CoDEx 2026 Research Talks
The following Research Talks will be presented at CoDEx 2026.
Sessions: 2:05 - 2:25 p.m.
Neutron Star Drag Race: Simulating Neutron Stars in a Common Envelope - Wildcat Room
Nicole Flors, PhD Student, Weinberg College of Arts and Sciences, et al.
Neutron stars are dense, rapidly rotating magnets that power many interesting astrophysical phenomena by funneling gas onto their surfaces and accelerating particles away at high energies. I perform large-scale simulations of fluid dynamics around neutron stars using high-performance computing, which will help us understand how gas flows in extreme physical environments and provide insights into the formation of events like gravitational waves.
iVox: Deep Interpretable Survival Prediction for Personalized Radiotherapy Dose Optimization - Lake Room
Sagnik Sarkar, Senior Research Technologist, Feinberg School of Medicine, et al.
Stereotactic body radiation therapy (SBRT) is a standard treatment for early-stage lung cancer, yet predicting local tumor recurrence remains challenging. Current clinical practice lacks quantitative tools to assess how patient-specific dose distributions influence individual outcomes, limiting opportunities for personalized treatment optimization.
Methods: A deep learning framework integrating multimodal imaging with survival analysis was developed to predict local failure risk following lung SBRT. The model was trained on a multi-institutional cohort of 822 patients, processing over 3,200 three-dimensional volumetric images (CT, dose, contours, fractionation) as four-channel 64³-voxel tensors. A 3D ResNet backbone extracts spatial features, with dual prediction heads providing radiomic auxiliary supervision (256 prognostically-selected biomarkers) and patient-specific log-hazard outputs. Training employed extensive data augmentation and GPU acceleration. Cox proportional hazards calibration converts predictions into absolute survival probabilities. Voxel-wise attribution maps are computed via Integrated Gradients (50 integration steps) and gradient backpropagation with SmoothGrad noise tunneling (32 samples), followed by spatial filtering to produce interpretable risk localization.
Results: The model achieved a concordance index of 0.73 on internal validation and 0.71 on an independent external test set, outperforming classical Cox baseline (C-index ~0.64). Attribution analysis revealed elevated hazard ratios in peritumoral regions for delivered dose, suggesting microscopic disease extension requiring dose escalation, while intratumoral CT features associated with recurrence risk reflect intrinsic biological susceptibility. The calibrated Cox framework establishes a quantitative relationship between prescription dose, geometric dose patterns, and projected local failure probability, enabling calculation of recommended dose modifications for target tumor control rates.
Conclusion: This framework bridges deep learning with interpretable survival analysis to provide personalized, spatially-resolved risk assessment for lung SBRT. By identifying high-risk regions and quantifying the dose-response relationship, the approach offers a pathway toward individualized radiotherapy planning guided by predicted treatment outcomes.
Designing Gen AI Interaction Paradigms to Support User Creativity in Music Production - Big Ten Room
Katherine O'Toole, PhD Candidate, School of Communication, et al.
Text-to-music AI tools make it possible for anyone to create music instantly, but how can we design AI systems that help people learn, experiment, and think more creatively, rather than simply generating outputs for them? This work introduces a computational framework for analyzing human–AI co-creativity in music generation, leveraging music information retrieval and text embedding models to directly compare the efficacy of interventions designed to help users learn how to intentionally explore different musical ideas and possibilities. This allows us to determine what kinds of interaction paradigms are most effective for co-creative systems that not only generate content, but also help users to learn more about the domain and how to explore it.
Sessions: 2:35 - 2:55 p.m.
Automatic Detection of Ultrasonic Vocalizations Using Deep Learning Methods: Exploring the Effects of Neonatal Hypoxia on Vocal Behavior in Rats - Wildcat Room
Yilan Wei, PhD Student, Department of Communication Sciences and Disorders, et al.
Rats can emit ultrasonic vocalizations (USVs) beyond the range of human hearing (>20 kHz), which are widely studied in research on communication, emotional changes, and language development. However, the short duration and high density of USV events make traditional manual annotation inefficient, limiting their application in large-scale behavioral studies. We are developing a deep learning-based automated detection tool to automatically detect USVs. We applied this method to behavioral analysis in a neonatal rat hypoxia model to explore the impact of hypoxia on early vocalization behavior. Preliminary results show that rats in the hypoxia group produced fewer USVs than those in the control group. These findings suggest that neonatal hypoxia may negatively affect early vocal behavior and social communication. This method improves data processing efficiency, demonstrating the potential of deep learning methods for USV research.
Comparative genomics of herpes simplex virus 2 isolated from a maternal-neonatal dyad reveals high consensus sequence homology as well as minor variant diversity - Lake Room
Reem Abu Rass, PhD, Feinberg School of Medicine, et al.
Neonatal herpes simplex virus (nHSV) disease occurs following mother-to-neonate transmission during childbirth and can be severe. However, little is known about the maternal viral population, how it differs from the neonatal viral population, what viral features contribute to successful transmission, and what new viral features arise in the infant. We performed deep whole-genome sequencing of HSV-2 clinical isolates obtained from a maternal-neonatal dyad to investigate viral population dynamics. Each clinical isolate consists of a viral population that has a level of genomic variability. Comparing the average sequence of each genomic population revealed near-identical viral genomes. However, analysis of alternative genomic sequences that make less than 50% of the viral population revealed maternal alternative populations that were not transmitted, and newly emerged neonatal viral populations. This work provides a first look at the genomic variation that arises during HSV-2 transmission, and a framework for future analysis of additional maternal-neonatal pairs.
