University of Austin’s Post

Our next Spark Talk is scheduled for next Friday, January 19th, at 10:30am! Register now here: https://lnkd.in/gK6xkQ2r Speaker: Rod Little (University of Michigan, Department of Biostatistics) Topic: “Handling Missing Data in Social Science Studies” Rod Little review methods for handling missing data in empirical studies in the social sciences. Rod defines missing data, provide a taxonomy of main approaches to analysis, including complete-case and available-case analysis, weighting, maximum likelihood (ML), Bayes, single and multiple imputation (MI), and augmented inverse-probability weighting (AIPW). Assumptions about the missingness mechanism are key to the performance of alternative methods; Rod define missingness mechanisms, which play a key role in the performance of methods. Approaches to robust inference, and to inference when the mechanism is potentially missing not at random, are discussed.

Yan Ma, PhD, Biostatistics Department chair, is the corresponding author for the newly published paper, "Smart data augmentation: One equation is all you need,” featured in the Statistical Analysis and Data Mining: The ASA Data Science Journal. The paper's co-authors are Lu Tang, PhD, vice chair for education in the Biostatistics Department; Yuhao Zhang, PhD, a mathematical statistician with the U.S Food and Drug Administration; and Yuxiao Huang, PhD, assistant professor of data science at George Washington University. In this paper, the authors propose an approach named smart data augmentation (SDA), which aims to augment imbalanced data in an optimal way to maximize downstream classification accuracy. Empirical results on a wide range of datasets demonstrate that SDA could significantly improve the performance of the most popular classifiers such as random forest, multi-layer perceptron, and histogram-based gradient boosting.

This Friday, October 6th, we have the honor of hosting our first guest speaker, Dr. Andrew Gelman of Colombia University. Dr. Andrew Gelman is a professor of statistics and political science and director of the Applied Statistics Center at Columbia University. Andrew has done research on a wide range of topics, including: why it is rational to vote; why campaign polls are so variable when elections are so predictable; why redistricting is good for democracy; reversals of death sentences; police stops in New York City, the statistical challenges of estimating small effects; the probability that your vote will be decisive; seats and votes in Congress; social network structure; arsenic in Bangladesh; radon in your basement; toxicology; medical imaging; and methods in surveys, experimental design, statistical inference, computation, and graphics. Dr. Gelman is known for his contributions to Bayesian statistics, data analysis, and social science research. is widely recognized for his work on Bayesian modeling, hierarchical modeling, and data visualization. If you have any questions for Dr. Gelman, please fill out the form below: https://lnkd.in/gsfVgVsq Meeting Link: https://lnkd.in/gJWCx29X

📊 The field of statistics is dynamic, continually evolving to adapt to new methodologies and perspectives. In the realm of behavioral sciences, this evolution is evident in the shift from merely assessing the reliability of experimental results to considering their broader significance and consistency across multiple studies. 📊 One significant advancement is the adoption of meta-analysis, which involves synthesizing findings from various studies to ascertain overarching trends or consensus within a given field. This approach, particularly prominent in medicine under the banner of evidence-based practice, underscores the importance of contextualizing individual studies within a broader research landscape. 📊 A concrete example elucidating these concepts is the study of drug tolerance, exemplified by Shepard Siegel’s research on morphine tolerance. Through innovative experimental design, Siegel demonstrated the influence of environmental context on drug response, with implications for understanding and mitigating drug overdose risks. 📊 This example not only elucidates statistical principles but also underscores the relevance of statistics in addressing pressing societal issues such as drug abuse. By delving into such case studies, we gain a deeper understanding of statistical concepts and their real-world applications. It will help if we think about what events in our own lives or the lives of people around us illustrate the phenomenon of tolerance. 📊 The example illustrates the concept of generalizing findings from experimental settings to real-world scenarios. While studies conducted in controlled laboratory environments provide valuable insights, their applicability to broader contexts, such as drug overdose cases in novel settings, may require careful consideration and further investigation. 📊 The example resonates with broader human experiences, such as the development of tolerance to various stimuli or behaviors. By reflecting on personal or observed instances of tolerance, individuals can gain insights into the underlying mechanisms and implications for behavior. #Statistics #BehavioralScience #MetaAnalysis #ResearchInsights #SAIST

How do you model a binomial probability, when all you see are zeros (or ones)? Proud to be part of a new approach led by Frank Tuyl and coauthored with Richard Gerlach. The main dish comes with a corresponding dessert of a posterior probability of “homogeneity” and posterior predictive distribution. The new recipe has just appeared in Statistical Science: "On the Certainty of an Inductive Inference: The Binomial Case" DOI: 10.1214/23-STS913 Thanks for this contribution to a long and interesting discussion in the literature, Frank! Learn more about Frank here: https://lnkd.in/etaZxeRz

Join us November 8th from 3:30 - 4:30 p.m. (MT) for our Big Data Seminar: Machine Learning. This is a free online talk. Access the event at https://buff.ly/2XQ2tVW University of Colorado Anschutz Medical Campus Colorado School of Public Health Colorado Clinical and Translational Sciences Institute University of Colorado Denver #biostatistics #research #teamscience #collaborativeresearch #healthdata #imagingdata #datawrangling

A terrific and inspiring discussion today from leaders in life sciences and healthcare. Here are my takeaways: 1. We can “solve” biology. It is a deterministic process. Fiendishly complex - yes - but we’re building technology tools that crack that complexity. Expect patients to see changes within 5-10 years. 2. The scientific method has been the most powerful innovation in human history. AI is being inserted into that loop, which will result in exponential returns. 3. Our computers now work with our own human languages instead of computer code. They also work in the language of genetics, of proteins, of images and other data. They’re able to draw conclusions across all this data in a way the human mind cannot. The upcoming inflection points will be based upon these technology multipliers that help us do better work faster than ever before. Thanks to the panelists for their time and insight! Rory Kelleher (NVIDIA): Director of HCLS for the Americas Jeff Denworth, CoFounder, VAST Piotr Sliz, Chief Research Information Officer, Boston Children’s Hospital William Mayo, SVP of Research at Bristol Myers Squibb

Biostatistics Journal Club: Machine Learning for Imputation of Missing-not-at-Random Data Led by Boyu Ren of McLean Hospital, this talk will focus on Missing-not-at-random (MNAR) data and machine learning techniques used by researchers. The paper “Identifiable Generative Models for Missing Not at Random Data Imputation” will be discussed. Register: hvrdct.me/6g4

🚀 Exciting new insights in medical machine learning! Learn how driving down Poisson error can offset classification error in clinical tasks. Explore the latest research in this thought-provoking blog post on arXiv:2405.06065v1. Dive into the mathematics of error trade-off and practical examples in malaria diagnosis and quantitation. Don't miss out on this critical information for ML system development. Read the full post here: https://bit.ly/3QIKF75. #MedTech #MachineLearning #ClinicalTasks