The brain can be viewed as an extremely complex and high-dimensional dynamical system. Despite its complexity, only very limited measures of brain activity are generally accessible to recording?e.g. the electroencephalogram (EEG). Nonlinear dynamics provides the tools to extract information from a limited measurement to determine the invariant nonlinear properties of the underlying dynamical system. In Delay Differential Analysis (DDA), a low-dimensional nonlinear functional embedding is built from the dynamical structure of the data; this serves as a basis onto which the data can be mapped. By constraining the models used to low dimensionality, we ensure that DDA is immune to overfitting, insensitive to noise, and generalizes well to new data. DDA has already been applied to human intracranial recordings of sleep to detect sleep spindles and characterize their spatiotemporal development. In the proposed project, this method will also be applied to EEG data from a large study of schizophrenia. In both of these datasets, distinct observed phenomena can be linked to different underlying cortical states. By finding DDA models which detect sleep spindles, insights can be gained into their dynamics, and this information can be used to refine sophisticated circuit models for their generation. Likewise, by finding models which reliably distinguish schizophrenia patients from control subjects, we can develop a better understanding of the dynamical differences that might give rise to sensory processing deficits and other symptoms of schizophrenia. Further extensions of this work could help to address aditional questions related to functionally distinct states of the brain including in additional neurological and psychiatric disorders.
Delay differential analysis (DDA) is an exciting new technique which can help to facilitate major advances in neuroscience as a tool for accessing dynamical properties of the brain. DDA has allowed for detailed study of the generation and development of sleep spindles in intracranial recordings, and in the proposed project this technique will be used to study mismatch negativity (MMN), a biomarker for schizophrenia, in the largest electroencephalogram (EEG) data set of schizophrenia patients and nonpsychiatric comparison subjects. This project will also lay the groundwork for future applications of DDA to the study of neuropsychiatric disorders.
