% 0期刊文章%一个亚伦了% Gamaleldin奥斯曼% Nishi Rampal %一个悉达多Biswal % Legros本杰明%劳伦斯·赫希% m .布兰登·威斯多佛%尼古拉斯% T加斯帕德多态生存分析癫痫预测连续脑电图(P2.229) % D J神经病学2017% % P P2.229 X % V % 88% N 16补充目的:确定风险因素和时间依赖性连续脑电图的癫痫发作的风险。首页背景:之前连续脑电图分析危险因素预测回顾和不受控制的审查/主题退出的影响。改正这一缺点,我们使用多态生存分析连续665年连续脑电图记录。设计/方法:回顾性分析665年预期获得数据库连续连续脑电图会议与相关的临床因素(> 24小时)和脑电图数据包括事件的时间。Elastic-net逻辑回归是用来确定预测风险因素的时间独立变量,随后使用Cox比例风险模型。与时间有关的变量被用来创建一个多态生存模型与三个州(输入、风险状态和扣押)。定义的风险状态的出现癫痫样的模式:单侧性的周期放电(lpd),两国独立定期排放(BIPDs),简短的有节奏的排放(惟妙惟肖),单侧性的有节奏的三角洲活动(LRDA),和/或零星的痫性放电(SED)。引导是用于生成95%的置信区间。结果:电记录的癫痫发生在23%的cEEG监测会议。昏迷时间预测价值最大的独立变量(31%癫痫脑电图;O。R 1.8 p<0.01) and any history of seizures: either remotely or acutely (34% had electrographic seizures; OR 3.0 p<0.001). Four multistate survival models were generated dependent on the time independent variables (coma, history of seizure). The overall 72-hour risk of seizures was between 0.09–0.36 if the subject did not develop epileptiform EEG patterns, and 0.18–0.64 if the subject developed epileptiform patterns. After 6hrs the risk of seizures declined from 0.04–0.16 at 1hour of recording to 0.02–0.09 if no epileptiform EEG patterns developed, and to 0.08–0.34 if they did.Conclusions: The risk of seizures on continuous EEG is dependent on history of seizure and presence of coma. The risk of developing seizures during a continuous EEG decays quickly if no epileptiform EEG patterns emerge.Disclosure: Dr. Struck has nothing to disclose. Dr. Osman has nothing to disclose. Dr. Rampal has nothing to disclose. Dr. Biswal has nothing to disclose. Dr. Legros has nothing to disclose. Dr. Hirsch has received personal compensation for activities with Ceribell, Marinus, Monteris, Neuropace, Sun Pharma, Sunovion, and Upsher-Smith as an speaker. Dr. Hirsch has received research support from Acorda, Eisai, Lundbeck, Sunovion and Upsher-Smith. Dr. Westover has nothing to disclose. Dr. Gaspard has nothing to disclose. %U

map