Kaidi Kang

The paper introduces a UGEE framework for analyzing multimodal neuroimaging data, showing it accurately estimates effects and is 100x faster than permutation methods.

The paper develops a new estimator to more accurately assess machine learning model performance in predicting brain-phenotype associations and shows it reduces bias in estimating predictive accuracy.

The paper compares brain-behavior associations across different sampling models using ABCD Study data, finding increased effect sizes with tail-sampled groups but mixed results with propensity-matched models.

This study shows that optimizing study design, such as increasing age variability and adding longitudinal measurements, improves replicability and effect sizes in brain-wide association studies.

The paper evaluates resampling methods for neuroimage topological features, finding that wild bootstrap and permutation perform well in larger samples, while GRF-based methods have inflated error rates.