Research

A Novel Testing Method for Random Functional Effects in A fMRI Pain Study

• Established a new functional linear mixed model for scalar on function regression using continuous fMRI brain activation data with random functional effects to predict pain response
• Proposed a novel equal-variance test to assess the existence of the random subject-specific functional effects
• The proposed test has good size and better power property compared to the Bonferroni-corrected test in 24 simulated cases. Proposed equal variance test and Bonferroni-corrected test are denoted as solid black line and red dashed line, respectively in the following figure.

• The data application on a fMRI thermal pain study indicates significant subject-specific effects of human brain hemodynamics related to pain and provides insights on how the effects might differ across subjects

Computional Spatial Data Analysis

• Built a spatial poisson model and conditional autogressive (CAR) model to evaluate the association between states homicide rate and LASSO selected state gun laws
• Established a spatio-temporal model for predicting the street crimes in August 2015 using a real-time large dataset on ``day call for service” records for street crimes in Portland (from 2012/03/01 to 2015/07/31). The predicted crime and original crime records in August 2015 for Portland are compared as follows: