Welcome to the LSU Department of Experimental Statistics

Seminar Announcement

Vasileios Maroulas

Associate Vice Chancellor for Research, Director of the AI Tennessee Initiative, Professor of Mathematics

University of Tennessee

"Grid Cell Navigation via Simplicial Convolutional Recurrent Neural Network"
November 8, 2024, 1:30pm to 2:20pm
145 Coates Hall

The brain's spatial orientation system relies on specific neuron groups for navigation, such as head direction cells for orientation and grid cells for mapping environments. These neurons work together in patterns, firing simultaneously to create directional and positional signals. To better understand and decode these patterns, we developed a new topological deep learning model that goes beyond traditional graph-based approaches. Our model, a simplicial convolutional recurrent neural network, uses topological structures to capture complex neural relationships. This method allows us to predict head direction and location from neural data without needing prior similarity measures, proving effective in head direction and trajectory prediction.

Seminar Announcement

Susan Murphy

Professor of Statistics and Computer Science

Harvard University

"Online Reinforcement Learning in Digital Health Interventions"
October 28, 2024, 1:30pm-2:20pm
134 Howe Russell East

In this talk, Murphy will discuss first solutions to some of the challenges we face in developing online RL algorithms for use in digital health interventions targeting patients struggling with health problems such as substance misuse, hypertension and bone marrow transplantation. Digital health raises a number of challenges to the RL community including different sets of actions, each set intended to impact patients over a different time scale; the need to learn both within an implementation and between implementations of the RL algorithm; noisy environments and a lack of mechanistic models. In all of these settings the online line algorithm must be stable and autonomous. Despite these challenges, RL, with careful initialization, with careful management of bias/variance tradeoff and by close collaboration with health scientists can be successful. We can make an impact!

Seminar Announcement

Michael Beck

Senior Chemist, Reference Atmospheres, LLC

"From Opticks to Optik: Parallel Histories in Statistics and Physics"

September 27, 2024, 1:30pm-2:20pm
145 Coates Hall

Statistics is a young discipline! One entertaining way to make this point is to map a chronology of landmark statistical publications in parallel with those of a more mature discipline, namely physics. What stands out is that physics transitioned from classical, Newtonian physics into modern physics during the same decades that saw the emergence of classical, or frequentist statistics.

All attending, will be eligible for dozens of door prizes (mainly used statistical books).

View September 27 Seminar Annoucement

Seminar Announcement

Tingting Hou

Department of Biostatistics, University of Florida

"Kernel Neural Networks for High-Dimensional Genetic Data Analysis"

April 26, 2024 at 1:30 p.m. to 2:30 p.m.

E134 Howe-Russell-Kniffen

The recent development of artificial intelligence (AI) technology, especially the advance of deep neural network (DNN) technology, has revolutionized many fields. While DNN plays a central role in modern AI technology, it has been rarely used in sequencing data analysis due to challenges brought by high-dimensional sequencing data (e.g., overfitting). Moreover, due to the complexity of neural networks and their unknown limiting distributions, building association tests on neural networks for genetic association analysis remains a great challenge.

In this talk, Hou will introduce recent efforts to address these challenges and fill the important gap of using AI in high-dimensional sequencing data analysis. In the first, Hou will introduce a new kernel-based neural network (KNN) test for complex association analysis of sequencing data. The test is built on our previously developed kernel-based neural network model (KNN) framework. Based on KNN, a MINQUE-based test is introduced to evaluate the joint association of high-dimensional genetic data with a disease phenotype of interest, considering non-linear and non-additive effects (e.g., interaction effects). In the second part, Hou will present a mixed neural network (MNN) framework for high-dimensional risk prediction research. MNN is a novel neural network approach that inherits features from both linear mixed models (LMM) and classical neural networks to model high-dimensional data and complex genotype-phenotype relationships. Through simulation studies, we demonstrate the advantages of MNN for high-dimensional risk prediction analysis in terms of robustness and accuracy.

