A Mobile Health Application to Detect Absence Seizures using Hyperventilation and Eye-Movement Recordings

摘要 Eysz公司正在开发一个用于诊断和监测的移动的健康（mHealth）应用程序和算法 失神性癫痫准确诊断和监测癫痫发作和治疗效果至关重要 有效治疗癫痫的要素。不幸的是，失神发作是出了名的难以识别， 导致诊断延迟和难以监测治疗。诊断缺席的黄金标准 癫痫发作是视频脑电图（VEEG），但这种方法是昂贵的，限于临床设置，并很难 access.监测失神癫痫的金标准是患者自我报告的数据，研究表明， 显示超过50%的不准确。其他远程监测策略，如动态脑电图，缺乏 VEEG的敏感性和特异性，并能增加癫痫患者的污名经历。已经 自20世纪90年代以来，没有批准用于失神癫痫的新疗法，部分原因是难以测量 结果。因此，迫切需要一种用于失神发作的远程诊断/监测工具。Eysz 因此，计划开发一个mHealth应用程序，使用（1）自愿引导过度通气（HV），（2）眼睛 运动和面部生物识别数据，以监测癫痫患者的癫痫易感性和治疗反应， 失神发作自愿HV触发>90%的失神癫痫患者癫痫发作，是一种标准 临床工具，以协助诊断和监测失神癫痫。HV也被证明是安全的， 有效时，每天进行激活癫痫发作，从而缩短VEEG监测会话。 因此，HV提供了一个很有前途的工具，用于在家里监测癫痫发作活动的背景下。艾斯群岛 开发利用眼球运动数据检测癫痫发作的软件和算法， 癫痫发作Eysz建议将基于视频的眼睛跟踪（和面部生物识别跟踪）的使用扩展到 基于智能手机的应用程序，包括软件引导的HV。第一阶段的建议侧重于初步测试 我们基于智能手机的工具，用于引导HV和视频数据收集。该项目的具体目标是：1） 从接受HV与VEEG同时进行的受试者收集眼动和面部生物特征数据; 2） 评估算法验证的新“黄金标准”指标的潜力，以促进移动健康的发展 3）开发机器学习（ML）算法，从眼睛检测癫痫发作 追踪和面部生物识别数据Eysz的目标是证明>75%的灵敏度检测癫痫发作>7秒， 持续时间，为未来评估应用程序的家庭使用和算法准确性提供了坚实的基础 在一个更大的患者群体中。

Abstract Eysz, Inc. is developing a mobile health (mHealth) application and algorithms for diagnosing and monitoring absence epilepsy remotely. Accurate diagnosis and monitoring of seizures and therapeutic effects are critical elements of effective epilepsy treatment. Unfortunately, absence seizures are notoriously difficult to identify, leading to diagnostic delay and difficulty monitoring treatments. The gold standard for diagnosing absence seizures is video EEG (VEEG), but this method is expensive, limited to clinical settings, and can be hard to access. The gold standard for monitoring absence epilepsy is patient self-reported data, which studies have shown to be more than 50% inaccurate. Other strategies for remote monitoring, such as ambulatory EEG, lack the sensitivity and specificity of VEEG, and can add to the stigma people with epilepsy experience. There have been no new therapy approvals for absence epilepsy since the 1990s, in part due to the difficulty of measuring outcomes. Thus, there is a critical need for a remote diagnostic/monitoring tool for absence seizures. Eysz therefore plans to develop an mHealth app that uses (1) voluntary guided hyperventilation (HV), with (2) eye movement and facial biometric data to monitor seizure susceptibility and treatment responses in people with absence seizures. Voluntary HV triggers seizures in >90% of people with absence epilepsy and is a standard clinical tool to assist in diagnosing and monitoring absence epilepsy. HV has also been shown to be safe and effective when performed on a daily basis to activate seizures and thereby shorten VEEG monitoring sessions. Thus, HV offers a promising tool for use in the context of at-home monitoring of seizure activity. Eysz is developing software and algorithms for detecting seizures using eye movement data, starting with absence seizures. Eysz proposes to extend the use of video-based eye-tracking (and facial biometric tracking) to a smartphone-based application that includes software-guided HV. This Phase I proposal focuses on initial testing of our smartphone-based tool for guided HV and video data collection. The Specific Aims of this project are: 1) Collect eye-movement and facial biometric data from subjects undergoing HV concurrently with VEEG; 2) Evaluate the potential for a new “gold standard” metric for algorithm validation to enable mHealth development in the home environment; and 3) Develop machine learning (ML) algorithms that detect seizures from eye tracking and facial biometrics data. Eysz aims to demonstrate >75% sensitivity for detection of seizures >7 s in duration, providing a strong foundation for future evaluation of at-home use of the app and algorithm accuracy in a larger cohort of patients.