Seminar Series

Date: Tuesday, April 21, 2026 | 12:00PM-1:00PM, BMI Classroom 4004 or on Zoom
Speakers: Camilo Valderrama, PhD
                 Assistant Professor, Applied Computer Science
                 The University of Winnipeg (Canada)

Abstract: Birth weight (BW) is a critical measure of infant health and nutritional status. Among the factors influencing BW, maternal height is one of the strongest, as taller mothers tend to deliver heavier newborns. In the US, infants born to Caucasian mothers, who, on average, are taller than mothers of other ethnicities, are more likely to exceed the World Health Organization’s (WHO) low birth weight (LBW) cutoff of 2,500 g. Since this threshold was originally established using predominantly Caucasian populations, its universal applicability across diverse ethnic groups is questionable. To address this, we investigated alternative methods for deriving LBW cutoffs that consider maternal differences. Using the 2022 CDC birth dataset, we conducted a two-stage analytical approach. First, we applied conditional inference trees and a fuzzy inference model to examine the relationship between maternal height, BW, and Apgar scores. Second, we used an ensemble of five machine learning regressors to estimate BW thresholds associated with normal Apgar outcomes. Our results indicate that under the fixed WHO standard, the likelihood of achieving a normal Apgar score decreases as maternal height increases. This suggests that some infants above the 2,500 g cutoff may still face health risks. We conclude that an adjustable LBW threshold incorporating maternal height is essential for more accurate newborn risk assessment and improved health outcomes.

Date: Tuesday, April 7, 2026 | 12:00PM-1:00PM, BMI Classroom 4004 or on Zoom
Speakers: Patricia C. Dykes, PhD, MA, RN, FAAN, FACMI
                  Professor and Executive Director of the Center for Data Science
                  Nell Hodgson Woodruff School of Nursing
                  Emory University

Abstract: Preventable patient harm remains a major challenge despite widespread adoption of electronic health records. This seminar will present a program of research that leverages clinical insight, data science, and human-centered design to develop digital tools that make safety risks visible and actionable. Through four case studies, Fall TIPS, CONCERN, DOVE, and the ongoing OPTIONS trial, Dr. Dykes will illustrate how workflow-integrated clinical decision support, artificial intelligence, and electronic clinical quality measures can detect risk earlier, engage patients and clinicians, and prevent harm across care settings. The talk will highlight lessons in implementation, scalability, and the role of informatics in translating frontline clinical knowledge into deployable solutions that improve patient safety.

Date: Tuesday, March 17, 2026 | 12:00PM-1:00PM, BMI Classroom 4004 or on Zoom
Speakers: Sean Alexander Ridout, PhD
                 Postdoctoral Fellow
                 Department of Physics, Emory University

Abstract: Diabetes, a disease characterized by high blood glucose, is the 8^th leading cause of death worldwide. I will discuss my ongoing work using mathematical and biophysical models to understand healthy glucose regulation and diabetes. In one line of work, we propose and test a mathematical model for the pathogenesis of ketosis-prone diabetes, a subtype of diabetes with unusually fast dynamics. By fitting our model to data from individual patients, we identify patient-specific rates of disease processes and propose optimized treatment protocols. In another project, I am working to understand the design principles of healthy glucose control. I will show how spatial variation in glucose concentrations places strong constraints on glucose regulation, revealing one of the basic principles which has forced evolution to regulate blood glucose using hormones.

Date: Tuesday, March 3, 2026 | 12:00PM-1:00PM, BMI Classroom 4004 or on Zoom
Speakers: Jenni Seale Reiff, PhD, OTR/L 
                 Assistant Professor, Geriatrics and Gerontology, 
                 Emory University, School of Medicine

