Dynamic Signatures of Molecular Disorders
Over the last few decades, new technologies have revolutionised our ability to identify faulty genes and other molecular causes of childhood conditions. Yet even where we identify genetic mutations, or abnormal auto-antibodies as the cause for a particular condition, understanding the link between these abnormalities at the smallest scale with the whole brain dysfunction they cause remains challenging.
One possible approach to improve our understanding is computational modelling. We can try out how well different models explain the EEG abnormalities we can observe, and link the model parameters back to disruptions at the scale of individual neurons. This work integrates recent advances in how models can be 'inverted' to explain EEG data (e.g. through dynamic causal modelling) and a rich history of models of neuronal populations (e.g. neural mass models).
We are applying this approach to patient cohorts, animal models, and some healthy study participants to understand the convergent paths towards specific EEG abnormalities. This may help us to develop biomarkers of specific disorders in the future, and in turn holds the potential to improve our ability to tailor treatments to those patients that are most likely to respond.
Synaptic inhibition shapes benzodiazepine responses in status epilepticus [preprint]
Using dynamic causal modelling we track synaptic parameters during administration of emergency medication in paediatric status epilepticus.
Fedele et al. (2023) medRxiv: https://doi.org/10.1101/2023.08.23.23294456
Reduced excitatory neurotransmission causing seizures in NMDAR-Ab encephalitis [paper]
Here we show through detailed in vitro, in vivo and in silico work that counterintuitively reduced excitation underlies seizures in a rat model of NMDAR-Ab encephalitis.
Wright et al (2021) Comms Biol: doi.org/10.1038/s42003-021-02635-8
NMDA-receptor antibodies alter microcircuit dynamics [PNAS]
The full paper from the preprint below is now out - applying computational models across mice and men to infer cortical circuitry in the presence of these pathological antibodies.
Using computational models allows us to translate between different types of observations: Here we link patient EEG recordings with observations made in an experimental mouse model.
Linking EEG abnormalities and model parameters [chapter]
Here we discuss the different ways computational models can be used to explain EEG abnormalities and illustrate one approach using NMDAR-Ab encephalitis as an example.
Rosch et al (2016) 10.1007/978-3-319-49959-8_6 [link to author manuscript]