Stochastic Population Model for Electrical Stimulation of the Auditory Nerve

We have developed a biophysical model of a population of electrically stimulated auditory nerve fibers. It can be used to interpret results from physiological and behavioral experiments with cochlear implants and propose novel stimulation strategies. Our model consists of myelinated internodes described by a passive resistor-capacitor network, membrane capacitance, and leakage current at the nodes of Ranvier, as well as stochastic representations of nodal voltage-dependent channels. To approximate physiological properties measured in the auditory nerve (AN) of an acutely deafened cat, electrical parameters of the model fiber were chosen based on literature-reported values. Using our model, we have replicated the following properties within 10% of the reported feline single-fiber measurements: relative spread (5.8%), spike latency (630 mus), jitter (93 mus), chronaxie (238 mus), relative refractory period (4.6 ms), and conduction velocity (14 m/s). Moreover, we have successfully matched response characteristics of a population of fibers with the same number of diameter-distributed model fibers, enabling us to simulate responses of the entire AN. To demonstrate the performance of our model, we compare responses of a population of ANs stimulated with two speech encoding strategies, continuous interleaved sampling and compressed analog.     

Prediction for human motion tracking failures.

We propose a new and effective method of predicting tracking failures and apply it to the robust analysis of gait and human motion. We define a tracking failure as an event and describe its temporal characteristics using a hidden Markov model (HMM). We represent the human body using a three-dimensional, multicomponent structural model, where each component is designed to independently allow the extraction of certain gait variables. To enable a fault-tolerant tracking and feature extraction system, we introduce a single HMM for each element of the structural model, trained on previous examples of tracking failures. The algorithm derives vector observations for each Markov model using the time-varying noise covariance matrices of the structural model parameters. When transformed with a logarithmic function, the conditional output probability of each HMM is shown to have a causal relationship with imminent tracking failures. We demonstrate the effectiveness of the proposed approach on a variety of multiview video sequences of complex human motion.