Post-Graduate Stories

Chi-Kuang Yeh Explores New Modelling Approaches for Early Diagnosis of Alzheimer’s Disease

Chi-Kuang Yeh’s research project, titled “Semi-supervised Learning for High-dimensional Functional Data Integration with Measurement Error” will explore new modelling approaches to deal with large-scale high-dimensional and functional data collected with different devices (and often contaminated with measurement error). He will apply the proposed methodologies to identify key imaging and genetic biomarkers for the early diagnosis of Alzheimer’s disease.

Program: CANSSI Distinguished Postdoctoral Fellowship
Region: National
Date: 2023–2025

Project Focus Areas

In the Alzheimer’s Disease (AD) Neuroimaging Initiative (ADNI) study, imaging, clinical, genetic and demographic datasets of patients are collected. However, because the data collection procedures are performed for different groups, when we attempt to integrate data from multiple sources, the differences in data quality can be a potential confounder and cause interested estimators to be biased.

The major goals of the proposed research are to develop powerful and innovative statistical learning methods and computational tools that can systematically handle different measurement error and misclassification mechanisms brought by the multisource/multimodality datasets. The project will address methodological and computational challenges in data integration including high-dimensionality, functional covariates, measurement error and statistical efficiency.

The project team is particularly interested in identifying key imaging and genetic biomarkers for the early diagnosis of AD where imaging data are of different qualities for some subjects.

Getting to Know Chi-Kuang

Chi-Kuang Yeh received his PhD in statistics from the University of Waterloo in 2023.

His research program operates mainly in two directions: optimal design (OD) and functional data analysis (FDA). For OD, his focus is on proposing flexible and fast computational algorithms to numerically solve OD problems, with particular emphasis on the robustness of the results and applications for drug studies. As for FDA, he has developed new statistical methods on densely observed curves. The effectiveness of these methods was demonstrated in applications to model selection for daily electricity price curves.

Chi-Kuang has more than a decade of experience as a teaching assistant and notes that he has always had a keen interest in teaching, as it provides an opportunity to positively impact others in a common discipline. In pursuing his goal of an academic career, he aims to emphasize research and teaching equally.

As a recipient of the CANSSI Distinguished Postdoctoral Fellowship, Chi-Kuang will work under the supervision of Archer Yi Yang and Qihuang Zhang at McGill University and Peijun Sang at the University of Waterloo. 

I have always been passionate about solving problems from real-life corners by integrating methods from different disciplines and delivering the findings to others.