Projects

Brain tumor patient functional connectivity and its test-retest reliability during awake and anesthesia

• Comparing the functional connectivity and its test-retest reliability between awake and anesthesia
• Using the developed variance estimation method for dynamic functional connectivity
• Poster at International Society for Magnetic Resonance in Medicine (ISMRM), Singapore, May 2024

Variance estimation for dynamic functional connectivity

• A variance estimator for sliding window correlation, considering varying window autocorrelation and cross-correlation
• Window size recommendation for varying window; Dynamic connectivity detection based on confidence interval
• Oral presentation at Eastern North American Region (ENAR), Baltimore, MD, US, March 2024

Early stopping rule for real-time fMRI dynamic experimentation through forecasting

• Forecasted fMRI signal by bootstrap aggregation (bagging) ARIMA model using STL Decomposition and Box-Cox Transformation
• Supported the early stopping rule of real-time fMRI dynamic experimentation
• Publication, Poster at 7th Annual BRAIN Initiative Meeting, Virtual, June 2021

Simulation-based comparisons between generalized spatially varying coefficient models

• Generated varying-coefficient simulation data based on logistic distribution, according to disease occurrence
• Compared the performance of generalized geographically weighted regression under different distributions
• Applied the result to tuberculosis in Shanghai, and Hand, Foot, and Mouth disease in 7 provinces in China

Effects of spatial autocorrelation and hospitals on tuberculosis treatment outcome: An implication of tuberculosis control in a highly integrated city

• Explored risk factors for poor tuberculosis treatment outcome in Shanghai, in both individual level and spatial aspect
• Found the effects of spatial autocorrelation and designated hospital on treatment outcome and the biased estimation of other risk factors if spatial information was ignored
• Visualized the preferences of patients in treatment hospitals and gave constructive suggestions to the government
• Publication

A Simulation on the Effectiveness of the Robustness Index of Propensity Score Estimation

• Collaborated with Dr. Wei Pan, an associate professor at Duke University on the simulation-based study to prove the effectiveness of the robustness index in the presence of uncontrolled confounders
• Implemented the robustness index, probability values and percentage points for Pearson distributions in R
• Improved the method by augmenting the sampling space and bootstrap