SUURPh funds cohorts of 8 doctoral fellows, who work with advisors in Oslo and at UCSD on projects in key fields of computational biology and medicine. The project topics range from computational electrophysiology and neuroscience to mechanics and fluid dynamics, and involve leading experts in these fields. These projects are described below.
Current SUURPh projects:
Improving machine learning techniques for predicting adaptive immune receptor:epitope recognition
This project seeks to develop and apply machine learning approaches to map how the genetic sequence of immune receptors determines their ability to target specific antigens on diseased cells or pathogens. This work will involve utilising large scale sequence databases with validated immune receptor-epitope interactions to define structure-blind methods for assessing how immune receptor sequence determines the range of epitopes that offer likely targets.
PhD candidate: Charlotte Würtzen completed her Masters work in Bioinformatics and Systems biology at the Technical University of Denmark. In a similar vein to the project she is pursuing with SUURPh, there she applied graph neural networks for predicting structural relationships between T-cell immune receptors and peptide major histocompatibility complex binding.
Using personalized cardiac models to predict patient risk to arrhythmias and optimal therapy in patients with congenital heart disease
The goal of this project is to extend the use of population-based personalized forward heart models to stratify risk treatment in congenital heart disease (CHD). The models take into account the unique structural and electrophysiological characteristics of the population group by utilizing imaging and clinical data from CHD patients in the Cardiac Atlas Project and the FORCE registry. This will involve GPU acceleration and other high performance computing techniques (including optimized finite element solvers) to create atlases of population-specific arrhythmia risk and predict outcomes for future patients.
PhD candidate: Alessandro Gatti studied Mathematics and Statistics for Life and Social Sciences at the University of Trento, Italy. There he modeled biophysical properties of myelination in neurons.
Decoding Human Memory and Perception through Machine Intelligence
As part of a collaboration between the Departments of Informatics and Cognitive Neurophysiology at UiO and the Voytek lab at UCSD, this project will apply machine learning to decompose recordings of brain activity into those signals most relevant for learning and memory. This will yield models that we hope will be capable of predicting the semantic category, success of memory storage and retrieval, and certain pathological characteristics from similar recordings.
PhD candidate: Amir Arfan is a data scientist with a background in developing machine learning and AI methods. He received his Masters from the Norwegian University of Life Sciences where he sought to develop alternative methods with performance comparable to convolutional neural networks.
Neurocognitive Mapping: Exploring Hippocampal Navigation and Memory with Neural Sampling
This PhD project will use Neural Sampling (NS) to explore the hippocampus's role in navigation and memory. By simulating hippocampal neural dynamics, the NS supports generative predictive coding in specialized neurons like place and grid cells during navigational tasks. Leveraging statistical and machine learning techniques, estimates of synaptic interactions and signal transmission from rat-brain measurements will be used to validate the model. This work aims to enrich our understanding of navigation and memory and offer applications in artificial intelligence systems.
PhD Candidate: Nicolai Haug is a computational neuroscientist, with a Masters specialty in imaging and biomedical computing from the Norwegian University of Life Sciences. His research to date has focused on building simulation-based inference for identifying parameters of mechanistic models of neural dynamics.
Mechanochemical interplay underlying human performance: computational modelling of skeletal muscle signaling and mechanics in response to exercise
This project seeks to define a new multiscale model of mechano-chemical coupling in skeletal muscle, specifically for the purpose of capturing and predicting the response of muscle to different forms of exercise-induced stress. This involves developing detailed ordinary differential equation models of electro-chamical signaling in muscle cells, and coupling these with models of larger scale deformation and mechanics of the intact muscle fiber.
PhD candidate: Ingvild Devold holds a Masters in Physics and Applied Mathematics from the Norwegian University of Science and Technology. She has worked in developing hybrid graph-based applications for partial differential equations in computational geosciences and is now applying those skills to biology in this interdisciplinary SUURPh project.
Computational models of growth and remodelling in Pulmonary Hypertension
This PhD project attempts to develop a framework for modelling the right ventricular hypertrophy and functional remodelling that contribute to compensated systolic and diastolic function, during pulmonary hypertension. His work mainly involves multi-scale computational modelling using data from murine experiments to validate the model-predicted myocardial growth and remodelling.
