Data Driven Modelling

We investigated the sensor-based and data system dynamic multi-scale models to track and predict the progression of deep brain structure related disorders such as Epileptic seizure, Alzheimer’s and Parkinson’s diseases. Specifically, we attempt to show how theoretical studies performed in physiologically-plausible computational models of neuronal assemblies (“neural mass models”) can enable us to set up some relationships between excitability-related parameters in models and some characteristic electrophysiological patterns typically observed in local field potentials (LFPs) or the EEG recorded under normal or epileptic conditions and other deep brain disorders.

We present an virtual instrumentation platform using real-time, lumped parameter dynamics model to investigate the coupled electrophysiological dynamics of the pathological processes in cardiovascular and neurological process with augmented input data from the bio-signal signals. The dynamical model has been developed based on nonlinear lump parameter models with personalized parameters optimized from the signal characteristics. Such virtual instrument flat form can be used to generate surrogate hemodynamic signals of the cardiovascular system and specify the deep brain activities of neurological disorders that alter deep brain structures.