This paper introduces an innovative framework designed for progressive (granular in time to onset) prediction of seizures through the utilization of a Deep Learning (DL) methodology based on non-invasive multimodal sensor networks. Epilepsy, a debilitating neurological condition, affects an estimated 65 million individuals globally, with a substantial proportion facing drug-resistant epilepsy despite pharmacolog-ical interventions. To address this challenge, we advocate for predictive systems that provide timely alerts to individuals at risk, enabling them to take precautionary actions. Our framework employs advanced DL techniques and uses personalized data from a network of non-invasive electroencephalogram (EEG) and electrocardiogram (ECG) sensors, thereby enhancing prediction accuracy. The algorithms are optimized for real-time processing on edge devices, mitigating privacy concerns and minimizing data transmission overhead inherent in cloud-based solutions, ultimately preserving battery energy. Additionally, our system predicts the countdown time to seizures (with 15-minute intervals up to an hour prior to the onset), offering critical lead time for preventive actions. Our multimodal model achieves 95% sensitivity, 98% specificity, and 97% accuracy, averaged among 29 patients.

This paper reports the design, modeling, and control of an MR-compatible actuation unit comprising pneumatic stepper mechanisms. One helix-shaped bellows and one toroid-shaped bellows were designed to actuate in pure rotation and pure translation, respectively. The actuation unit is a two degree- of-freedom needle driver that translates and rotates the base of one tube of a steerable needle like a concentric tube robot. For safety, mechanical stops limit needle motion to maximum unplanned step sizes of 0.5 mm and 0.5 degrees. Additively manufactured by selective laser sintering, the flexible fluidic actuating (FFA) mechanism achieves two degree-of-freedom motion as a monolithic, compact, and hermetically-sealed device. A second novel contribution is sub-step control for precise translations and rotations less than full step increments; steady- state errors of 0.013 mm and 0.018 degrees were achieved. The linear FFA produced peak forces of 33 N and -26.5 N for needle insertion and retraction, respectively. The rotary FFA produced bidirectional peak torques of 68 N-mm. With the FFA's in full motion in a 3T scanner, no loss in signal-to-noise ratio of MR images observed.

Concentric tube robots can enable new clinical interventions if they are able to pass through soft tissue, deploy along desired paths through open cavities, or travel along winding lumens. These behaviors require the robot to deploy in such a way that the curved shape of its shaft remains unchanged as the tip progresses forward (i.e., "follow-the-leader" deployment). Follow-the-leader deployment is challenging for concentric tube robots due to elastic (and particularly torsional) coupling between the tubes that form the robot. However, as we show in this paper, follow-the-leader deployment is possible, provided that tube precurvatures and deployment sequences are appropriately selected. We begin by defining follow-the-leader deployment and providing conditions that must be satisfied for a concentric tube robot to achieve it. We then examine several useful special cases of follow-the-leader deployment, showing that both circular and helical precurvatures can be employed, and provide an experimental illustration of the helical case. We also explore approximate follow-the-leader behavior and provide a metric for the similarity of a general deployment to a follow-the-leader deployment. Finally, we consider access to the hippocampus in the brain to treat epilepsy, as a motivating clinical example for follow-the-leader deployment.