Research

My research interests mainly lie in precision medicine, machine learning, random forests, survival analysis, and genomics.

Research questions

• Some people are cautious about having physical activities due to a lack of some homeostasis capability, such as hypoglycemia. To remedy this, some behavioral and nutritional planning should be accompanied. What reinforcement learning framework would be good for solving this problem?
• There could be two latent subgroups on each side of a plane that have different regression models. How can we estimate the plane and make inferences for it? Kang, Cho, Song, Banerjee, Laber and Kosorok 2022 “Change-plane regression”
• When patients go through multiple rounds of treatments, what treatment rule would patients be benefited the most in terms of their overall survival status? Cho et al. 2022 Survival Dynamic Treatment Regimes
• Survival time is not often exactly observed but only the interval is provided. If one is interested in estimating the survival probability for different covariate values, what tool can we use? Cho et al. 2021 ICRF
• Say we are interested in modeling the bivariate distribution of overdispersed count data. Moreover, the data have an excess number of zeros, or is zero-inflated. The Cho et al. 2021 BZINB model provides a bivariate model.

Ongoing projects

• An mHealth methodology development regarding lifestyle improvement for Type 1 diabetes patients
• Non-regular M-estimator asymptotics (change-plane analysis, outcome weighted learning)
• Precision medicine methodology development with non-standard data (non-compliance, competing risks)
• Discovering metagenomic association with children’s dental caries (“what bacteria and what functionality are associated with dental caries?”)

Selected publications

• Kang, C., Cho, H., Song, R., Banerjee, M., Laber, E.B., Kosorok, M.R. Inference for Change-Plane Regression. (2022) arXiv.

• Cho, H., Holloway, S.T., Kosorok, M.R. (2022) Multi-stage optimal dynamic treatment regimes for survival outcomes with dependent censoring. Biometrika R package: dtrSurv on CRAN. (JSM 2021 award-winning paper)

• Cho, H., Jewell, N., Kosorok, M.R. (2021). Interval Censored Recursive Forests. Journal of Computational and Graphical Statistics. R package: icrf on CRAN.

• Cho, H., et al. (Submitted) Pathobiont-mediated spatial structuring enhances biofilm virulence in childhood oral disease

• Cho, H., Liu, C., Preisser, J., Wu, D. (bioRxiv). Bivariate zero-inflated negative binomial model for identifying underlying dependence. R package: bzinb on CRAN.

• Cho, H., Matthews, G. J., & Harel, O. (2018). Confidence intervals for the area under the receiver operating characteristic curve in the presence of ignorable missing data. International Statistical Review.

 

Software

• R package: dtrSurv (dynamic treatment regime for dependent survival outcomes) to appear on CRAN.

• R package: icrf (Interval Censored Recursive Forests) on CRAN.

• R package: bzinb (Bivariate Zero-Inflated Negative Binomial Model Estimator) on CRAN.