Walczak S, Nowack W J
University of Colorado at Denver, College of Business and Administration, Campus Box 165, P.O. Box 173364, Denver, Colorado 80217-3364, USA.
J Med Syst. 2001 Feb;25(1):9-20. doi: 10.1023/a:1005680114755.
Determining the cause of seizures is a significant medical problem, as misdiagnosis can result in increased morbidity and even mortality of patients. The reported research evaluates the efficacy of using an artificial neural network (ANN) for determining epileptic seizure occurrences for patients with lateralized bursts of theta (LBT) EEGs. Training and test cases are acquired from examining records of 1,500 consecutive adult seizure patients. The small resulting pool of 92 patients with LBT EEGs requires using a jack-knife procedure for developing the ANN categorization models. The ANNs are evaluated for accuracy, specificity, and sensitivity on classification of each patient into the correct two-group categorization: epileptic seizure or non-epileptic seizure. The original ANN model using eight variables produces a categorization accuracy of 62%. Following a modified factor analysis, an ANN model utilizing just four of the original variables achieves a categorization accuracy of 68%.
确定癫痫发作的病因是一个重大的医学问题，因为误诊会导致患者发病率增加甚至死亡。所报道的研究评估了使用人工神经网络（ANN）来确定具有偏侧化θ波爆发（LBT）脑电图的患者癫痫发作情况的有效性。训练和测试病例来自对1500名连续成年癫痫患者的检查记录。最终得到的92名具有LBT脑电图的患者样本量较小，需要使用留一法来开发ANN分类模型。对ANN在将每个患者正确分类为癫痫发作或非癫痫发作这两组分类中的准确性、特异性和敏感性进行评估。使用八个变量的原始ANN模型的分类准确率为62%。经过改进的因子分析后，仅使用四个原始变量的ANN模型的分类准确率达到了68%。
