Biosensor Data Fusion for Real-Time Monitoring of Global Neurophysiological Function

生物传感器数据融合实时监测整体神经生理功能

• 批准号: 1719388
• 负责人: Prahalada Rao
• 金额: $ 21.3万
• 依托单位: University of Nebraska-Lincoln
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-10-21 至 2020-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1719388&HistoricalAwards=false
• 关键词: Biosensor; Data; Fusion; Real; Time

Real-time detection of acute changes in neurophysiological state, such as epileptic seizures, lapses in cognitive ability, acute stress, etc., can ultimately serve to prevent accidents in high-risk occupations that require unwavering focus. Such professions include hazardous cargo trucking, heavy machinery operation, security and defense, air traffic control, etc. Indeed, technology for acquiring rich biosensor data streams that capture brain function, e.g., electroencephalography, are becoming increasingly portable and noninvasive. These developments present an opportunity for implementing not only real-time monitoring, but also providing pre-emptive alerts (e.g., smart phone displays), which can be used to indicate degradation in physiological states. This research has direct applications in biomedical settings - for instance, epilepsy, is one of the most common neurological disorders afflicting over 50 million people worldwide, including 3 million people in the U.S. In about 25 percent of these patients, epileptic seizures are not controlled using available medications. Being able to detect (or predict) the onset of epileptic seizures would significantly enhance the patient's quality of life. In a proof-of-concept study, the novel analytical approaches by the research team detected the onset of epileptic seizures within 2.5 seconds. In contrast, existing approaches have a detection delay exceeding 7 seconds. From a broader perspective, the findings of this research can transform the status quo in real-time monitoring of neurophysiological function. The multidisciplinary research team will strive to provide state-of-the-art research and training opportunities for a diverse group of students that bridges the gap from engineering to the life and brain sciences. The research team will develop a sensor data fusion approach based on graph theoretic topological mapping to combine data acquired from multiple biosensors for neurophysiological change point detection. Unlike existing approaches, which rely on complex signal pre-processing, the graph theoretic approach eschews these computationally demanding steps and is therefore more viable in a practical setting. The research team will exploit this framework using a data library of high-resolution neurophysiological recordings acquired from end users in realistic settings that induce shifts in global functional states (e.g., acute stress, cognitive exhaustion, and fatigue and so on). The research team will integrate automated decision-making approaches in the overall schema to synthesize the information and provide easily interpretable feedback to the end user (e.g., displays on a smart device). Furthermore, the PIs will customize biosensors to accommodate the patient's lifestyle.

实时检测神经生理状态的急性变化，例如癫痫发作、认知能力下降、急性应激等，最终可以预防需要高度集中注意力的高风险职业发生事故。这些行业包括危险货物卡车运输、重型机械操作、安全和防御、空中交通管制等。事实上，获取丰富的生物传感器数据流以捕获大脑功能（例如脑电图）的技术正变得越来越便携和非侵入性。这些发展不仅为实施实时监测提供了机会，而且还提供了先发制人的警报（例如智能手机显示屏），可用于指示生理状态的退化。这项研究在生物医学领域有直接应用 - 例如，癫痫是最常见的神经系统疾病之一，全世界有超过 5000 万人受其影响，其中美国有 300 万人。其中约 25% 的患者无法通过使用药物来控制癫痫发作。可用的药物。能够检测（或预测）癫痫发作将显着提高患者的生活质量。在一项概念验证研究中，研究小组采用的新颖分析方法在 2.5 秒内检测到癫痫发作。相比之下，现有方法的检测延迟超过 7 秒。从更广泛的角度来看，这项研究的结果可以改变神经生理功能实时监测的现状。多学科研究团队将努力为多元化的学生群体提供最先进的研究和培训机会，弥合工程与生命和脑科学之间的差距。研究团队将开发一种基于图论拓扑映射的传感器数据融合方法，结合从多个生物传感器获取的数据进行神经生理变化点检测。与依赖于复杂信号预处理的现有方法不同，图论方法避开了这些计算要求较高的步骤，因此在实际环境中更可行。研究团队将利用从最终用户在现实环境中获取的高分辨率神经生理学记录数据库来开发该框架，这些记录会引起全局功能状态的变化（例如急性压力、认知衰竭和疲劳等）。研究团队将在整体架构中集成自动化决策方法，以综合信息并向最终用户提供易于解释的反馈（例如，在智能设备上显示）。此外，PI 将定制生物传感器以适应患者的生活方式。

项目成果

The role of low-level image features in the affective categorization of rapidly presented scenes

低级图像特征在快速呈现场景的情感分类中的作用

• DOI: 10.1371/journal.pone.0215975
• 发表时间: 2019-05
• 期刊: PLOS ONE
• 影响因子: 3.7
• 作者: Rhodes, L. Jack;Ríos, Matthew;Williams, Jacob;Quiñones, Gonzalo;Rao, Prahalada K.;Miskovic, Vladimir;D'Mello, Sidney
• 通讯作者: D'Mello, Sidney

Paired Trial Classification: A Novel Deep Learning Technique for MVPA

配对试验分类：一种新颖的 MVPA 深度学习技术

• DOI: 10.3389/fnins.2020.00417
• 发表时间: 2020-04
• 期刊: Frontiers in Neuroscience
• 影响因子: 4.3
• 作者: Williams, Jacob M.;Samal, Ashok;Rao, Prahalada K.;Johnson, Matthew R.
• 通讯作者: Johnson, Matthew R.