EEG Biomarkers Derived from Dynamical Network Models Enable Rapid Paths to Accurate Diagnosis and Effective Treatment of Epilepsy

摘要 癫痫是一种神经系统疾病，其特征是突然反复发作异常的电活动。 大脑，即所谓的癫痫。这种疾病困扰着全球6000多万人，背负着同样的负担 女性罹患乳腺癌，男性罹患肺癌。癫痫患者的一线治疗 是抗癫痫药物(AEDs)。如果AEDs在抑制癫痫发作方面无效，那么患者可能会考虑 替代治疗包括手术切除大脑中的致痫区域，脑电波 刺激，或迷走神经刺激。有几种治疗方法可供选择，人们可能会认为癫痫是 在控制之下。然而，事实远非如此。准确诊断癫痫，进而找到有效的治疗方法 治疗可能需要数年甚至一生，在此期间，患者和家人遭受癫痫的耻辱，侧面- 无效的AEDs的影响，住院时间长且昂贵，不可逆转的外科治疗结果不佳， 和/或不太令人满意的刺激疗法，其疗效在生理上是无法测量的。我们 提出一项计划，以建立新的脑电生物标记物和计算工具，使快速 而准确的癫痫诊断紧随其后的是一条快速有效的治疗途径。这样的一个节目 需要在癫痫皮质网络如何表现和变化的概念性知识方面取得重大进展 刺激治疗将从脑电的动态网络建模(DNM)中收集。有很多 随着临床工作流程的开始，癫痫的诊断和治疗面临着挑战 病人的第一次癫痫发作。首先，癫痫的准确诊断可能需要几个月到几年的时间，头皮脑电图 可以被用来确认诊断。然而，黄金标准是寻找符合以下条件的脑电异常 癫痫的指标(例如，棘波)，通常没有捕捉到或误读。其次，这需要几个月到几年的时间 寻找有效的AED治疗方法，因为目前尚无药效的生理学衡量标准。对于这两个痛点， 我们将利用我们实验室从颅内脑电中发现的一种新的生物标记物，称为源-汇度量 它旨在捕获始终仅在癫痫患者中存在的病理网络属性。 对于30%的患者来说，AEDs不起作用，他们的替代治疗包括手术治疗 致癫痫区(EZ)和电刺激治疗。然而，耐药手术的成功率 患者平均为50%，目前还没有神经刺激治疗的有效性衡量标准，剩下一半 接受治疗的患者中无反应的患者。对于这些耐药患者，我们将利用源汇指数， 源自DNMS和EEG，以帮助更准确地定位EZ，以提高手术成功率，并 从FDA批准的RNS设备跟踪刺激治疗的疗效。拟议的R35将针对主要 来自动态网络模型的新型脑电生物标记物的挑战。如果成功，该计划将 引领突破性技术，使所有癫痫都能得到准确的诊断和最佳治疗 患者更快(从几年到几周)，包括服务不足的耐药队列。

SUMMARY Epilepsy is a neurological disorder that is marked by sudden recurrent episodes of abnormal electrical activity in the brain, known as seizures. This disease plagues more than 60 million people globally, with the same burden of disease as breast cancer in women and lung cancer in men. First line of treatment for patients with epilepsy are anti-epileptic drugs (AEDs). If AEDs are not effective in suppressing seizures, then patients may consider alternative treatments including surgical resection of the epileptogenic zone in the brain, electrical brain stimulation, or vagus nerve stimulation. With several treatment options available, one may think that epilepsy is under control. However, this is far from true. Accurately diagnosing epilepsy and then finding an effective treatment can take years to a lifetime, during which patients and families suffer from the stigma of epilepsy, side- effects of ineffective AEDs, extensive and costly hospital stays, poor outcomes of irreversible surgical treatment, and/or less than satisfactory stimulation therapies whose efficacies are physiologically unmeasurable. We propose a program to establish novel EEG biomarkers and computational tools that will enable rapid and accurate diagnosis of epilepsy followed by a rapid path to an effective treatment. Such a program entails major advances in conceptual knowledge of how epileptic cortical networks behave and change during stimulation treatment that will be gleaned from dynamic network modeling (DNM) of EEG. There are many challenges with diagnosing and treating epilepsy that unfolds as one considers the clinical workflow beginning with a patient’s first seizure. First an accurate diagnosis of epilepsy can take months to years, where scalp EEG can be leveraged to confirm diagnosis. However, the gold standard is to look for EEG abnormalities that are indicators of epilepsy (e.g., spikes), which are often not captured or misread. Second, it takes months to years to find effective AED treatment as there is no physiological measure of drug efficacy. For these two pain points, we will leverage a new biomarker that our lab discovered from intracranial EEG called the source-sink metric which is designed to capture pathological network properties that are always present only in epilepsy patients. For 30% of the patient population, no AEDs work, and their alternative treatments include surgical treatment of the epileptogenic zone (EZ) and electrical stimulation therapy. However surgical success rates for drug resistant patients averages 50%, and there is currently no measure of efficacy of neurostimulation treatment, leaving half of treated patients nonresponsive. For these drug resistant patients, we will leverage the source-sink index, derived from DNMs and EEG, to help more accurately localize the EZ to improve surgical success rates, and to track efficacy of stimulation treatment from the FDA approved RNS device. The proposed R35 will address major challenges with novel EEG biomarkers stemming from dynamic network models. Is successful, the program will lead to breakthrough technologies enabling getting to accurate diagnosis and optimal treatment for all epilepsy patients more rapidly (from years to weeks), including the underserved drug-resistant cohort.