index (BMI), chronic obstructive pulmonary disease (COPD), and history of transient ischemia attack (TIA)/stroke and had an AUC of 0.737 and AIC of 285.4. Model B used perivascular adipose tissue (PVAT) volume, lumen area, calcified volume, and target lesion length and had an AUC of 0.644 and AIC of 304.8. Finally, model C combined both clinical and software variables of age, sex, matrix volume, history of TIA/stroke, BMI, PVAT, lipid rich necrotic core, COPD and hyperlipidemia and had an AUC of 0.759 and an AIC of 277.6. Model C was the most predictive because it had the highest AUC and lowest AIC.
Conclusions
Our study demonstrates that combining both clinical factors and plaque morphology creates the best predication of a patient's risk for all-cause mortality or stroke from carotid artery stenosis. Additionally, we found that for patients with even 3 points in our risk score model has a 20% chance of stroke/death. Further prospective studies are needed to validate our findings.
期刊介绍:
Annals of Vascular Surgery, published eight times a year, invites original manuscripts reporting clinical and experimental work in vascular surgery for peer review. Articles may be submitted for the following sections of the journal:
Clinical Research (reports of clinical series, new drug or medical device trials)
Basic Science Research (new investigations, experimental work)
Case Reports (reports on a limited series of patients)
General Reviews (scholarly review of the existing literature on a relevant topic)
Developments in Endovascular and Endoscopic Surgery
Selected Techniques (technical maneuvers)
Historical Notes (interesting vignettes from the early days of vascular surgery)
Editorials/Correspondence