Skip to content

Regression Overview + A Study of Medical Treatments for COVID-19

Photo of
Hosted By
Madeline and 4 others
Regression Overview + A Study of Medical Treatments for COVID-19


# Schedule
Tuesday, November 17th, 6:30-8:30pm
6:30 - Zoom Breakout Rooms Networking Activity
6:55 - Announcements
7:00 - Regression Overview
7:35 - A Study of Medical Treatments for COVID-19
8:30 - Meeting Adjourned

# Regression Overview
Speaker: Madeline Bauer, PhD

Bio: Madeline Bauer holds a Ph.D. in Mathematical Statistics. She has collaborated with medical and surgical oncologists in the design and analysis of large-scale Phase III cancer clinical trials, using survival and Cox proportional hazards regression analysis. She has also worked closely with radiation biologists in developing and implementing innovative dose response designs for Phase I and II studies to evaluate dose escalation of radiation therapy, with immunologists developing designs for Phase II studies to evaluate combination therapy with biologic agents, and has used robust regression and data visualization tools in S-Plus and R to visualize dose response surfaces for combination antifungal drug therapy in murine models and in vitro drug susceptibility testing. Now retired, Dr. Madeline Bauer helps high school students with their science projects, mentors undergraduate statistics and data science students preparing for ASA DataFest, and serves as a co-organizer for R-Ladies Irvine.

Abstract: The general purpose of developing a regression model is to understand the association between a response Y and X and/or to predict Y given X, where X is a set of independent variables. In this talk, we give an overview of regression and highlight some of the regression packages available in R. We start with examples of suggested best workflow practices that are useful regardless of the specific regression model. For model-specific workflows, we take advantage of the "methods" associated with the model, such as "summary", "print", and especially "plot." These methods illustrate the individual package author’s approaches to check the model assumptions, estimate the association between the response and the covariates and evaluate the fit and accuracy of the resulting model. We will demonstrate these methods with real datasets. The data and R code will be available to download.

# A Study of Medical Treatments for COVID-19
Speaker: Song Zhai

Bio: Song Zhai is a Ph.D. candidate in the Department of Statistics at UC, Riverside. After finishing his master degree in the Department of Mathematics at Texas A&M University, Song Zhai joined UCR in Fall 2016, under the supervision of Dr. Xinping Cui. Song’s previous projects include a retrospective study for biomarker selection with Parkside Medical Group; efficient polygenic risk score developing at Merck; and best predictive biomarker identification with Xinping. Song’s research focus is on pharmacogenomics, with an emphasis on applications rooted in biomarker study, multiple testing, and causal inference.

Abstract: To the end of this study, various medical treatments for COVID-19 are attempted. A total of 417 patients were considered and 414 of them were included in this study (3 deaths) with mild-to-critical COVID-19. Patients were followed up for recurrence for 28 days quarantine after being discharged. We applied the Synthetic Minority Oversampling Technique (SMOTE) to overcome the rare recurring events in certain patient subgroups. Virtual Twins Matching (VTM) analysis, facilitated by random forest regression, was performed for medical treatment-recurrence classification, while minimizing effects of confounding factors. We used the Multiple Comparisons with the Best (MCB) as the test of the best drug combination in each patient subgroup. The insights into combinatorial therapy found in this study shed light on the use of a combination of (biological and chemical) anti-virus therapy and/or anti-cytokine storm as a potentially effective therapeutic treatment for COVID-19.

# Meeting Materials
All meeting materials will be sent to registered attendees after the event.
Online event
This event has passed