Joint Institution Seminar Series

Joint Institution Seminar Series

The Biostatistics, Epidemiology, and Research Design (BERD) groups at Duke, UNC, and Wake Forest are all supported by the Clinical and Translational Science Award and have teamed together to share educational and training materials. The following seminars will be shared across all three institutions.

Introduction to Ontology in the Digital Era

Friday, January 14th at 1:30-2:30pm

Zoom link (no registration necessary): https://duke.zoom.us/j/93778474406?pwd=VEpYc0d4eVhmSEZsNGVINHE3SlZzQT09

Anna Maria Masci, PhD, MS
Research Scientist, Biostatistics & Bioinformatics
Duke University School of Medicine

In the digital era, most groups and organizations are dealing with a fast-growing massive amount of data in almost all knowledge domains, with biomedicine, business, and social interactions leading the way. Heterogeneity, unstructuredness, and incompleteness are relevant limitations of the current data sets, while common standards represent a powerful strategy to improve data access and reuse. Ontologies are controlled vocabularies providing standard definitions for the elements of data sets, which facilitate data retrieval by allowing the grouping of annotations. Of note, ontologies allow data integration within and across domains (e.g., biomedical, social, environmental) and granularity levels (e.g., organism, organ, tissue, cell, and molecule). The topic of the talk is to introduce ontologies, explain their differences with vocabulary, thesaurus, taxonomies, while highlighting their role in data accessing and maintenance.

Audience: quantitative methodologists/analysts and clinical investigators

Technical level: beginner

Statistical Power: Understanding the Inputs and How You Come Up with Them

Wednesday, January 19th at 12:00-1:00pm

Registration: https://redcap.wakehealth.edu/redcap/surveys/?s=J4J8WXD7WTHEW7WJ

Walter T. Ambrosius, PhD
Professor, Biostatistics and Data Science
Wake Forest School of Medicine

This session will cover approaches to estimate statistical power, sample size, and effect estimates to be considered in the study design phase.

Audience: clinical investigators

Technical level: beginner

Interaction and Effect Modification: What Are They and How Are They Different

Wednesday, February 2nd at 12:00-1:00pm

Registration: https://redcap.wakehealth.edu/redcap/surveys/?s=R3DNT78YAC4EAR9E

Mike Bancks, MPH, PhD
Assistant Professor of Epidemiology and Prevention
Wake Forest School of Medicine

This advanced level session will introduce, describe, and distinguish causal interaction and effect measure modification.

Audience: clinical investigators

Technical level: intermediate

Biostatistics: Analysis of a Continuous Outcome

Wednesday, February 16th at 12:00-1:00pm

Registration: https://redcap.wakehealth.edu/redcap/surveys/?s=FXACWFL4XYMALNJN

Joseph Rigdon, PhD
Assistant Professor, Biostatistics and Data Science
Wake Forest School of Medicine

This introductory level session will present the basic statistical concepts and approaches for analyzing outcome data in continuous form.

Audience: clinical investigators

Technical level: beginner

Biostatistics: Analysis of Repeated Measure Outcomes

Wednesday, March 2nd at 12:00-1:00pm

Registration: https://redcap.wakehealth.edu/redcap/surveys/?s=P7AMFJ7CE83CJTP4

Joseph Rigdon, PhD
Assistant Professor, Biostatistics and Data Science
Wake Forest School of Medicine

This session will present approaches for analyzing repeated measures outcome data.

Audience: clinical investigators

Technical level: beginner

Presenting Your Results

Wednesday, March 16th at 12:00-1:00pm

Registration: https://redcap.wakehealth.edu/redcap/surveys/?s=E3PT7NMJRRMCK8FX

Mike Bancks, MPH, PhD
Assistant Professor of Epidemiology and Prevention
Wake Forest School of Medicine

This session will provide guidance on the basic “Dos and Don’ts” of presenting your results that can be used in manuscript/grant writing and poster/oral presentations.

Audience: clinical investigators

Technical level: beginner

Study Design: Screening Methodology and Statistics

Wednesday, April 6th at 12:00-1:00pm

Registration: https://redcap.wakehealth.edu/redcap/surveys/?s=NY3JPTJMT9M973XX

Beverly Levine, PhD
Assistant Professor, Social Sciences and Health Policy
Wake Forest School of Medicine

This session will cover fundamentals of screening methodology and common statistical measures used in screening for disease prevention (e.g., Sens, Spec, NPV, and PPV).

