Search

Short-time Fourier transform (STFT) is the most common window-based approach for analyzing the spectrotemporal dynamics of time series. To mitigate the effects of high variance on the spectral estimates due to finite-length, independent STFT windows, state-space multitaper (SSMT) method used a state-space framework to introduce dependency among the spectral estimates. However, the assumed time-invariance of the state-space parameters makes the spectral dynamics difficult to capture when the time series is highly nonstationary. We propose an adaptive SSMT (ASSMT) method as a time-varying extension of SSMT. ASSMT tracks highly nonstationary dynamics by adaptively updating the state parameters and Kalman gains using a heuristic, computationally efficient exponential smoothing technique. In analyses of simulated data and real human electroencephalogram (EEG) recordings, ASSMT showed improved denoising and smoothing properties relative to standard multitaper and SSMT approaches., Comment: IEEE Signal Processing Letters (2022)

Burst suppression is an electroencephalography (EEG) pattern associated with profoundly inactivated brain states characterized by cerebral metabolic depression. Its distinctive feature is alternation between short temporal segments of near-isoelectric inactivity (suppressions) and relatively high-voltage activity (bursts). Prior modeling studies suggest that burst-suppression EEG is a manifestation of two alternating brain states associated with consumption (during a burst) and production (during a suppression) of adenosine triphosphate (ATP). This finding motivates us to infer latent states characterizing alternating brain states and underlying ATP kinetics from instantaneous power of multichannel EEG using a switching state-space model. Our model assumes Gaussian distributed data as a broadcast network manifestation of one of two global brain states. The two brain states are allowed to stochastically alternate with transition probabilities that depend on the instantaneous ATP level, which evolves according to first-order kinetics. The rate constants governing the ATP kinetics are allowed to vary as first-order autoregressive processes. Our latent state estimates are determined from data using a sequential Monte Carlo algorithm. Our neurophysiology-informed model not only provides unsupervised segmentation of multi-channel burst-suppression EEG but can also generate additional insights on the level of brain inactivation during anesthesia., Comment: To appear in the proceedings of the 2020 IEEE Asilomar Conference on Signals, Systems, and Computers

To capture the slowly time-varying spectral content of real-world time-series, a common paradigm is to partition the data into approximately stationary intervals and perform inference in the time-frequency domain. However, this approach lacks a corresponding nonstationary time-domain generative model for the entire data and thus, time-domain inference occurs in each interval separately. This results in distortion/discontinuity around interval boundaries and can consequently lead to erroneous inferences based on any quantities derived from the posterior, such as the phase. To address these shortcomings, we propose the Piecewise Locally Stationary Oscillation (PLSO) model for decomposing time-series data with slowly time-varying spectra into several oscillatory, piecewise-stationary processes. PLSO, as a nonstationary time-domain generative model, enables inference on the entire time-series without boundary effects and simultaneously provides a characterization of its time-varying spectral properties. We also propose a novel two-stage inference algorithm that combines Kalman theory and an accelerated proximal gradient algorithm. We demonstrate these points through experiments on simulated data and real neural data from the rat and the human brain., Comment: Uncertainty in Artificial Intelligence (UAI), 2021

Reinforcement Learning (RL) can be used to fit a mapping from patient state to a medication regimen. Prior studies have used deterministic and value-based tabular learning to learn a propofol dose from an observed anesthetic state. Deep RL replaces the table with a deep neural network and has been used to learn medication regimens from registry databases. Here we perform the first application of deep RL to closed-loop control of anesthetic dosing in a simulated environment. We use the cross-entropy method to train a deep neural network to map an observed anesthetic state to a probability of infusing a fixed propofol dosage. During testing, we implement a deterministic policy that transforms the probability of infusion to a continuous infusion rate. The model is trained and tested on simulated pharmacokinetic/pharmacodynamic models with randomized parameters to ensure robustness to patient variability. The deep RL agent significantly outperformed a proportional-integral-derivative controller (median absolute performance error 1.7% +/- 0.6 and 3.4% +/- 1.2). Modeling continuous input variables instead of a table affords more robust pattern recognition and utilizes our prior domain knowledge. Deep RL learned a smooth policy with a natural interpretation to data scientists and anesthesia care providers alike., Comment: International Conference on Artificial Intelligence in Medicine 2020

Significant effort toward the automation of general anesthesia has been made in the past decade. One open challenge is in the development of control-ready patient models for closed-loop anesthesia delivery. Standard depth-of-anesthesia tracking does not readily capture inter-individual differences in response to anesthetics, especially those due to age, and does not aim to predict a relationship between a control input (infused anesthetic dose) and system state (commonly, a function of electroencephalography (EEG) signal). In this work, we developed a control-ready patient model for closed-loop propofol-induced anesthesia using data recorded during a clinical study of EEG during general anesthesia in ten healthy volunteers. We used principal component analysis to identify the low-dimensional state-space in which EEG signal evolves during anesthesia delivery. We parameterized the response of the EEG signal to changes in propofol target-site concentration using logistic models. We note that inter-individual differences in anesthetic sensitivity may be captured by varying a constant cofactor of the predicted effect-site concentration. We linked the EEG dose-response with the control input using a pharmacokinetic model. Finally, we present a simple nonlinear model predictive control in silico demonstration of how such a closed-loop system would work., Comment: 7 pages, 6 figures. This work has been submitted to IFAC for possible publication

