PhD Thesis Defense

https://caltech.zoom.us/j/83727307317 Meeting ID: 837 2730 7317

Advancements in medical devices are pivotal for enhancing patient care and assisting physicians in the management of chronic conditions like epilepsy and spinal cord injuries. Our work explores the integration of artificial intelligence (AI) and machine learning (ML) into medical devices, focusing on applications such as brain-machine interfaces (BMIs) for spinal cord rehabilitation, seizure detection, and the classification of cardiac arrhythmias. We have developed a dynamic recurrent neural network (DRNN) decoder and a Feature Extraction Network (FENet) to enhance the functionality of BMIs, aiming to reduce power and memory usage while maintaining consistent performance across various users. In the area of cardiac health, we introduced EKGNet, a convolutional network that combines analog computing with deep learning for the detection of arrhythmias, addressing the need for real-time processing efficiency. Furthermore, we employed an XGboost-based method for seizure detection, achieving fast and precise diagnostics with reduced energy demands. These efforts demonstrate how AI and ML can potentially contribute to addressing challenges in medical data analysis and assist in the development of medical devices that are more dependable, efficient, and versatile.