Web-Based Lectures
Title: Accounting for Bias In Big Data
Presenter: David Marker, PhD, Marker Consulting, LLC
Date and Time: Monday, October 23, 12:00 p.m. – 1:30 p.m. EST
Sponsor: Survey Research Methods Section
Registration Deadline: October 23, at 11:30 a.m. EST
Description:
Government agencies are excited by the prospect of using Big Data sets to supplement and/or replace surveys in the production of official statistics. But Big Data are known to suffer from many sources of potential bias. Do their large sizes overcome these biases so that they can be useful for official data? Recent research by Xiao-Li Meng and others has demonstrated the limitations of Big Data with examples from COVID-19 vaccinations and elections. This talk will summarize some of that research and then look at whether electronic health records can be used to improve on the Behavioral Risk Factor Surveillance System estimates of diabetes prevalence and control at the state level. This research has wide application to Big Data in general, and the potential usefulness of a wide range of electronic health records data sets like All of Us and IQVia.
Registration Fees:
Survey Research Methods Section Members: $20
ASA Members: $30
Student ASA Member: $25
Nonmembers: $45
Register
Access Information
After registering, you will receive a confirmation email. In the body of the confirmation email under the “Additional Information” header is information on how to access the webinar. Save this email or register with Zoom right away.
Title: Deep Learning Methods for Survival Analysis
Presenter: Ying Ding, Department of Biostatistics, University of Pittsburgh, PA, USA
Date and Time: November 28, 2023, 12:00 p.m. – 2:00 p.m. ET (This webinar will be taught via Zoom)
Sponsor: Lifetime Data Science Section
Registration Deadline: TBA
Description:
This webinar covers recent developments in deep learning-based methods for survival data analysis and provides case studies to apply these methods. Part 1 will introduce various neural networks for analyzing time-to-event data under different censoring mechanisms. Part 2 will introduce deep learning methods for estimating individualized/conditional average treatment effects for survival outcomes under the causal inference framework. In each part, we will demonstrate the implementation of the methods using R and Python and use case studies to illustrate the applications of these methods for biomedical and health research.
Part 1 – Deep Learning for Survival Analysis and Predictions
---------------------------------------------
- Neural networks for right-censored survival data with time-independent or time-dependent covariates
- Case Study 1: Prediction of Progression of AMD (Age-related Macular Degeneration)
- Neural networks for interval-censored (and left truncated) survival data
- Case Study 2: Prediction of Development of AD (Alzheimer’s Disease)
Part 2 – Deep Learning for Causal Survival Analysis
---------------------------------------------
- CATE (conditional average treatment effect) for survival outcomes
- Deep learning approaches for estimating CATE with survival outcomes
- Case Study 3: Childhood Asthma EHR data analysis
Registration Fees:
Lifetime Data Science Section Members: $20
ASA Members: $30
Student ASA Member: $25
Nonmembers: $45
Register
Access Information
After registering, you will receive a confirmation email. In the body of the confirmation email under the “Additional Information” header is information on how to access the webinar. Save this email or register with Zoom right away.