Seizures and the Brain: The Role of Perturbed Dynamic Networks

癫痫的准确诊断、预后和治疗是一个重要的未得到满足的医学需求。癫痫是一种严重的大脑疾病，易感人群反复发作。在生活中的某个时候，它会影响到近1%的英国人口。每年有12.5万人在英国各地的首次癫痫诊所就诊。在这些患者中，40,000人最终被确诊为癫痫，通常在平均延迟一年之后(癫痫联合理事会，2011年9月)。虽然这些确诊病例中的三分之二最终可以通过抗癫痫药物(AED)控制，但大约50%在开始治疗一年后仍然没有得到控制，没有选择AED或剂量的明确指标。治疗反应目前是一种“观察等待”的情况，观察是否会出现进一步的癫痫发作，并相应地调整药物和剂量的选择。对于那些没有反应的人，他们继续癫痫发作，显然是随机的，这导致生活质量和数量的显著下降。在英国，每年有超过1000人死于癫痫的直接后果。我提出了一个令人兴奋的研究计划，其中将开发数学模型和计算机算法，以提高我们对癫痫发作如何发生的理解。这些模型将描述大脑不同区域如何相互作用，以及内部和外部刺激如何影响这些相互作用。大脑区域内的神经活动和区域之间的联系之间的相互作用，至关重要地决定了癫痫是否会发生以及发作的频率。有了这个基本的理解，计算机算法就可以开发出来，直接从癫痫患者的临床记录中告知这些模型的关键参数。这使得这些模型变得个性化，因此适合于为(疑似)癫痫患者解决关键问题：-诊断准确吗？-开出的药物有效吗？-暴露在压力下会增加癫痫发作的可能性吗？-我能知道我的癫痫发作最有可能发生在什么时候吗？-手术能阻止我癫痫发作的发生吗？我将与一名软件工程师和癫痫患者合作，将这种理解转化为解决这些问题的原型工具。例如，开发一种工具，可以帮助多学科临床团队确定大脑癫痫发作的来源，并使用这些信息来更好地计划手术。或者，一款基于智能手机的应用程序，它从无线耳机接收脑电记录，以提供癫痫发作风险的预测。与有生活经验的人和临床医生共同设计和开发这些原型，将最大限度地增加它们导致基础研究影响的可能性。

Accurate diagnosis, prognosis and management of epilepsy is a significant unmet medical need. Epilepsy is a serious brain condition whereby susceptible individuals have recurrent seizures. It affects almost 1% of the UK population at some point in their lives. Every year 125,000 people attend first seizure clinics across the UK. Of these, 40,000 eventually receive a confirmed diagnosis of epilepsy, typically following an average delay of a year (Joint Epilepsy Council, September 2011). Whilst two-thirds of these confirmed cases can ultimately be controlled by anti-epilepsy drugs (AED), approximately 50% remain uncontrolled a year after commencing treatment, with no clear indicators of choice of AED or dose. Treatment response is currently a case of "watchful waiting" to see whether further seizures occur and adjusting choice of medication and dose accordingly. For those who do not respond they continue to have seizures, apparently at random, which leads to a significant reduction in quality and quantity of life. Every year over 1,000 people die in the UK as a direct consequence of epilepsy.I propose an exciting programme of research in which mathematical models and computer algorithms will be developed to improve our understanding of how seizures occur. These models will describe how different regions of the brain interact with each other and how internal and external stimuli can influence these interactions. The interplay between neural activity within brain regions and the connections between regions critically determines whether seizures can occur and how frequently they are likely to happen. Having developed this fundamental understanding, computer algorithms can be developed to inform key parameters of these models directly from clinical recordings collected from people with epilepsy. This makes the models personalised and therefore suitable to address key questions for people with (suspected) epilepsy: - is the diagnosis accurate?- are the drugs being prescribed effective?- will exposure to stress make seizures more likely?- can I know when my seizures are most likely to happen? - will surgery stop my seizures happening? I will work with a software engineer and people with epilepsy to translate this understanding into a prototype tools to address these questions. For example, developing a tool that can aid a multi-disciplinary clinical team in determining where in the brain seizures originate from and using this information to better plan for surgery. Alternatively, a smart-phone based app that receives electrical brain recordings from a wireless headset to provide a forecast of seizure risk. Co-designing and developing these prototypes with people with lived experience and clinicians will maximise the likelihood of their leading to impact of the fundamental research.