基于机器学习的算法可在接受冠状动脉钙化积分的患者中排除阻塞性冠状动脉疾病
研究纳入435例低至中度冠状动脉疾病(CAD)概率并接受临床指征冠状动脉钙化积分(CACS)及冠状动脉CT血管造影的患者,构建梯度提升机(GBM)模型预测阻塞性CAD,并在126例连续患者中验证。GBM模型在435例患者中分层10折交叉验证敏感度100±0%、特异度69.8±3.6%;在126例患者中真阴性73例、假阴性0例、真阳性20例、假阳性33例,阴性预测值100%、阳性预测值38%。
要点
- GBM模型在435例交叉验证中敏感度100±0%、特异度69.8±3.6%。
- 在126例验证组中假阴性0例、真阴性73例,阴性预测值100%。
- 验证组真阳性20例、假阳性33例,阳性预测值38%。
诊断标准与关键数值
- 在435例患者中分层10折交叉验证,GBM模型敏感度100±0%、特异度69.8±3.6%。
- 在126例患者中临床应用混淆矩阵:真阴性73例、假阴性0例、真阳性20例、假阳性33例。
- GBM模型阴性预测值100%、阳性预测值38%。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
RATIONALE AND OBJECTIVES: An application of artificial intelligence to screen for obstructive coronary artery disease (CAD) after coronary artery calcium scoring (CACS) test.
MATERIALS AND METHODS: As an initial step we analyzed a group of 435 patients (23% male, mean age 61 ± 10) with low to moderate probability of CAD, who underwent clinically indicated CACS and coronary computed tomography angiography. Based on those data we elaborated a gradient boosting machine (GBM) model for prediction of obstructive CAD. Later the model was evaluated on a control group of 126 consecutive patients (31% male, mean age 59 ± 10).
RESULTS: Stratified 10-fold cross-validation performed on the group of 435 patients demonstrated the GBM model's sensitivity at 100 ± 0% and specificity at 69.8 ± 3.6%, while the outcomes (confusion matrix) of a clinical application on the group of 126 patients were: 73 true negative, 0 false negative, 20 true positive, and 33 false positive.
CONCLUSION: The GBM algorithm showcased a considerably high discriminatory power for excluding the presence of obstructive CAD, with negative predictive value and positive predictive value of 100% and 38%, respectively.
引用
- Głowacki J, Krysiński M, Czaja-Ziółkowska M, et al. Machine Learning-based Algorithm Enables the Exclusion of Obstructive Coronary Artery Disease in the Patients Who Underwent Coronary Artery Calcium Scoring. Acad Radiol. 2019. PMID: 31839566. doi:10.1016/j.acra.2019.11.016
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。