Dodrill C B, Wilkus R J, Ojemann G A, Ward A A, Wyler A R, van Belle G, Tamas L
Ann Neurol. 1986 Jul;20(1):2-12. doi: 10.1002/ana.410200103.
Preoperative variables from a full range of medical specialties were used to predict degree of seizure relief from cortical resection surgery as treatment for epilepsy in 100 patients. General, seizure history, electroencephalographic (EEG), radiological, surgical, and psychological/neuropsychological data were considered. The patients were divided into one large predictive group (n = 75) and a smaller independent cross-validation sample (n = 25). Eight predictive variables emerged: single EEG focus; anterior-midtemporal lobe discharges; discharges only from the side of surgery; rate of occurrence of discharges in surgical area; Wechsler Adult Intelligence Scale Digit Symbol subtest; Marching Test, preferred hand, time; Minnesota Multiphasic Personality Inventory (MMPI) Hysteria scale score; and MMPI Paranoia scale score. By use of multivariate procedures, increased predictability of surgical outcome was obtained not only with the predictive group but with the independent cross-validation sample. The results demonstrate that predictions of seizure relief from epilepsy surgery can be made with 80% accuracy using multiple, rather than single, predictors.
来自各个医学专业的术前变量被用于预测100例癫痫患者接受皮质切除术治疗后癫痫发作缓解的程度。考虑了一般情况、癫痫病史、脑电图（EEG）、放射学、手术及心理/神经心理学数据。患者被分为一个较大的预测组（n = 75）和一个较小的独立交叉验证样本（n = 25）。出现了八个预测变量：单个EEG病灶；颞叶中前部放电；仅手术侧放电；手术区域放电发生率；韦氏成人智力量表数字符号分测验；行进测试、优势手、时间；明尼苏达多相人格调查表（MMPI）癔症量表得分；以及MMPI偏执量表得分。通过使用多变量程序，不仅在预测组，而且在独立交叉验证样本中，手术结果的可预测性都得到了提高。结果表明，使用多个而非单个预测指标，可以80%的准确率预测癫痫手术的发作缓解情况。
