Differential DNA methylation patterns can serve as biomarkers for allergic diseases such as pediatric asthma and rhinitis, but age-dependent variability in epigenetic profiles undermines the reliability of predictive models. This thesis addresses that challenge by introducing a graph-based deep learning approach for \textbf{age-independent} allergy prediction from DNA methylation data. Each subject’s DNA methylation profile is represented as an individualized graph constructed via an extended Weighted Gene Co-expression Network Analysis (WGCNA) that captures global co-methylation structure and subject-specific patterns, thus balancing population-level relationships with individual epigenetic heterogeneity. Edges between CpG sites are assigned weights using a Gaussian kernel on methylation values, ensuring the graph reflects personalized similarity while maintaining biologically meaningful connections. A Graph Neural Network (GNN) with an Edge Convolution (EdgeConv) architecture is then trained on these subject-specific graphs to predict allergy outcomes. We evaluated this framework on DNA methylation data from three harmonized pediatric cohorts (PIAMA, MAKI, COPSAC) processed with the MEFFIL pipeline for cross-cohort normalization and quality control. An Epigenome-Wide Association Study (EWAS) identified key CpG features associated with asthma, rhinitis and IgE, which were used to guide feature selection for model training. Our graph-based model outperformed conventional methods like ElasticNet and XGBoost in certain cohorts and maintained robust predictive accuracy between the ages of 6 and 16, demonstrating a certain resilience to age-related methylation differences. Furthermore, we applied gradient-based saliency analysis to the trained GNN to highlight influential methylation features, providing interpretability and revealing plausible epigenetic markers of allergy. The proposed pipeline is scalable and interpretable, and its ability to deliver reliable, age-invariant risk predictions from early-life epigenetic data underscores its potential clinical utility for early allergy diagnostics in children.

This research is conducted in the framework of a Master’s thesis within the Marine Engineering department of the Maritime Transport and Technology branch and in joint collaboration with the Robotics branch of the Mechanical Engineering Faculty of Tu Delft.
The precise control of robotic systems in granular underwater environments is essential for applications such as deep-sea mining, sediment sampling, and seabed infrastructure maintenance. In such environments, the interaction between the robot and deformable substrates like sand and clay plays a crucial role in operational efficiency and system stability. This research investigates how fine-tuning joint stiffness in an impedance controller influences penetration depth accuracy, horizontal force distribution, and force consistency along a trajectory mapped using 3D camera point cloud data. Understanding these relationships is critical for optimizing force control strategies in unstructured and dynamic underwater settings.
Experiments were conducted using a KUKA iiwa 7 robotic arm equipped with an impedance controller, following a mapped trajectory over a real sandbed in both dry and submerged conditions. The point cloud data from a 3D camera provided accurate environmental mapping, ensuring precise trajectory tracking. The results indicate a significant correlation between joint stiffness and penetration accuracy: higher stiffness improved depth accuracy and reduced external disturbances but compromised adaptability in cases where the robot encountered hard obstacles. Conversely, lower stiffness increased compliance, allowing for smoother interactions but at the cost of greater sensitivity to force fluctuations.
Fluid damping in submerged conditions was found to reduce penetration error variability, highlighting the stabilizing influence of water on force interactions. The study also revealed that current robotic systems for deep-sea applications differ significantly from the 7-degree-of-freedom (DOF) KUKA arm used in this research. In practical scenarios, deep-sea mining robots typically feature a single actuated DOF (pitch), with other degrees of freedom facilitated by passive flexibility rather than active control. These structural differences influence force distribution and overall system behavior, emphasizing the need for future studies tailored to real-world deep-sea mining configurations.
Future research should extend beyond sand to softer seabed sediments such as clay, which behaves more like a Bingham fluid and exhibits significantly lower shear strength—potentially by a factor of 5 to 10 compared to sand. Additionally, exploring adaptive stiffness strategies that dynamically adjust control parameters in real time could enhance the efficiency and robustness of underwater robotic systems. These advancements would contribute to optimizing force control for precise, adaptable interactions in unstructured marine environments.

Shoulder injuries, prevalent worldwide, often occur from ageing and accidents. In Western countries, these injuries primarily afflict the elderly population, while in rural regions of Bangladesh, Iran, India, and Pakistan, they affect younger individuals who are often the family's primary earners. Due to that, preventing and aiding the recovery of shoulder injuries is crucial. To address this, strain maps with vibrotactile feedback, emerge as a promising solution. However, the feedback system must be affordable, compact, comfortable, user-friendly, easily understood, and portable to suit the local environment. Vibrotactile feedback appears promising but can distract the user from work. Hence, this study seeks to investigate if vibrotactile feedback can be paired with strain maps to guide users in maintaining healthy postures and reducing the risk of shoulder injuries in rural areas, where visual feedback is used as a benchmark. To provide feedback using strain maps, shoulder angles are determined using Python's OpenCV and MediaPipe libraries. PyGame is utilized to display the strain maps, and OpenCV helps delineate boundaries between regions of high and low strain within the shoulder. Visual feedback is integrated into the strain map display, while vibrotactile feedback is delivered through a wearable haptic device. Despite challenges related to axial rotation accuracy and the camera-dependent nature of shoulder angle measurements, user experiments, conducted independently for shoulder elevation and planar elevation, reveal that vibrotactile feedback shows better performance compared to visual feedback. Consequently, this study concludes that vibrotactile feedback has the potential to prevent shoulder injuries with strain maps, but also still needs to improve for future work.

In this work, we propose a method for monitoring and management of rotator-cuff tendon strains in human-robot collaborative physical therapy for rotator cuff rehabilitation. The proposed approach integrates a complex offline biomechanical model with a collaborative, industrial robot arm and an impedance controller. The model is used for computing rotator-cuff tendon strain as a function of human shoulder configuration, muscle activation and external forces. This subject- and injury-specific data is stored in \textit{strain maps}, which represent the relationship between the strains and shoulder DoFs. In our previous work, we implemented strain maps to preplan minimal strain, safe trajectories using two shoulder DoFs, and used the corresponding robot-mediated movement for passive trajectory following for healthy subjects. This work expands on that by implementing two novel functionalities: 1) patient-led movement, and 2) adding the third shoulder DoF and the corresponding control complexities, while still controlling for safe rotator-cuff tendon strains. For patient-led movement, we precomputed unsafe zones for each strain map by clustering and fitting ellipses to the clusters. These unsafe areas with increased risk of (re-)injury are then used to set the impedance control parameters and reference pose for real-time biomechanical safety control. By linearly interpolating between strain maps, smooth and safe movement of the third shoulder DoF is added. The resulting robot control torques guide the patient away from unsafe, high strain shoulder poses in real-time during patient-led movement. The proposed method has the potential to improve the safety, Range of Motion, and muscle activity that the patients receive through robot-mediated physical therapy. The main advantage of this approach is that the patient is free to use and explore their full shoulder RoM, while the robot controls and manages biomechanical safety in real-time. To validate the proposed method, we performed two experiments showcasing two novel functionalities, and a third experiment as proof-of-concept displaying the full method.