Student Seminar by Dr. Qihao Wu on Oct 06, 2026, 3:00PM

Title: Data-driven Prediction and Dynamic Decision Making for Hospital Operations

Speaker: Dr. Qihao Wu (Rotman School of Management, University of Toronto; Department of Data and Systems Engineering, HKU)

Date: Oct 06, 2026 (Tuesday)

Time: 3:00 pm - 4:00 pm

Venue: Tam Wing Fan Innovation Wing Two, G/F, Run Run Shaw Building, HKU 

Registration Link: https://forms.cloud.microsoft/r/1TtEiDWygf

ITS Student Committee will provide snacks and beverages for registered participants.

Abstract:

Modern healthcare operations are becoming increasingly complicated. Improving the efficiency and the quality of healthcare delivery is a significant research agenda that advances public welfare. This work aims to bridge the research gap of deploying both OR and AI solutions in prediction and decision-making tasks for hospital operations, with particular considerations about explainability, robustness, equity, and the curse of dimensionality. At macrolevel, patient boarding across Hong Kong public hospitals during the Omicron surge has been evaluated and interpreted.

Vector autoregression with lagged variables is used to predict patient waiting times and analyze the causality. At mesolevel, a robust patient-redirecting policy for coordinating multiple inpatient units has been proposed to address the imbalance between patient demand and healthcare resource supply. With the structural properties of the formulated constrained Markov decision process (MDP), the distributionally robust optimization can be efficiently solved.

At microlevel, an effective outpatient scheduling scheme that ensures equitable access to care for heterogeneous patients has been developed based on MDP and an optimal stopping mechanism. The contributions can be summarized into two aspects. Methodologically, the novel OR and AI solutions developed for healthcare operations, such as explainable AI frameworks, constrained MDPs, and model-informed RL, can also be applied to other domains, particularly in terms of explainability, robustness, equity, and large-scale optimization. Regarding healthcare applications, this work comprehensively traverses macrolevel, mesolevel, and microlevel. The key findings, such as elderly admission leading to patient boarding, robust inpatient redirect policies, and equitable outpatient scheduling, serve as managerial implications for stakeholders' decision support.

Bios: Dr. Wu is joining the Rotman School of Management as a postdoctoral fellow, with a cross-appointment in the Department of Psychiatry, University of Toronto. He received his Ph.D. in the Department of Data and Systems Engineering from HKU, his M.Sc. from Imperial College, and a dual B.Eng. (honours) from Beijing Jiaotong-Wollongong joint program. Qihao’s research focuses on healthcare management, OR, and AI, particularly data-driven operations and decisions in complex systems.

Student Seminar by Ms. Ziyi He on Sep 23, 2026, 2:30PM

Title: Uneven Housing-Market Capitalization of Public Electric Vehicle Chargers in Los Angeles

Speaker: Ms. Ziyi He (Department of Geography)

Date: Sep 23, 2026 (Wednesday)

Time: 2:30 pm - 3:30 pm

Venue: 1104, 11/F Jockey Club Tower, HKU

Registration Link: https://forms.cloud.microsoft/r/rkG5PXWrnG

ITS Student Committee will provide a beverage for registered participants.

Abstract:

Public EV charging is typically treated as mobility infrastructure, yet its accessibility may also be capitalized into housing values. Where these premiums emerge remains poorly understood. Using housing transactions in Los Angeles during 2010-2024, we constructed a network-based Generalized Two-Step Floating Catchment Area measure of public charger accessibility. We then estimated property and time fixed-effects models within a continuous-exposure difference-in-differences design, exploiting within-property changes in charging accessibility. Across preferred specifications, a one-unit increase in the log-transformed accessibility index is associated with a 1.2%–1.6% increase in property value. Capitalization is stronger in commercial and mixed-use zones and varies nonlinearly with neighborhood income and racial composition, weakening at the most affluent and highest-White-share margins. Public charging deployment should therefore assess effective access, land-use context, and potential housing-affordability consequences. 

Bios: Ziyi is currently an MPhil student in Prof. Junshi Xu’s group. She received a B.Econ. in Economics from HKU in 2026. Her current research focuses on how transportation shapes urban accessibility, environmental externalities, housing markets, and distributional outcomes.

Distinguished Transport Lecture by Prof. Srinivas Peeta, 06 Oct 2026 (Tue) 21:00 (HKT)

AI-enabled Travel Choice Modeling and Autonomous Driving Decision-Making 

SPEAKER Professor Srinivas Peeta Frederick R. Dickerson Chair and Professor

School of Civil and Environmental Engineering

H. Milton Stewart School of Industrial and Systems Engineering

Georgia Institute of Technology ​ DATE AND TIME 06 Oct (Tuesday) 2026, 21:00 – 22:00 (Hong Kong Time) REGISTRATION Please register by using this link: https://forms.cloud.microsoft/r/B5NbjzsnYy Confirmation emails with ZOOM link will be sent to successfully registered participants. This Distinguished Transport Lecture will be held on ZOOM only. ORGANISED BY Institute of Transport Studies, The University of Hong Kong ​ ABSTRACT Advances in large language models (LLMs), vision-language models, and other foundation models are creating new opportunities for travel choice modeling and autonomous driving through their capabilities for information synthesis, contextual reasoning, and complex decision support. However, their outputs can be inconsistent, difficult to verify, and uncertain, creating challenges for transportation applications where decisions must be interpretable and reliable. This talk will discuss approaches that integrate Generative AI methods with travel choice models, traffic rules, human driving behavior, and autonomous driving decisions. Applications will span travel choice modeling, where LLMs support the development of explicit choice models, and autonomous driving, where foundation models are combined with traffic-rule representation and uncertainty-aware driving decisions. These approaches illustrate how AI can be integrated with domain knowledge to develop interpretable travel choice models and support safer, context-aware autonomous-driving decisions. SPEAKER’S BIO Dr. Srinivas Peeta is the Frederick R. Dickerson Chair and Professor in the School of Civil and Environmental Engineering and the H. Milton Stewart School of Industrial and Systems Engineering at Georgia Institute of Technology. He is also Principal Research Faculty at the Georgia Tech Research Institute. He is the Co-Editor-in-Chief of the journal Transportation Research Part B: Methodological. He was a past Chair of the Transportation Network Modeling Committee of the Transportation Research Board (TRB) of the U.S. National Academies. Prior to Georgia Tech, he was the Jack and Kay Hockema Professor at Purdue University up to 2018, where he served as the Founding Director of the NEXTRANS Center, the USDOT Region 5 University Transportation Center (2006-2018). Dr. Peeta has authored over 530 technical publications, including about 460 in peer-reviewed journals and conference proceedings. He has received 10 best paper awards and 7 students have received best dissertation awards. His research interests are multidisciplinary and broadly span transportation and infrastructure systems as well as several methodological domains. He received his B.Tech., M.S. and Ph.D. degrees from the Indian Institute of Technology Madras (IIT Madras), Caltech, and The University of Texas at Austin, respectively. He is a Fellow of ASCE and a Distinguished Alumnus of IIT Madras.