December 4, 2025
Authors: Maria Semkovska, Bonnie Liefting Publisher: Psychiatry Research, Volume 356, 2026, 116883, ISSN 0165-1781, DOI: 10.1016/j.psychres.2025.116883
Abstract: Depression’s direct and indirect health effects compound the disease burden of normal aging. Personalised prevention is limited by existing diagnostic classifications. The network theory conceptualises individual depressive symptoms and associated protective factors (e.g., cognition) as a complex system of interrelated components, which enables the identification of potential personalisation targets. We evaluated the temporal stability and connectivity of extended networks including symptoms, cognitive functions and daily activities in elderly co-twins discordant for past depression, and explored possible genetic contributions to the structure of associations between the networks’ elements. Gaussian graphical models estimated the networks of 228 like-sex (104 monozygotic and 124 dizygotic) co-twins aged >70 from the Danish Twin Registry at three timepoints, set two years apart. The networks of both the affected and unaffected by past depression co-twins showed stable global strength and global structure across time. At each timepoint, the affected co-twins showed stronger global network associations and stronger local associations among both depressive symptoms and cognitive functions than the unaffected networks. In affected co-twins, anhedonia was the most central network element (i.e., with the strongest independent associations with remaining variables), and learning – the most central cognitive function. In unaffected co-twins, memory functions were the most influential elements in the extended networks of depressive symptoms. Frequency of engagement in daily activities was not central in either networks at any timepoint. No significant differences were observed when comparing monozygotic to dizygotic networks. Depression preventative strategies in the elderly should target anhedonia and stimulating/maintaining memory functions.
January 24, 2024
Abstract: By designing coupling to control populations of oscillators, we can control their synchonisation behaviour. Oscillators (e.g. neurons) can be coupled on different levels. The most basic level is through links between pairs of oscillators. However, using graphs with only pairwise links is not necessarily a satisfactory approximation of reality as nonpairwise interactions can be found in many dynamical systems including social networks and the human brain. Even though the effects of these nonpairwise interactions have been observed, described and modeled in a wide range of oscillatory systems, controlling nonpairwise interactions in arbitrary populations of oscillators has remained a relatively unexplored area. In this thesis we generalise synchronisation engineering to control nonpairwise interactions in arbitrary systems. We designed a nonlinear time-delayed coupling that can be used to match the phase reduction of a system of oscillators to a target phase model. The contribution of this thesis is allowing for nonpairwise interactions in the target phase model. We used an optimisation proceidure to find coupling parameters to match a nonpairwise target phase model that has the collective behaviour we aim to introduce to the system We found that we need one additional filter to find the parameter sets that match the bifurcation of both in-phase and splay configuration in to the nonpairwise target phase model.
Supervisors: Kyle Wedgwood (University of Exeter) and Christian Bick (Vrije University of Amsterdam)
June 14, 2022
An outreach article for Videnskab.dk, find it here (in Danish).
May 27, 2021
Abstract: By designing feedback to control populations of oscillators, synchronisation in neurons causing epileptic seizures can be broken up. Oscillators (e.g. neurons) can be coupled on different levels. The most basic level is through links between pairs of oscillators. These pairwise links fail to explain phenomena such as peer pressure. The nonpairwise ‘links’ make such phenomena possible. Even though the effects of these nonpairwise interactions have been observed, described and modelled in a wide range of oscillatory systems, controlling nonpairwise interactions in arbitrary systems has remained a mainly unexplored area. We generalize synchronisation engineering to control nonpairwise interactions in arbitrary systems. As a first step, we design nonlinear time-delayed feedback that introduces bifurcations away from splay configuration into arbitrary systems. Controlling nonpairwise interactions might advance the design of minimum-power stimuli for the treatment of epilepsy.
Session Abstract: Phase reduction is a powerful technique for the analysis and modelling of oscillatory dynamics in various applications, including power grids and biological systems, such as neural systems. A phase reduction describes the dynamics of weakly coupled limit-cycle oscillators in terms of their phase, a single variable for each oscillator. The effect of coupling between nodes in the network on the evolution of these phase variables is then captured by a single, phase response function, which can be inferred from data, or can be computed for given dynamical models. A barrier to the application of phase reduction approaches is that obtaining them is not straightforward, particularly when the oscillators cannot be isolated. Additionally, phase reductions are formally defined for weak coupling, that is, when the strength of coupling between units is small compared to the intrinsic oscillatory dynamics of each node. When the weak assumption does not hold, higher-order phase reduction techniques, or inclusion of additional dynamic variables, are often necessary to extend the validity of phase reduction. This minisymposium will highlight recent advances in the applicability of phase reduction techniques, covering extensions to phase-amplitude coordinates, efficient approaches for phase response function inference, and approaches for understanding and controlling higher order network interactions.
