PFI-RP: Smart Seizure Prediction System based on AI-enabled Implantable Sensor Networks

该创新-研究伙伴关系（PFI-RP）项目的更广泛影响/商业潜力是为抗癫痫药物无效的神经系统疾病患者提供更好的生活质量。 虽然该技术将专注于耐药性癫痫的新型治疗方法，但许多其他医疗应用，包括治疗心脏起搏器，帕金森病或创伤后应激障碍（PTSD）患者可能是可能的。根据疾病控制中心（CDC）的数据，美国人口的1.2%（约340万人）患有活动性癫痫，全球有超过6500万人受到影响。虽然癫痫患者已从植入式医疗器械（IMD）中受益，但目前美国食品药品监督管理局（FDA）批准的用于缓解耐药性癫痫患者癫痫发作强度的神经刺激器均未提供预防性治疗或提前、准确、预测性警报。该项目可以使医疗专业人员准确预测癫痫发作，以便患者采取预防措施，避免严重伤害或死亡等不良后果。拟议的项目将通过嵌入式人工智能（AI）算法为患有抗癫痫药物无法治疗的神经系统疾病的患者提供现场智能医疗推理。 这种治疗可以消除对皮下布线和高速通过身体的无线链路的需要，提供更安全和更节能的解决方案。目前，商业神经刺激器系统使用来自皮下发生器的有线引线将电信号发送到目标刺激部位。 这种接线是术后并发症的主要原因。此外，这些系统中的大多数都是开环的，边缘计算能力有限。它们需要大量的手动临床医生输入来调整它们的响应参数，并且一些仍然使用连续刺激，因为它们不能预测癫痫发作。该项目可以通过开发具有嵌入式原位AI处理的智能和无线供电植入物来改进深部脑和其他神经刺激技术。这种发展可以消除预测模型对外部云计算的依赖性，这需要将稳定的数据流从可植入传感器传输到外部设备。通过低功率超声波，系统将增加无线连接、可充电性和可重新编程性。这项创新可能消除电极和植入物之间的皮下布线以及大型植入电池的需要。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact/commercial potential of this Partnerships for Innovation - Research Partnerships (PFI-RP) project is to provide a better quality life for patients affected by neurological disorders for whom antiepilepsy drugs are not effective. While the technology will focus on novel treatments for drug-resistant epilepsy, a number of additional medical applications, including the treatment of patients with cardiac pacemakers, Parkinson disease, or post-traumatic stress disorder (PTSD) could be possible. According to the Centers for Disease Control (CDC), 1.2% of the US population (about 3.4 million people) has active epilepsy with more than 65 million people affected globally. While epilepsy patients have benefited from Implantable Medical Devices (IMDs), none of the current Food and Drug Administration (FDA)-approved neurostimulators that are used to alleviate seizure intensity in patients with drug-resistant epilepsy offer preventive treatment or well-in-advance, accurate, predictive alerts. This project may enable medical professionals to accurately predict a seizure so that patients could take precautions and avoid adverse outcomes such as serious injuries or death.The proposed project will enable in-situ smart medical inference via embedded Artificial Intelligence (AI) algorithms for patients affected by neurological disorders not treatable with antiepilepsy madications. The treatment may eliminate the need for subcutaneous wiring and highspeed through-body wireless links, offering a safer and more energy efficient solution. Currently, commercial neurostimulator systems use wired leads from a subcutaneous generator to send electrical signals to the targeted stimulation site. This wiring is a major cause of post-operative complications. Additionally, most of these systems are open-loop and have limited edge-computing capabilities. They require significant manual clinician input to adjust their response parameters and some still use continuous stimulation since they cannot predict a seizure. This project may improve deep brain and other neurostimulation technologies by developing intelligent and wirelessly-powered implants with embedded, in-situ, AI processing. This development may remove the dependency of the prediction models on external cloud computing, which requires transmission of a steady stream of data from implantable sensors to external devices. Wireless connectivity, re-chargeability, and re-programmability will be added to the system through low-power ultrasonic waves. This innovation may eliminate the needs for subcutaneous wiring between electrodes and implants and for large, implanted batteries.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.