January 30, 2026 1:00 pm
In this quarterly IBT/ABC-UTC Research Seminar, University of Washington professors Marc Eberhard (Civil and Environmental Engineering) and John Choe (Industrial and Systems Engineering) present work related to the Use of Machine Learning to Predict Key Bridge Properties. This work supports the improvement of Bridge Asset Management practices by identifying subsets of bridges most likely to have seismic vulnerabilities.
presentation begins: 5:28
Q & A begins: 38:48
Presentation Documents:
IBT-ABC January 2026 Quarterly Research Seminar Presentation Slides
Intro & News Slides
Description:
We explore the use of machine learning (ML) to predict key structural characteristics of bridges using data from the U.S. National Bridge Inventory (NBI). The publicly available NBI (2023) documents more than 100 characteristics of approximately 617k bridges in the United States, but few characteristics directly relate to structural or earthquake engineering. To provide ground-truth values, key structural characteristics not documented in the NBI were extracted from structural drawings of 796 bridges within Washington State. Using the corresponding NBI characteristics (features) and extracted properties (targets), we trained and evaluated standard ML algorithms and an automated ML (AutoML) approach using AutoGluon to develop predictive models for four target characteristics. We demonstrate that ML can feasibly generate a more detailed and comprehensive bridge inventory, supporting regional-scale planning, resource prioritization, and post-earthquake response. The predicted characteristics were accurate enough to improve the identification of shear-critical columns. The practical value of quick identification lies in enhancing resilience by enabling rapid decision-making, such as prioritizing pre-earthquake mitigation and post-earthquake inspections, particularly when reviewing plans for every bridge in a region is not feasible.
Presenters:
Marc Eberhard, Ph.D.
Professor
Department of Civil & Environmental Engineering
University of Washington, Seattle
Email: eberhard@uw.edu
Marc Eberhard received his Bachelor of Science in Civil Engineering from the UC Berkeley in 1984. After working for the Bridge Design Division of the California Department of Transportation, he attended the University of Illinois at Urbana-Champaign, where he received his Master’s Degree (1987) and PhD (1989). Marc Eberhard teaches course on structural analysis, reinforced and prestressed concrete structures, and earthquake engineering. His current research focuses on the rapid construction and performance of reinforced and prestressed concrete building and bridges, subjected to gravity loads, earthquakes and tsunamis.
John Choe, Ph.D.
Associate Professor
Department of Industrial & Systems Engineering
University of Washington, Seattle
Email: ychoe@uw.edu
Dr. John Y. Choe is the Director of the Disaster Data Science Lab and the Deputy Director of the Center for Disaster Resilient Communities. He received his Ph.D. in Industrial & Operations Engineering (Concentration: Quality Engineering & Applied Statistics) and M.A. in Statistics from the University of Michigan, Ann Arbor. He holds bachelor’s degrees in Physics and Management Science from KAIST in Korea.
Presentation Graphics:
Figure 1. Machine learning framework overview.
Figure 2. Text embedding framework overview.