STTR Phase I: Patient-Specific System for Early Detection and Identification of Epileptic Seizures

这个小企业技术转让（STTR）第一阶段项目的更广泛的影响/商业潜力是为癫痫患者及其护理人员提供一个智能系统，可以在癫痫发作之前预测癫痫发作。 美国有超过300万成年人和100万儿童，全世界有超过5000万人患有癫痫。 反复和不可预测的癫痫发作严重影响癫痫患者的生活质量。这些癫痫发作仍然是癫痫患者及其护理人员经济，情感和身体伤害的主要原因。设计、开发和集成人工智能（AI）模型与检测脑波异常（如脑电图（EEG））的仪器，以实时预测癫痫发作，可能会为这些患者及其护理人员带来改善。这项技术有望在美国快速增长的60亿美元人工智能医疗解决方案市场中占据一部分份额。这项小企业技术转让（STTR）第一阶段项目支持开发一种新型消费产品，与护理人员合作，主动降低癫痫患者癫痫发作事件的风险。目前的商业解决方案大多是被动的，只有在癫痫发作事件发生后才能提供支持。该公司将通过开发、测试、集成和评估应用于EEG数据的机器学习（ML）模型来填补这一空白，以预测癫痫发作。科学方法将利用固有的异质性和复杂的边缘技术。与第三方供应商EEG帽、微控制器、智能手机和云服务的数据连接依赖于许多不同的操作技术和通信标准。这项研究将通过硬件和软件解决方案来克服这些挑战，这些解决方案将这些服务集成到边缘设备中，以实现应用程序的可移植性并简化部署。诸如对有限计算能力和能源设备的推断及其对预测准确性/灵敏度的影响等挑战将通过强大的交叉验证技术、广泛的测试和使用社区标准的基准测试来解决。这项研究的技术产品将促进护理人员的知识和增加对癫痫发作的理解，以及增加病人的福祉。该奖项反映了NSF的法定使命，并已被认为是值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估的支持。

The broader impact/commercial potential of this Small Business Technology Transfer (STTR) Phase I project is to provide epileptic patients, and their caregivers a smart system that can predict seizures before they occur. There are more than 3 million adults and 1 million children in the US, and more than 50 million people worldwide, suffering from epilepsy. Repeated and unpredictable seizures significantly affect the quality of life of people suffering from epilepsy. These seizures remain the leading cause of economic, emotional, and physical injuries for people with epilepsy and their caregivers. Design, development, and integration of artificial intelligence (AI) models with instruments that detect abnormalities in brain waves like electroencephalogram (EEG) for real-time seizure prediction may bring improvements for these patients and their caregivers. This technology is poised to capture a portion of the rapidly growing $6 billion US market of AI healthcare solutions.This Small Business Technology Transfer (STTR) Phase I project supports the development of a novel consumer product that works with caregivers to proactively mitigate the risk of seizure events in people with epilepsy. Current commercial solutions are mostly reactive, and support is available only after a seizure event. The company will fill this gap by developing, testing, integrating, and evaluating machine learning (ML) models - applied to EEG data - for epileptic seizure prediction. The scientific approach will leverage inherently heterogenous and complex edge technologies. Data connectivity with third party vendor EEG caps, microcontrollers, smart phones, and cloud services rely on many different operational technologies and communication standards. This research will overcome these challenges with hardware and software solutions that will integrate these services within an edge device to enable application portability and simplify deployment. Challenges such as inference on limited computational power and energy devices, and its effects on the accuracy/sensitivity of the predictions will be solved using robust cross-validation techniques, extensive testing, and benchmarking using community standards. The technical product of this research will advance caregiver knowledge and increase understanding of epileptic seizures as well as increase patient well-being.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.