View September 27 Seminar Annoucement

Seminar Announcement

Dr. Kai Kang

Research Scientist within Computer Science and Artificial Intelligence Laboratory (CSAIL)
Massachusetts Institute of Technology (MIT)

"Deciphering Biological Complexity: A Statistical Approach"

April 19, 2024 at 1:30 p.m. to 2:30 p.m.

E134 Howe-Russell-Kniffen

Understanding the intricate biological mechanisms underlying diseases, such as cancer and neurodegenerative disorders, is crucial for developing effective therapies. Genomics and transcriptomics offer invaluable insights into disease pathology. Next-generation sequencing (NGS) and third generation sequencing (TGS) ease high-throughput RNA/DNA sequencing, revolutionizing biomedical research. The unprecedented amount of data generated daily needs an interdisciplinary effort to address computational challenges. In this talk, Kang will discuss a recent project that highlights the challenges and opportunities in computational biology, highlighting how statistical approaches can lead to deeper biological insights.

He will share recent work on developing statistical methods that computationally untangle complex mixtures of cell types from bulk RNA sequencing data. A notable feature of their approach is its capability to simultaneously estimate individual-level, cell type-specific transcriptomics profiles and cell-type proportions. The method uses Bayesian hierarchical modeling to capture the heterogeneous nature of mixture data, employing a Markov Chain Monte Carlo procedure for parameter estimation and hyperparameter tuning. By deciphering cellular heterogeneity in complex tissues and improving downstream analyses of bulk RNA-seq data and spatial transcriptomics, the method holds significant promise for affecting biomedical research.

Seminar Announcement

Dr. Enayet Talukder

Retired from Pfizer

"Navigating the Complexities: Drug Development and Innovative Clinical Trial Designs for Next-Generation Oncology Therapies"

Friday, April 12, 2024, 1:30 pm – 2:30 pm

E134 Howe-Russell-Kniffen

This presentation will explore the prescription drugs development process, from initial discovery to market approval, focusing on clinical trial designs for each development phase for Oncology therapies. The landscape of cancer treatment is rapidly evolving. The rise of targeted cancer therapies necessitates more intricate study designs. Unlike traditional chemotherapies, these therapies have distinct mechanisms of action, demanding innovative approaches to design to capture their effects. The objective of a Phase 1 clinical trial for chemotherapy is to obtain the maximum tolerated dose (MTD), while a targeted therapy may seek an optimum biological dose (OBD). Similarly, tumor shrinkage (endpoint: tumor response) may not be an appropriate efficacy measure for a targeted therapy when it slows the disease progression and helps patients survive (endpoint: progression-free survival) longer. This presentation highlights the key considerations for designing clinical trials, from selecting the appropriate patient population to determining sample size (by controlling Type I and Type II Error probabilities). We will review real-life examples of innovative designs in cancer patients.

Seminar Announcement

Christopher Peters, MApStat

Chief Data Scientist at Touch, employee 9 and former Principal Data Scientist at Zapier

"AI-Driven Business Innovation: Harnessing Bayesian Multilevel Models and Probabilistic Programming at the Intersection of Statistics, Decision Theory, and Game Theory"

Friday, February 2, 2024, 1:30pm-2:30pm

E134 Howe-Russell-Kniffen

This presentation delves into the innovative integration of artificial intelligence (AI) with advanced statistical methods, decision theory, and game theory, emphasizing their application in business for enhanced profitability and strategic advantage. The focus is on Bayesian multilevel models and their pivotal role in powering AI systems alongside LLMs and other methods. Attendees will gain insights into how these models effectively manage complex data structures and their importance in AI applications. Additionally, the talk will shed light on probabilistic programming, illustrating its role as a critical link that unites statistical modeling with machine learning in business contexts. Moreover, the synergy of decision theory and game theory with AI and Bayesian methodologies will be discussed, underscoring how these combined approaches inform optimal decision-making and strategy development in competitive business environments. Practical examples and real-world case studies will be presented to demonstrate the transformative potential of these integrated methods in shaping the future of data-driven business strategies.

View February 2 Seminar Annoucement