MSI – Physical Sciences and Engineering

Grand Prize: Argyro Sasli
Poster: Listening for the Unknown: A Multimodal Anomaly Detection for Gravitational Waves
Abstract:
Current gravitational-wave detection frameworks rely primarily on matched filtering with waveform templates, which are highly effective for known sources but limited in detecting unexpected or poorly modeled signals. This work develops a multimodal anomaly detection framework designed to identify gravitational-wave events without predefined templates, offering a promising path toward model-agnostic discovery.

Runner-Up: Vivasvaan Aditya Raj
Poster: Machine-Learning Segmentation of Galaxy Mergers Using Volunteer Classifications
Abstract:
Galaxy mergers are key drivers of galaxy evolution but are difficult to identify at scale. This project develops a machine-learning pipeline that transforms volunteer classifications into segmentation masks and trains models to identify merger features and predict vote fractions. Using MSI for large-scale computation, the work lays the foundation for a Euclid-based merger catalogue and future studies of galaxy evolution processes such as gas flows and AGN feedback.

Runner-Up: Armand Lannerd
Poster: Thermodynamics of proton insertion across the perovskite-brownmillerite transition in La0.5Sr0.5CoO3-δ
Abstract:
Using density functional theory and machine-learning interatomic potentials, this study investigates protonation thermodynamics in La0.5Sr0.5CoO3−δ across structural phase transitions. Results show hydrogen insertion is thermodynamically favorable but produces unstable hydrogenated phases, providing a thermodynamic explanation for experimentally observed acid etching during electrochemical cycling in humid environments.

Data Science and AI in Sustainability, Health Care, and Social Sciences

Grand Prize: Minzhu Zhao
Poster: Reliability of Functional Connectivity for Estimating tDCS Treatment Response in Small fMRI Cohorts of Addiction
Abstract:
This study examines functional connectivity (FC) reliability and individualized treatment-effect estimation in a small-cohort tDCS addiction study. While graph-based network representations are structurally stable across resampling, this robustness does not translate into stable machine-learning treatment-effect estimates in small datasets, highlighting key limitations in clinical neuroimaging ML applications.

Runner-Up: John Bellamy
Poster: Bayesian Optimization Guided Lab-in-the-loop Design of Porous Molecular Materials
Abstract:
This work introduces a structure-aware Bayesian optimization framework that uses chemical fingerprints to guide experimental design of porous materials for carbon capture. The system efficiently explores large chemical spaces and identifies high-performing materials that may be unintuitive to human designers, demonstrating the value of AI in accelerating sustainable materials discovery.

MSI – Life Sciences and Engineering

Grand Prize: Pitchaya Santaivongchai
Poster: Mapping the Turkey Reproductive Tract: A Single Cell RNA Sequencing Atlas for Organoid Validation
Abstract:
To study avian influenza in turkey hens, this research develops turkey oviductal organoids validated through single-cell RNA sequencing. Computational subclustering and multi-modal integration reveal region-specific gene expression and viral receptor distribution consistent with native tissue, producing the first single-cell atlas of the turkey reproductive tract and advancing organoid-based infectious disease research.

Runner-Up: Thomas Hodder
Poster: A deep-learning enabled assay of heat stress resistance in C. elegans identifies novel geroprotective drugs
Abstract:
This work develops an AI-enabled imaging assay using C. elegans heat stress resistance as a predictor of lifespan extension. The system identifies 12 novel compounds that extend worm lifespan, offering a scalable approach for accelerating geroprotective drug discovery.

Runner-Up: Adam Genda
Poster: Single cell transcriptional atlas of banked leukemia samples reveals novel clustering of leukemia stem cells
Abstract:
This study constructs a single-cell transcriptomic atlas of AML, ALL, and MDS samples, revealing novel clustering of TP53-mutant versus wild-type leukemia stem cells. Findings are supported by differential expression and gene set enrichment analyses, offering new insights into leukemia heterogeneity and disease progression.