PhD Candidate: Oscar Odeigah has a background in mechanical engineering and engineering design from Ahmadu Bello University, Nigeria, and University of Tromsø, Norway. For his master thesis, he used patient-specific bi-ventricular models to investigate the differences (and similarities) between the active stress and active strain methods of mathematically modelling the active contraction of cardiac muscle. His interest in cardiovascular biomechanics and patient-specific modelling, led him to join the SUURPh program, where he will have the opportunity to continue pursuing these interests, in a highly stimulating research environment.
Using Personalized Cardiac Models to Predict Patient Risk for Arrhythmias
The goal of the PhD project is to extend the use of current personalized heart models to address the need for risk stratification and treatment planning in patients with congenital heart disease (CHD). To achieve this goal Lisa will utilize images and clinical data that were collected as part of a large longitudinal study of CHF patients. Lisa will use MRI imaging data to create structural models, personalize the electrophysiology using recorded ECGs, and then perform simulations to predict risk to arrhythmia. The simulations can then be used to predict the optimal treatment for each patient (surgery, ablation, ICD implantation, etc.).
PhD Candidate: Lisa Pankewitz. Lisa received her M. Sc. Biochemistry with a focus on Biomedicine from Leipzig University in Germany. During her thesis she worked for 6 months in the laboratory of Jens Meiler in the center for structural biology at Vanderbilt University on developing a novel approach for membrane protein structure determination. By joining the SUURPH program Lisa can use her interdisciplinary background in biophysics, biochemistry and computation to develop tools for biomedical applications.
BrainCircus - From Brain Circuit Models to Brain Insight
As part of an ongoing collaboration with the Allen Institute for Brain Science, Atle will work on validating and developing a biophysically detailed model of the mouse primary visual cortex. With this project, he will contribute both to a better understanding of visual processing in the brain and to improving our insight into the neuronal processing underlying the electrical signals measured with EEG, ECoG, and multielectrodes.
PhD Candidate: Atle Rimehaug, with a B.Sc. in cognitive psychology from the University of Oslo and an M.Sc. in applied physics from NTNU, was searching for a field where he could apply all his skills and scientific knowledge. He found this in computational neuroscience at CINPLA and the SUURPh-program, where strong multidisciplinarity is combined with an excellent international research environment.
Behavior Dependent Neural Stimulation in Freely Moving Animals
The objective of the project is to combine computational modelling with animal experiments to reveal new insight about memory processing in the hippocampal-entorhinal circuit. Maria will use closed-loop stimulation of neurons in the hippocampal-entorhinal circuit involved in various memory tasks, such as spatial navigation in freely moving animals. While experiments with this system can examine the changes in the stimulus responses of target neurons, many questions about the underlying mechanism of those changes can only be answered in computational models. In particular, artiﬁcial intelligence (AI) approaches will be used to study the hippocampus-entorhinal circuit function, which has recently opened new perspectives to understand cognitive behaviors.
PhD Candidate: Maria Perona Fjeldstad holds a B.Sc. in Biomedical Sciences with Pharmacology and a M.Sc. in Integrative Neuroscience from the University of Edinburgh. She is interested in the nervous system, and how different types of neuronal and glial cells communicate in an immensely complicated way to give rise to complex behaviour in living animals. Her Masters degree research involved experimentally analysing sciatic nerve conduction properties.
Previous SUURPh projects
The second cohort of SUURPh students are completing and defending their dissertations in the period from Autumn of 2023 through Summer of 2024. They join in the success of the inaugural SUURPh cohort, which began the program in 2015 and 2016.
Scaling kernel-based learning for big data
Non-Invasive Microwave BOdy-Sensing (NIMBO)
Quantifying the Impact of Germline Gene Variation on Immune Repertoire Architecture
Directing neural learning via brain-machine interfaces
Ultrastructural modelling of cerebrospinal fluid flow
PhD Candidate: Karl Erik Holter
Thesis available here: https://www.duo.uio.no/handle/10852/86036
Supervisors: Kent Andre Mardal (UiO, Mathematics / Simula, BioComp), Anders Dale (UCSD, Radiology and Neurosciences)
VIBRAM: VIrtual BRAin Measurements (EEG, ECoG, MEG)
Computing the origin of fluid flow pathologies in the brain
Coding memories in time
PhD Candidate: Tristan Stöber
Thesis available here: https://www.duo.uio.no/handle/10852/86524
Supervisors: Marianne Fyhn (UiO, IBV), Aslak Tveito (Simula Research Laboratory), Jill Leutgeb (UCSD, Biological Sciences)