Micro-magnetic stimulation is a promising technique for the treatment of neurological conditions, allowing long-term implant stability and effective neuronal regulation. Despite this potential, further research is essential to fully understand its mechanisms. The earthworm, with its primitive nervous system comprising two linear giant fibers, the medial giant fiber (MGF) and the lateral giant fiber (LGF), is proposed as a suitable model for investigating magnetic stimulation. The goal of this thesis is to compare tactile, electrical, and magnetic stimulation in an earthworm model. Electrical stimulation triggered both giant fibers a single time and also a third longer spike of unknown origin. In contrast, tactile stimulation selectively triggered one type of fiber depending on stimulation site. The fiber was usually triggered multiple times and the third spike was absent. Several action potentials from the MGF were recorded after magnetic stimulation. The stimulus artifact blocked the first 10 ms of recording and therefore a direct relation between stimulus and stimulation could not be established. The stimulus was 1.4 μs long and reached 2.45 kV/m. The peak electric field required to achieve stimulation for a pulse of that length was estimated at 22.6 ± 10.5 kV/m. Magnetic stimulation did not reach this value and therefore it is unclear whether this action potentials were triggered by the magnetic pulse. In tactile stimulation, crosstalk is negligible and in electrical stimulation, it becomes a concern only when the distance is less than 2 cm. In contrast, magnetic stimulation presents significant crosstalk issues, which can be effectively mitigated by increasing the distance between the stimulation and recording sites. These findings suggest that the earthworm is an appropriate model for future research exploring the phenomenon of magnetic stimulation.

This report presents the design process of a project aimed at the automatic recognition of 3D structures formed by magnetic spheres in a turbulent water-filled cylinder. This field of research holds promise for future technologies, as macroscopic self-assembly might be the key to three-dimensional storage on chips. With the macroscopic setup, the microscopic self-assembly process is imitated. To automatically recognise a 3D structure, this report is divided into three subgroups that research the optimal test setup, develop an image processing program and create a deep learning model. A labelled result of the 3D formation will be outputted, all while limiting the data size, computation time and inaccuracies. The subgroup responsible for the setup and the underlying physics produces images of the magnetic spheres forming a structure in the test setup. The Image Processing subgroup extracts the properties of the spheres from the image. Finally, the subteam for deep learning, in combination with data management, gives the extracted properties as input to a neural network model, which determines the structure of the spheres. Each submodule has demonstrated successful functionality on its own. However, due to time constraints, a fully integrated system with high accuracy has not been achieved yet. Future work will involve expanding the dataset to enhance the robustness of the recognition algorithms.

Intense precipitation can have extensive economic outcomes, from disrupting outdoor activities to causing severe infrastructural damage, such as landslides, and endangering public safety. The urgency to mitigate these impacts underscores the need for improved early warning systems. Enhanced short-term weather prediction, or nowcasting, is critical for addressing these severe weather events effectively. Traditional meteorological forecasting methods, while foundational, are often constrained by simplistic physical assumptions and fail to capture the complex, nonlinear patterns of intense weather events. These methods also struggle with high computational demands and lack the resolution needed to detect crucial microscale atmospheric phenomena for accurate short-term forecasts.
To address these challenges, this research introduces a novel deep learning approach utilizing a streamlined architecture that combines a Vector Quantized Variational Autoencoder (VQVAE) and an Autoregressive (AR) Transformer. This model aims to predict weather conditions up to 180 minutes ahead, using data analyzed at 30-minute intervals. The proposed model displays comparable performance with the state-of-theart conventional methods and other deep learning nowcasting models in predicting precipitations and sometimes extreme events. This study seeks to enhance forecasting accuracy and efficiency, providing valuable contributions to the field of meteorological nowcasting.

Implant-associated infections are a severe concern affecting 1−3 % of patients undergoing primary joint arthroplasty [1]. Magnetic hyperthermia, which can be non-invasively induced by applying an alternating magnetic field (AMF) around a metallic implant, is a new approach with great potential. This research project explores the application of non-contact induction heating (IH) of metallic implants in orthopedic surgery. It focuses on the heating behavior of paramagnetic biomaterials relevant to this application in alternating magnetic fields.
Empirical testing plays an inherent role in research. However, it is often very time-consuming, especially when assessing many specimens. With in silico simulations, data can be yielded much faster. An analytical model simulating the heating dynamics of paramagnetic materials in the AMF was constructed to provide a tool for faster suitability assessment of metallic biomaterials for magnetic hyperthermia. The model was then compared with empirical data obtained in a controlled environment.