Audience: clinical investigators

Technical level: beginner

Prediction and Causation in EHRs

Friday, April 8th at 1:30-2:30pm

Zoom link (no registration necessary): https://duke.zoom.us/j/98947384406?pwd=MnJlekV1SnpPN3JFRG9LazV5alYydz09

David Page, PhD
Professor, Biostatistics & Bioinformatics
Duke University School of Medicine

Some applications of machine learning in medicine need only accurate prediction, while others require accurate attribution of cause and effect.  This talk provides some empirical and theoretical results of both types of applications, including showing the following.  First occurrence of thousands of ICD codes can be predicted with average AUC above 0.7, and accuracies can be further significantly improved by using family histories constructed entirely automatically from de-identified patient data.  Beneficial and harmful side effects of drugs can be identified accurately by machine learning, but only when algorithms are modified to consider unobserved and partially-observed confounders, especially time-varying confounders.

Audience: quantitative methodologists/analysts

Technical level: advanced


Fall 2021 Seminars

Experimental Study Design

Wednesday, October 6th, 12pm-1pm EST

Mike Bancks, MPH, PhD
Assistant Professor of Epidemiology and Prevention
Wake Forest School of Medicine

This session provides an overview of the experimental study design and fundamental concepts including randomization, equipoise, and masking.

Level/audience: Clinical and translational researchers who have basic quantitative training in biostatistical methods

Neural Networks for Survival Outcomes Applied to Medical Images

Friday, October 8, 2021, 1:30-2:30pm EST

Samantha Morrison, PhD
Biostatistician III
BERD Methods Core

Neural networks have become widely used for development of risk prediction models. In particular, convolutional neural networks offer a promising method for incorporating imaging data into risk prediction models.  However, these algorithms cannot be directly applied to situations of incomplete outcome data which poses a problem for development of risk prediction models for survival outcomes.  Standard neural networks build prediction models through estimating an unknown weight vector by minimizing a loss function.  To account for censored data, we proposed an extension to these algorithms that replaces the standard loss function with censoring unbiased loss functions. 

In this talk, we provide background on neural networks for imaging data.  Then, we discuss the methodological and practical complications associated with using convolutional neural networks for censored data.  We propose our censoring unbiased loss functions and illustrate the performance through an analysis of a histology dataset of gliomas.

Level/audience: Applied biostatisticians

Observational Study Designs

Wednesday, November 3rd, 12pm-1pm EST

Mike Bancks, MPH, PhD
Assistant Professor of Epidemiology and Prevention
Wake Forest School of Medicine

This introductory-level session will provide an overview of the fundamental observational study designs used in clinical and epidemiological research.

Level/audience: Clinical and translational researchers who have basic quantitative training in biostatistical methods

Biostatistics Seminar Series: Causal inference with observational data: A gentle introduction

Friday, November 05, 2021, 10:30 am - 12:00 pm

Michael Hudgens, PhD
Professor and Associate Chair, Department of Biostatistics
Gillings School of Global Public Health, UNC-Chapel Hill

Biomedical researchers often want to answer causal questions, but they often have access to observational data, not clinical trials. In this session of the TraCS Biostatistics Seminar series, you’ll learn why causal inference is difficult with observational data and what can be done to allow for valid causal inferences if you have observational data.

Level/audience: Clinical and translational researchers who have basic quantitative training in biostatistical methods

Getting Started on Literature Reviews

Wednesday, November 17th, 12pm-1pm EST

Brandy W. Hardy
Acquisitions and e-Resource Librarian
Wake Forest School of Medicine

The goal of this session is be to provide a balance of beginner and intermediate-level skills and resources (both general and local) for researchers when undertaking a literature review.

Level/audience: Clinical and translational researchers

Tips for Effective Data Visualization

Friday, December 10, 2021, 1:30-2:30pm EST

Eric E Monson, PhD
Data Visualization Specialist
Duke Libraries Center for Data and Visualization Sciences

Visualization is a powerful way to reveal patterns in data, attract attention, and get your message across to an audience quickly and clearly. But, there are many steps in that journey from exploration to information to influence, and many choices to make when putting it all together to tell your story. I will cover some basic guidelines for effective visualization, point out a few common pitfalls to avoid, and run through a critique and iterations of an existing visualization to help you start seeing better choices beyond the program defaults.

Level/audience: Applied biostatisticians, statistics for clinicians