SOFTWARE TOOL FOR SEIZURE ANALYSIS AND PREDICTION

描述(由申请人提供)：缺乏强大的、随时可用的计算工具来分析不规则间隔的数据，如癫痫发作和治疗干预的效果，并通过定量分析智能地优化治疗。为了分析抗癫痫疗法(脑电刺激或BES)的效果，提出了一种统计方法的改进和验证。这一应用结合了一个统计模型，该模型可以利用临床研究期间经过验证的癫痫发作检测算法的自动输出，得出趋势、生物节律对癫痫易感性的影响以及治疗参数变化对癫痫活动的影响，从而改进治疗方法。该方法将在BES临床试验数据库上进行测试，以可能作为分析和优化治疗的软件工具实施。将使用类似的框架开发一种将趋势提取与经验模型相结合的经验癫痫发作预测算法(ESPA)，以测试受试者癫痫发作的可预测性。Espa将在一个广泛的人类长时间序列数据库上进行评估。所提出的方法将确定受试者从特定数据集癫痫发作的可预测性，以及ESPA是否可以在便携式/植入式警报设备中实施。

DESCRIPTION (provided by applicant): There is a dearth of powerful, readily available computational tools for analyzing irregularly spaced data such as seizures and the effect of therapeutic interventions, and for intelligent optimization of therapies through quantitative analysis. The refinement and validation of a statistical methodology for analyzing the effects of an anti-seizure therapy (brain electrical stimulation or BES) are proposed. This application incorporates a statistical model that can elicit trends, influences of biological rhythms on seizure susceptibility and the effects of changes in treatment parameters on epileptic activity using the automated output of a proven seizure detection algorithm over the duration of a clinical study, thus leading to improved therapies. The method will be tested on a database of clinical BES trials for possible implementation as a software tool for analysis and optimization of therapy. An empirical seizure prediction algorithm (eSPA) incorporating trend extraction with empirical models will be developed using a similar framework to test the predictability of a subject's seizures. The eSPA will be evaluated on an extensive human database of long time series. The proposed methods will determine the predictability of a subject's seizures from a certain data set, and whether the eSPA can be implemented in a portable/implantable warning device.