PhD candidate in Statistics

Research interests

Statistical methodologies: High-dimensional data analysis, spatial statistics, machine learning
Scientific applications: Imaging, neuroscience, data harmonization

Developed a sparse tensor decomposition method for localizing and mitigating inter-scanner/-site variability in functional connectivity, while preserving functional activity patterns after harmonization.
Proposed an enhanced EM algorithm for estimation.

Developed the Spatial Autocorrelation Normalization (SAN) technique to ensure spatially homogeneous neuroimaging data across different sites and MRI scanners by reducing unwanted technical variations at the vertex level.
Demonstrated through real data and simulations that SAN outperforms existing methods in reducing covariance variations, improving data quality and reproducibility.
Developed an efficient R package (SAN) to share simulated examples and codes.

Developed RELIEF (REmoval of Latent Inter-scanner Effects through Factorization), a novel method that reduces dimensions and factors interlinked matrices, correcting inter-scanner biases while preserving biological associations, and significantly increasing statistical power compared to existing methods.
Developed an efficient R package (RELIEF) and built a GitHub repository for sharing simulated examples and codes publicly.

Proposed a regularized Neural Network model for Scalar on Image problems, incorporating Total Variation and L-1 penalties to recover sparse spatial patterns in nonlinear systems.
Developed a scalable algorithm for efficient parameter estimation and established a generalized framework for prediction and variable selection across various data types, including continuous, binary, and multi-categorical variables.

Registered fMRI 3D volume to brain Atlas to segment the fMRI anatomy into distinct regions.
Identified active ROIs using Temporal Contrast-to-noise Ratio (tCNR), applied tensor regression to detect activation, and localized task-activated regions with post hoc statistical mapping.
Co-developed an efficient R package (TCIU) and built a GitHub repository for sharing simulated examples and codes publicly.

Teaching Assistant @ University of Toronto (September 2021 - Current)

Data Analyst Intern @ Bayer Healthcare Co. Ltd. (August 2018 - January 2019)

Established a sales forecasting ARIMA model based on 4-year monthly data to predict the sales amount of a selected drug in the following year, and optimized the model based on the AIC.
Distributed sales quota to each sales representative by simulations based on historical sales, market potential and sales forecast data.

Student Paper Award (runner-up) in the 2022 Statistical Methods in Imaging (SMI) conference
Data Sciences Institute Doctoral Student Fellowship in 2023 (CAD $25,000 per year up to 3 years)

Talks

2023 Joint Statistical Meetings (JSM)

2022 Statistical Methods in Imaging (SMI) conference

Poster

2023 Statistical Methods in Imaging (SMI) conference

2023 Eastern North American Region (ENAR) meeting

2024 Eastern North American Region (ENAR) meeting

Weinstein, S.M., Tu, D., Hu, F., Pan, R., Zhang, R., Vandekar, S.N., Baller, E.B., Gur, R.C., Gur, R.E., Alexander-Bloch, A.F. and Satterthwaite, T.D., 2024. Mapping individual differences in intermodal coupling in neurodevelopment. bioRxiv, pp.2024-06.
Zhang, R., Chen, L., Oliver, L.D., Voineskos, A.N. and Park, J.Y., 2024. SAN: mitigating spatial covariance heterogeneity in cortical thickness data collected from multiple scanners or sites. Human Brain Mapping, 45(7), p.e26692.
Zhang, R., Oliver, L.D., Voineskos, A.N. and Park, J.Y., 2023. RELIEF: A structured multivariate approach for removal of latent inter-scanner effects. Imaging Neuroscience, 1, pp.1-16.
Zhang, Y., Shen, Y., Zhang, R., Liu, Y., Guo, Y., Deng, D. and Dinov, I.D., 2023. Numerical methods for computing the discrete and continuous Laplace transforms. arXiv preprint arXiv:2304.13204.
Zhang, R., Zhang, Y., Liu, Y., Guo, Y., Shen, Y., Deng, D., Qiu, Y.J. and Dinov, I.D., 2022. Kimesurface representation and tensor linear modeling of longitudinal data. Neural Computing and Applications, pp.1-20.