Abstract: Preoperative rehabilitation, or prehabilitation, is an emerging preventive approach to rehabilitation service delivery that is intended to prepare patients physiologically, psychologically, and logistically to recover well from an anticipated elective surgery. While prehabilitation has been broadly recommended to support vulnerable populations such as older adults and cancer patients in preparing for elective surgery, small underpowered sample sizes in trial and pilot data limit capacity to assess both effectiveness and generalizability, resulting in a lack of consensus regarding an optimal prehabilitation protocol. There is growing interest in trying to leverage secondary data (e.g., administrative claims, medical records) to study prehabilitation, but data limitations pose a challenge to measurement. In claims, there is no dedicated procedural code designating that the service was delivered with the purpose or intent of preparing the patient for surgery. Several recently published studies attempting to study prehabilitation using administrative claims have defined prehabilitation using different criteria centered on preoperative timing and a defined set of procedural and diagnostic code criteria, but the risk for false-positive misclassification using this method is high, and the sensitivity and specificity of these varying criteria is unknown. An integrated delivery system and extensive linkable data infrastructure in the Veterans Affairs Health Care system offer an opportunity to leverage electronic health record data as a gold standard comparison to validate a claims-based algorithm for prehabilitation if methods can be devised to verify preoperative therapy session content (often recorded in free-text components) as prehabilitation in EHR data.

Date: Tuesday, February 17, 2026 | 12:00PM-1:00PM, BMI Classroom 4004 or on Zoom
Speakers: Anqi Wu, PhD
                 Assistant Professor, School of Computational Science and Engineering (CSE)
                 Georgia Institute of Technology

Abstract: When a mouse freely explores its environment, it continuously perceives sensory inputs, selects actions, and generates behavior. Understanding how these processes unfold and are coordinated in the brain has long been a central goal in neuroscience. Historically, this problem has been studied primarily in highly controlled experimental settings, with constrained tasks, simplified stimuli, and stereotyped behaviors. While these approaches have yielded important insights, extending them to naturalistic scenarios introduces substantial new modeling challenges.

Naturalistic behavior is inherently more complex, involving self-paced, unstructured, and variable actions that unfold across multiple timescales. In this talk, I will present our recent efforts to develop reinforcement learning and inverse reinforcement learning–based models that capture this complexity and uncover meaningful cognitive and motor representations linked to underlying brain function. Although the work is grounded in animal behavior, the modeling framework is designed to be general, emphasizing principles that extend beyond species or task boundaries.

More broadly, emerging experimental paradigms increasingly rely on rich, multidimensional stimuli and spontaneous behavior, calling for a shift in how neural and behavioral data are analyzed. By advancing data-driven approaches for freely moving, naturalistic settings, this work aims to contribute to a more general understanding of the neural basis of cognition and behavior, with direct relevance to both animal and human neuroscience.

Date: Tuesday, February 3, 2026 | 12:00PM-1:00PM, BMI Classroom 4004 or on Zoom
Speakers: Satish E. Viswanath, PhD
               Associate Professor, Pediatrics and Biomedical Engineering, Emory University

Abstract: Developing artificial intelligence (AI) schemes to assist the clinician towards enabling precision medicine approaches requires development of objective markers that are predictive of disease response to treatment or prognostic of longer-term patient survival. The solutions being developed in my group in this regard involve designing computational imaging features together with histology or molecular data for detailed tissue and disease characterization in vivo as well be associated with patient outcomes. The key innovation in this approach lies in “handcrafting” unique tools that can capture biologically relevant and clinically intuitive measurements from routinely acquired imaging (MRI, CT, PET) or digitized images of tissue specimens. Further, by conducting cross-scale associations between imaging, pathology, and -omics, we can not only “unlock” and integrate the information captured by these different, disparate data modalities but also develop an interpretable and intuitive understanding of what drives their performance. Specific problems addressed via the new computerized imaging markers we have developed include: (a) predicting response to treatment to identify optimal therapeutic pathways, as well as (b) evaluating therapeutic response to guide follow-up procedures. We will further examine how to account for differences between sites, scanners, and acquisition parameters to ensure generalizable performance of AI tools and computational imaging features; crucial for wider clinical translation and widespread adoption. These will be discussed in the context of clinical applications in colorectal and renal cancers, digestive diseases, as well as pediatric conditions.