Talk at the SIAM Conference on Applications of Dynamical Systems 2021 (DS21).
February 25, 2021
Abstract: We will go through the concepts of oscillators, phase, isochrones, phase response curves and phase reduction. We then use phase reduction techniques to design weak nonlinear time-delayed feedback to control arbitrary oscillatory systems. The collective behaviour that we introduce to a system of oscillators in this way is described by a target phase model. In our generalisation of this (“synchronisation engineering”) approach, the target phase model can have nonpairwise interactions. Controlling nonpairwise interactions might advance the design of minimum-power stimuli for the treatment of epilepsy.
Talk at the Exeter Mathematics Postgraduate Seminars.
August 27, 2020
Abstract: By designing feedback to control populations of oscillators, synchronisation in neurons causing epileptic seizures can be broken up. Oscillators (e.g. neurons) can be coupled on different levels. The most basic level is through links between pairs of oscillators. These pairwise links fail to explain phenomena such as peer pressure. The nonpairwise ‘links’ make such phenomena possible. Even though the effects of these nonpairwise interactions have been observed, described and modeled in a wide range of oscillatory systems, controlling nonpairwise interactions in arbitrary systems has remained a mainly unexplored area. We generalize synchronisation engineering to control nonpairwise interactions in arbitrary systems. As a first step, we design nonlinear time- delayed feedback that introduces bifurcations away from splay configuration into arbitrary systems. Controlling nonpairwise interactions might advance the design of minimum-power stimuli for the treatment of epilepsy.
Talk at Dynamics Days Digital 2020.
You can find a video of this talk on Youtube here.
June 19, 2020
Talk at the International Women in Engineering Day (INWED) at the University of Exeter.
April 9, 2020
Talk at the Dynamics Internal Seminar at the University of Exeter.
March 28, 2019
Talk at the Dynamics Internal Seminar at the University of Exeter. See Chapter 2 of my master thesis.
June 1, 2018
Abstract: Fænomenet synkronisering er blevet undersøgt ved hjælp af Kuramoto modellen samt med forskellige tilpasninger deraf. Denne model beskriver en stor population af koblede oscillatorer. Tilpasningerne som er undersøgt her inkluderer en bimodal frekvens fordeling, tilføjelse af hvid støj, kobling afhængig af beliggenhed og en uniform faseforskydning. Vi reproducerer analyser og udfører simulationer, hvor blandt andet Monte Carlo metoder er brugt. Ved hjælp af tid-frekvens analyse baseret på wavelets er vi i stand til at opdage partiel synkronisering.
Supervisor: Kristian Debrabant (University of Southern Denmark)
External supervisor: Holger Waalkens (University of Groningen)
Have a look at my master thesis here
Code from Appendix B (Runs simulations for the different adaptations of the Kuramoto model. Uses Monte Carlo method for the noisy Kuramoto model.)
Code from Appendix C (Makes phase plots.)
June 1, 2016
Abstract: The phenomenon of synchronisation is studied by means of the Kuramoto model. This model describes a large population of coupled oscillators with natural frequencies taken from a narrow distribution. It is assumed that the coupling between the oscillators is mean-field and purely sinusoidal. We follow Kuramoto’s analysis to obtain a formula for the critical coupling. Then the properties of the Kuramoto model are studied with the aid of Poincaré maps. We then conclude with a time-frequency analysis of the order parameter. With the aid of this time-frequency analysis we were able to detect partial synchronisation.
Supervisor: Holger Waalkens (University of Groningen)
Have a look at my thesis here (pdf)
Code from Appendix B.1 (Computes the trajectory of the order parameter of the Kuramoto model.)
Code from Appendix B.2 (Calculates the wavelet transform of a single trajectory of the 3D system.)
Code from Appendix B.3 (Converts the data to images.)