Signal Extraction from an Innovative Noninvasive Muscular Dystrophy Biomarker Using Functional Mixed-Effects Models

• Activity Number: 243
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Kush Kapur* and Basil Darras and Seward Rutkove
• Companies: Boston Children's Hospital and Boston Children's Hospital and Beth Israel Deaconess Medical Center
• Keywords: Functional data ; State space approach ; Kalman filtering and smoothing ; Mixed-effects models

Abstract

Duchenne muscular dystrophy (DMD) affects approximately 1 in every 3,500 live male births. Electrical impedance myography (EIM) is an innovative method to assess the progression of neuromuscular diseases such as DMD non-invasively. The technique relies on the application of a very low intensity, alternating electrical current at a variety of frequencies between 1 kHz to 10 MHz to a localized area of the muscle. It captures changes in the resulting voltages of the applied current as the current travels through the disease-affected tissue. In this work, we develop a functional mixed-effects model for the EIM longitudinal outcomes while taking into account of the complex frequency-temporal correlations. This model allows us to differentiate the smooth surface of the underlying signal over the pre-specified frequency range and various time-points between the DMD and control groups efficiently. Additionally, it allows us to differentiate the subject-specific deviations from the population averaged response surface. We present results using computationally efficient state space formulation in order to estimate model parameters.