Masters candidate in Biostatistics, Tingyang Zhou, will present:
“Functional Data Analysis on Magnetic Resonance Spectroscopy Data of Spinocerebellar Ataxias”
Plan B Adviser: Lynn Eberly
Abstract: This paper aims to implement functional data analysis methods on the entire magnetic resonance spectroscopy (MRS) spectrum acquired from three regions of the brain, pons, cerebellar vermis and cerebellar white matter (CWM), to predict Scale for the Assessment and Rating of Ataxia (SARA) and Activities of Daily Living (ADL) score for different types of spinocerebellar ataxia (SCA) patients and compare results between full spectrum model and metabolites-specific quantification. Pooled SCA group and sub-group analysis were conducted. The results show functional data analysis can be used to predict SARA and ADL scores. Compared to using the full spectrum data and separate metabolite data for prediction, the quantified metabolites treated as functional data performed better in prediction but are scientifically the least interpretable quantitative predictor. Moreover, functional linear regression was more efficient than multiple linear regression in terms of MSE most of the time.
Refreshments will be served prior to the presentation.