Learning as Graph Traversal - Big Ten Room
Jacob Puthipiroj, PhD Student, School of Education and Social Policy
The spiral curriculum, proposed originally by Jerome Bruner, has long served as a powerful metaphor for revisiting ideas at increasing levels of complexity, but it has remained under-specified as a formal model of learning. In this paper, we develop a graph-theoretic account of the spiral curriculum, conceptualizing learning as traversal through a hierarchical, semantically structured network of factual and conceptual micro-units. We introduce quizbowl as a uniquely suited environment for studying why such a spiral is appropriate: its structured yet expansive canon and graded question difficulty naturally instantiate repeated returns to the same ideas at increasing depth. In our framework, “spiraling” can be expressed as parameterized movement upward in difficulty and outward across related concepts, enabling precise discussion of sequencing strategies. To make the model empirically tractable, we detail the creation of two tools: knowledge graph visualization to render the underlying structure, and graph traversal–based spaced repetition to operationalize spiraling over time. Together, these contributions offer a foundation for designing personalized curricula in a data-driven manner.
Sessions: 3:05 - 3:25 p.m.
CalibrAI: Evaluating and Tuning Safety-Usability Tradeoffs in LLM-Based Systems - Wildcat Room
Dheeptha Rai, Graduate Student, McCormick School of Engineering and Applied Science
Safety mechanisms in large language model-based systems are often evaluated using aggregate metrics that obscure how systems fail in practice. CalibrAI introduces a calibration-based evaluation framework that separates user behavior generation from outcome assessment, making false positive and false negative tradeoffs explicit across safety thresholds. The talk examines how evaluation design choices shape our understanding of risk in deployed AI systems, and why aggregate safety scores can conceal meaningful shifts in system behavior.
Ultrasound-Mediated BBB Disruption Promotes Microglial Reprogramming and Enhanced Endothelial Crosstalk in the Human Brain - Lake Room
Víctor Andrés Arrieta, Clinical Research Associate, Health and Biomedical Informatics (HBMI), Feinberg School of Medicine, et al.
The brain has a natural “protective wall” called the blood–brain barrier (BBB). It keeps harmful things out, but it also blocks many medicines from reaching brain tumors like glioblastoma (GBM). In a clinical trial, we used gentle ultrasound plus tiny bubbles in the bloodstream to briefly open this barrier in patients with recurrent GBM. A key advantage of our study is that we collected two matched tissue samples from each patient: one from the ultrasound-treated area and one from a nearby untreated area, only 45 minutes later.
In this talk, I’ll show how we used computation to combine several types of data, single-cell sequencing, gene-regulation data, spatial maps of where cells are in the tissue, and multiplex imaging, to build a clear picture of what changes right after BBB opening. We find a rapid immune response, especially in microglia, and stronger communication between immune cells and blood vessels. This “BBB-opening response signature” could help doctors better time drug delivery and design smarter combination therapies.
Human vs. Generative AI: A Comparison of Creative Strategies - Big Ten Room
Yulin Yu, Postdoc Scholar, Northwestern Institute on Complex Systems & Kellogg School of Management, et al.
Generative AI can now produce highly creative outputs—but does it create the same way humans do? In this talk, I analyze 5,000 human responses and matched outputs from 20 leading language models on the widely used Divergent Association Task, examining how ideas unfold step by step rather than relying only on final creativity scores.
Although average human and AI performance can appear similar, top human creators (the top 10%) remain substantially more creative than current AI systems, and their creative processes differ in meaningful ways. Humans explore more flexibly and broadly, while AI systems follow more optimized but structurally constrained paths. Understanding these differences helps us design human–AI systems that combine machine-scale search with uniquely human depth and originality.
Sessions: 3:35 - 3:55 p.m.
Contempt - Wildcat Room
Nathan Reitinger, Postdoctoral Fellow, Pritzker School of Law, et al.
If you believe what you read, contempt is a weapon for impetuous judges to punish parties for offending the judge or misbehaving in court. But this perception, shared by nearly all scholars, is built on a few headline-grabbing examples and reviews of published opinions. Relying on a dataset of every single document in every case filed in one year in a federal court—over 700 million words of text, upon which we used advanced machine learning OCR techniques combined with low-level qualitative coding practices—this Article finds that contempt is something completely different.
Probabilistic Phylodynamic Models for Viral Evolution Across Biological Scales - Lake Room
Seth Borrowman, Research Technologist, Feinberg School of Medicine, et al.
Our talk shows how probabilistic models can uncover hidden patterns in how viruses evolve and spread using large genomic datasets. By applying a flexible phylodynamic framework across scales – from infections within a host to international transmission – we improve understanding of disease processes and strengthen the evidence base for public health decisions.
Gait Analysis Through the Deployment of Markerless Motion Capture in Routine Clinical Practice - Big TenRoom
Irina Djuraskovic, PhD Student, Interdepartmental Neuroscience Program (NUIN), The Graduate School, et al.
How a person walks encodes critical information about neurological health, disease progression, and recovery potential. The motor recovery trajectory following neurological impairments and injuries is highly variable and heterogeneous, involving changes in both functional ability and movement quality. Despite this complexity, current clinical motor recovery measures rely on coarse performance-level outcomes – such as speed and endurance – providing “stopwatch” assessments that overlook movement quality and the underlying biomechanical strategies. To overcome these limitations, this work leverages the Markerless Motion Capture system implemented directly into clinical workflow to capture continuous detailed longitudinal gait kinematics in individuals across numerous clinical etiologies as they undergo their rehabilitation. This approach enables the first large-scale longitudinal characterization of motor recovery trajectories, with a level of temporal and biomechanical detail previously unattainable in traditional clinical settings.