Rautaharju P M, Warren J W, Jain U, Wolf H K, Nielsen C L
Circulation. 1981 Aug;64(2):249-56. doi: 10.1161/01.cir.64.2.249.
A multivariate decision-theoretic electrocardiogram (ECG) classification scheme called Cardiac Infarction Injury Score (CIIS) was developed using ECGs of 387 patients with myocardial infarction (MI) and 320 subjects without infarction. The most accurate and stable classification was achieved by using a combination of eight binary (single threshold), three ternary (two thresholds), and four ECG features measured on a continuous scale. For practical visual coding of ECGs, the CIIS coding procedure uses a checklist containing 12 items measured from the conventional 12-lead ECG. The CIIS test results indicate that, in comparison with conventional ECG criteria for MI used in clinical trials, the diagnostic accuracy can be considerably improved by optimizing feature and threshold selection and by multivariate analysis. The CIIS detected MI with a sensitivity of 85% and a specificity of 95%. Using a higher severity level, a specificity of 99% was achieved, with a sensitivity of 71%. One of the primary uses of the CIIS is coding of significant worsening of the ECG with new coronary events from annually recorded ECGs in clinical trials and epidemiologic studies.
一种名为心肌梗死损伤评分（CIIS）的多变量决策理论心电图（ECG）分类方案，是利用387例心肌梗死（MI）患者和320例无梗死受试者的心电图开发的。通过结合八个二元（单阈值）、三个三元（双阈值）以及四个连续测量的心电图特征，实现了最准确和稳定的分类。为了对心电图进行实际的视觉编码，CIIS编码程序使用了一份包含从传统12导联心电图测量的12项内容的检查表。CIIS测试结果表明，与临床试验中使用的传统MI心电图标准相比，通过优化特征和阈值选择以及多变量分析，诊断准确性可得到显著提高。CIIS检测MI的灵敏度为85%，特异性为95%。使用更高的严重程度级别时，特异性达到99%，灵敏度为71%。CIIS的主要用途之一是对临床试验和流行病学研究中每年记录的心电图中新发冠状动脉事件导致的心电图显著恶化进行编码。