The empirical data showed a quadratically increasing temperature rise with the field amplitude and a linearly growing temperature rise with the frequency. The constructed analytical model confirmed that the heat generated in the materials increases quadratically with the increasing AMF amplitude. Additionally, increasing the frequency of the AMF also affects heat generation. The model predicted different trends depending on what domain a material is assigned to. However, empirical data indicates a consistent linear relationship across several assessed frequencies and materials.

Based on the obtained knowledge of the material's behavior in AMF, a new hip implant intended for magnetic hyperthermia treatment was designed. It utilized a coating on the neck of the femoral stem, designed to enable targeted IH, particularly to regions most susceptible to bacterial infections. The experiment featured a multi-material specimen consisting of Ti Gr. 23 and ST. 37, which mimicked the neck of a hip implant, the proof of concept was successfully demonstrated, showing a significant temperature increase for specimens with coating compared to the ones without at high magnetic field strength and frequency. However, targeted IH was not detected.

Better computational models can further explore the obtained knowledge, and the proof-of-concept can be further tested to investigate its contribution to treating resistant bacterial biofilms.

The device that is designed and described in this thesis aims to non-invasively stimulate neural activity in an earthworm at multiple, specific locations. This is done by inducing an electric field from two inductors with a specific electric field strength in mind. The response of the earthworm is recorded and amplified to enable analysis of this signal. Initially, requirements are set that need to be met by this device. From these requirements a design process was started and simulations were made. After this, a device was created on which the requirements could be evaluated. Due to the physical trials not being completed the requirements could not be fully verified.

In the context of prosthetic devices, the absence of natural sensory feedback poses a significant challenge, leading to limited proprioceptive sensation and reduced control over artificial limbs. This study explores the potential of establishing a magnetic connection between prosthetic devices and muscles to enhance proprioceptive feedback and facilitate more intuitive control of prosthetic limbs. The primary objective is to ascertain the safe range of forces that can be transmitted magnetically from a prosthesis to a muscle, with ethical considerations guiding the allocation of muscle resources.
Preliminary experiments conducted using Silicone Ecoflex 0030, chosen for its stress behaviour similarity to natural muscle tissue, serve as the cornerstone for this research. A novel methodology is introduced to predict the forces generated when magnets snap, particularly when one magnet is embedded within an elastic medium like silicone, replicating the force distribution characteristics of muscle tissue. The theoretical model guiding these predictions is based on empirical data collected from experiments that investigate the magnetic force interactions between magnets and the distinct properties of silicone materials. Validation of the theoretical model for predicting the snapping point is achieved through a comprehensive series of final experiments.
This study demonstrates that the monopole model, tailored for magnets with parallel-aligned poles, yields highly accurate predictions when compared to empirically measured data concerning magnet force modelling. Additionally, within silicone, the force-displacement relationship exhibits linearity, with stiffness primarily contingent on the length of the implanted object. Force transmission experiments involving a magnet embedded in silicone reveal that magnets snap at forces of less than 3 Newtons.
A significant outcome of this research is the development of a validated methodology for predicting the force when a pair of magnet become unstable and snap. This methodology is applicable to scenarios involving magnets embedded in elastic media like silicone and holds the potential for extending predictions to snap forces in muscle tissue. The implications of this study are promising, suggesting the feasibility of employing magnets as sensors to detect muscle intent and as activators to provide feedback by mobilizing implanted magnets, thereby stimulating muscle spindles. These findings open new avenues for enhancing proprioception and control in prosthetic devices.

The objective of this project was to develop a real-time restlessness detection algorithm for Rapid Eye Movement Sleep Behavior Disorder (RBD). This project was completed in collaboration with Momo Medical, a fast-growing start-up, and developer of the BedSense device. Our project aimed to design a system capable of detecting RBD episodes and bringing patients to a lighter sleep stage. To accomplish this, proprietary data and existing data from an RBD patient were collected through the BedSense device. Additionally, data was extracted using a hardware system (developed by another subgroup) from a non-RBD test subject. The collected data was utilized to design and implement a restlessness detection algorithm. Given the limited data collected for this project, a classical detection algorithm was developed instead of machine learning techniques. The software system demonstrated impressive performance, achieving a balanced accuracy of 95.94%, an average sensitivity of 100%, and an average specificity of 91.88%.
The results of this project were integrated with another project seeking to develop a sensor system that brings an RBD patient to a lighter sleep stage. This project built a sock system with embedded sensors and a vibration module to achieve this. The proof of concept of the final integrated system demonstrated a non-invasive method of detecting and preventing episodes caused by RBD.