Closed loop anesthesia delivery (CLAD) systems can help anesthesiologists efficiently achieve and maintain desired anesthetic depth over an extended period of time. A typical CLAD system would use an anesthetic marker, calculated from physiological signals, as real-time feedback to adjust anesthetic dosage towards achieving a desired set-point of the marker. Since control strategies for CLAD vary across the systems reported in recent literature, a comparative analysis of common control strategies can be useful. For a nonlinear plant model based on well-established models of compartmental pharmacokinetics and sigmoid-Emax pharmacodynamics, we numerically analyze the set-point tracking performance of three output-feedback linear control strategies: proportional-integral-derivative (PID) control, linear quadratic Gaussian (LQG) control, and an LQG with integral action (ILQG). Specifically, we numerically simulate multiple CLAD sessions for the scenario where the plant model parameters are unavailable for a patient and the controller is designed based on a nominal model and controller gains are held constant throughout a session. Based on the numerical analyses performed here, conditioned on our choice of model and controllers, we infer that in terms of accuracy and bias PID control performs better than ILQG which in turn performs better than LQG. In the case of noisy observations, ILQG can be tuned to provide a smoother infusion rate while achieving comparable steady-state response with respect to PID. The numerical analyses framework and findings, reported here, can help CLAD developers in their choice of control strategies. This paper may also serve as a tutorial paper for teaching control theory for CLAD., Comment: Accepted in the IFAC2020 Conference

Dynamic functional connectivity (DFC) analysis involves measuring correlated neural activity over time across multiple brain regions. Significant regional correlations among neural signals, such as those obtained from resting-state functional magnetic resonance imaging (fMRI), may represent neural circuits associated with rest. The conventional approach of estimating the correlation dynamics as a sequence of static correlations from sliding time-windows has statistical limitations. To address this issue, we propose a multivariate stochastic volatility model for estimating DFC inspired by recent work in econometrics research. This model assumes a state-space framework where the correlation dynamics of a multivariate normal observation sequence is governed by a positive-definite matrix-variate latent process. Using this statistical model within a sequential Bayesian estimation framework, we use blood oxygenation level dependent activity from multiple brain regions to estimate posterior distributions on the correlation trajectory. We demonstrate the utility of this DFC estimation framework by analyzing its performance on simulated data, and by estimating correlation dynamics in resting state fMRI data from a patient with a disorder of consciousness (DoC). Our work advances the state-of-the-art in DFC analysis and its principled use in DoC biomarker exploration., Comment: Presented in 53rd Annual Asilomar Conference on Signals, Systems, and Computers, Pacific Grove, CA

MEG/EEG are non-invasive imaging techniques that record brain activity with high temporal resolution. However, estimation of brain source currents from surface recordings requires solving an ill-posed inverse problem. Converging lines of evidence in neuroscience, from neuronal network models to resting-state imaging and neurophysiology, suggest that cortical activation is a distributed spatiotemporal dynamic process, supported by both local and long-distance neuroanatomic connections. Because spatiotemporal dynamics of this kind are central to brain physiology, inverse solutions could be improved by incorporating models of these dynamics. In this article, we present a model for cortical activity based on nearest-neighbor autoregression that incorporates local spatiotemporal interactions between distributed sources in a manner consistent with neurophysiology and neuroanatomy. We develop a dynamic Maximum a Posteriori Expectation-Maximization (dMAP-EM) source localization algorithm for estimation of cortical sources and model parameters based on the Kalman Filter, the Fixed Interval Smoother, and the EM algorithms. We apply the dMAP-EM algorithm to simulated experiments as well as to human experimental data. Furthermore, we derive expressions to relate our dynamic estimation formulas to those of standard static models, and show how dynamic methods optimally assimilate past and future data. Our results establish the feasibility of spatiotemporal dynamic estimation in large-scale distributed source spaces with several thousand source locations and hundreds of sensors, with resulting inverse solutions that provide substantial performance improvements over static methods.

MEG and EEG are noninvasive functional neuroimaging techniques that provide recordings of brain activity with high temporal resolution, and thus provide a unique window to study fast time-scale neural dynamics in humans. However, the accuracy of brain activity estimates resulting from these data is limited mainly because 1) the number of sensors is much smaller than the number of sources, and 2) the low sensitivity of the recording device to deep or radially oriented sources. These factors limit the number of sources that can be recovered and bias estimates to superficial cortical areas, resulting in the need to include a priori information about the source activity. The question of how to specify this information and how it might lead to improved solutions remains a critical open problem. In this paper we show that the incorporation of knowledge about the brain's underlying connectivity and spatiotemporal dynamics could dramatically improve inverse solutions. To do this, we develop the concept of the \textit{dynamic lead field mapping}, which expresses how information about source activity at a given time is mapped not only to the immediate measurement, but to a time series of measurements. With this mapping we show that the number of source parameters that can be recovered could increase by up to a factor of ${\sim20}$, and that such improvement is primarily represented by deep cortical areas. Our result implies that future developments in MEG/EEG analysis that model spatialtemporal dynamics have the potential to dramatically increase source resolution.

Determining the magnitude and location of neural sources within the brain that are responsible for generating magnetoencephalography (MEG) signals measured on the surface of the head is a challenging problem in functional neuroimaging. The number of potential sources within the brain exceeds by an order of magnitude the number of recording sites. As a consequence, the estimates for the magnitude and location of the neural sources will be ill-conditioned because of the underdetermined nature of the problem. One well-known technique designed to address this imbalance is the minimum norm estimator (MNE). This approach imposes an $L^2$ regularization constraint that serves to stabilize and condition the source parameter estimates. However, these classes of regularizer are static in time and do not consider the temporal constraints inherent to the biophysics of the MEG experiment. In this paper we propose a dynamic state-space model that accounts for both spatial and temporal correlations within and across candidate intracortical sources. In our model, the observation model is derived from the steady-state solution to Maxwell's equations while the latent model representing neural dynamics is given by a random walk process., Comment: Published in at http://dx.doi.org/10.1214/11-AOAS483 the Annals of Applied Statistics (http://www.imstat.org/aoas/) by the Institute of Mathematical Statistics (http://www.imstat.org)