EAGER: Exploring Artificial Intelligence Techniques for Energy-Efficient Arrhythmia Detection and Identification in Connected Implantable Cardiac Devices

点击翻译按钮获取中文摘要

Cardiovascular disease is one of the major causes of all human deaths. Irregular heart rhythms or arrhythmias are one of the most common causes for cardiovascular death. Arrhythmia-related cardiac morbidity and mortality can be reduced by the implantation of permanent cardiac devices that are designed to monitor the heart rhythm for serious abnormalities and to deliver immediate therapy when necessary. In the United States, nearly 225,000 pacemakers or other cardiac rhythm management devices (CRMD) are implanted annually. Device manufacturers have recently started incorporating advanced features to make the cardiac devices smarter and more connected. In the treatment of arrhythmia, the identification of the specific nature of arrhythmia is critical. At present the implantable cardiac devices have the capability only to detect the presence of an arrhythmia not to identify which kind of arrhythmia. Because of this shortcoming in current implantable devices, it is possible that they may administer the wrong corrective measure for the type of arrhythmia. This lack of specific treatment can have devastating consequences. Thus, energy efficient arrhythmia detection and identification is critical to next generation implantable cardiac devices. The results of this research work will have the potential to significantly impact the way patients suffering from arrhythmia are diagnosed and treated. CRMDs are severely energy constrained devices. Any new added features will drain more charge from the battery and will reduce the battery life of the implanted cardiac devices. In this research, the team will explore energy-efficient novel artificial intelligence (AI) techniques for real-time detection and identification of arrhythmia in connected smart implantable cardiac devices. Further, the feasibility of ultra-low power hardware implementation of the developed algorithms will be explored. The main novelty of this research lies in using statistical models and algorithms and in using energy efficient deep neural networks (DNNs) for arrhythmia detection and identification. The developed AI algorithm hardware can be optimized for energy efficiency by reduction of computation size and by reduction of number of computations. The research results are expected to enable the cardiac device manufacturers to develop the next generation of implantable cardiac devices capable of identifying arrythmias.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.