Wanyu Huang

     Wanyu Huang
     Ph.D. Student
     Department of Civil and Systems Engineering
     Johns Hopkins University

     Email: wanyu.huang at jhu dot edu
     3400 North Charles Street, Baltimore, MD

     Resume / Google Scholar

About me

I am currently a Ph.D. student in the Department of Civil and Systems Engineering, Johns Hopkins University, advised by Takeru Igusa. Before joining Johns Hopkins University, I obtained my M.S. degree in Industrial Engineering from Iowa State University and B.E. degree in Automation from Shanghai Jiao Tong University.

Research

• My research interests lie in the field of statistical approaches for simulating real-world complex systems.
• My current research topics include: 1) systems dynamics and agent-based model for healthcare intervention implementations, 2) machine learning for studying maternal and children health.

Skills

• Python, R, MATLAB, NetLogo, SQL, HTML, Tableau

Selected Publications

• In-utero co-exposure to toxic metals and micronutrients on childhood risk of overweight or obesity: new insight on micronutrients counteracting toxic metals
  Wanyu Huang, Tak Igusa, Guoying Wang, Jessie P. Buckley, Xiumei Hong, Eric Bind, Andrew Steffens, Jhindan Mukherjee, Douglas Haltmeier, Yuelong Ji, Richard Xu, Wenpin Hou, and Xiaobin Wang.
  International Journal of Obesity, 2022.[Nature]
  
• Agent-based modeling for implementation research: An application to tobacco smoking cessation for persons with serious mental illness
  Wanyu Huang, Chia-Hsiu Chang, Elizabeth A. Stuart, Gail L. Daumit, Nae-Yuh Wang, Emma E. McGinty, Faith B. Dickerson, and Takeru Igusa.
  Implementation research and practice, 2021.[SAGE Pub]
  
• Analyzing Residential Weatherization Decisions using Hybrid Simulation Modeling
  Wanyu Huang, Caroline C. Krejci, Michael C. Dorneich, Ulrike Passe and Jacklin Stonewall.
  Building Simulation, 2019.[SpringerLink]
  
• Weatherization Adoption in A Multilayer Social Network: An Agent-based Approach
  Wanyu Huang, Caroline C. Krejci, Michael C. Dorneich and Ulrike Passe.
  Computational Social Science Conference, 2017.
  Best paper finalist.[ACM DL]
  

Projects

• Scale-Up of Evidence-Based Tobacco Smoking Cessation Treatment in Community Mental Health Clinics
  Johns Hopkins University - ALACRITY Center
  
  • Led the development of conceptual and numerical models for simulating smoking cessation intervention in vulnerable populations with serious mental illness, incorporating behaviors and interactions of patients, health providers, site administrators, and policy makers. One paper was published (first author)
  • Employed regression methods to identify determinants of smoking abstinence behavior and generated dynamic prediction models of abstinence outcome based on intervention interim performances, achieving high performance scores of 0.8 as measured by area under the ROC curve
  • Proposed an adaptive intervention strategy for health providers and site administrators to adjust treatments based on smokers' estimated abstinence likelihoods, and conducted a cost-benefit analysis to determine the optimal time to intervene due to the trade-off between prediction accuracy and treatment duration. One working paper (first author)
• Assessment of Group Counseling Therapies in Smoking Cessation Intervention
  Johns Hopkins University - ALACRITY Center
  
  • Proposed and led a study on the impact of group sessions on smoking behavior, which was inspired by a focus group with a therapist at a Baltimore clinic who expressed patients' intentions to attend group counseling sessions
  • Employed an opinion dynamics model to assess the impact of group sessions on participants' smoking behavior, which was modified to fit the trial setting and was recognized by clinical experts
  • Identified a statistically significant trend that individual participants' smoking behavior approached the group average level and was more influenced by peers who performed better (smoked less) compared to those who smoked more
  • Demonstrated strong critical thinking skills by applying social network theorems and analytical techniques to this project and is highly acknowledged by other members. One working paper (first author) and one NIH proposal
• Machine Learning on Early Risk Assessment of Autism
  Johns Hopkins University
  
  • Developed a personalized prediction and prevention model for Autism Spectrum Disorder (ASD) using logistic regression, elastic net, and multilayer perceptron techniques
  • Conducted feature selection using stepwise logistic regression or elastic net to identify 8 significant predictors for ASD
  • Achieved over 0.7 AUC using only the 8 factors collected at birth
• Effects of Maternal Toxic Co-exposures and Micronutrients Intake on Offspring's Obesity
  Johns Hopkins University
  
  • Conducted research to investigate the association between in-utero exposure to toxic metals and childhood overweight or obesity, utilizing a machine learning tool, Bayesian kernel machine regression, to estimate joint health effects of multiple exposures
  • Identified a positive dose-response association between maternal co-exposure to toxic metals and childhood overweight or obesity, with the association being more pronounced in children born to overweight/obesity mothers
  • Published research (first author) in a top 3 journal among 4043 journals on the Obesity topic
• Childhood Body Mass Index Trajectory
  Johns Hopkins University
  
  • Led a research project on identifying children's BMI trajectories and their early-life determinants, with a focus on addressing the challenge of longitudinal BMI records with different amounts, lengths, and intervals
  • Designed and executed a data analysis plan that incorporated a time-limit scheme for data imputation, allowing observations only at limited time windows to impute missing data. Utilized big data processing and analyzing skills to effectively manage and analyze large longitudinal datasets
  • Applied time-series K-means cluster analysis and latent class growth analysis to identify four distinct growth patterns and investigated their causes and effects. One paper (first author) is under review
• Innovative Weatherization Adoption Decision Simulations
  Iowa State University - Sustainable Cities Team
  
  • Led development of agent-based models to study energy conservation decision-making processes in urban neighborhoods and their impact on energy-related outcomes, using building energy simulation, agent-based modeling, and social network modeling
  • Proposed a novel agent-based model that incorporated a multilayer social network and opinion dynamics based on the Susceptible-Exposed-Infected-Recovered epidemic model to simulate residents' socially motivated energy conservation decision
  • Conducted iterative model refinement through weekly discussions with my advisor and monthly meetings with the team to obtain feedback and improve the model
  • Published one conference paper in the Computational Social Science Conference and was a finalist for the Best Paper award, as well as one journal paper as the first author, based on this work
• Development of Expansion Planning Methods and Tools for Handling Uncertainty
  Iowa State University
  
  • Developed a novel iterative method for identifying a small number of high-quality scenarios in generation expansion by using linear programming and bi-level optimization
  • Applied the methodology to a case study on the U.S. eastern and western interconnections, demonstrating the effectiveness of the scenario selection algorithm with a more accurate estimation of the economic value of transmission plans
  • Gained valuable experience in developing solutions to complex problems in the energy sector, and developed expertise in optimizations and addressing uncertainty in generation and transmission planning
• Information Diffusion in Social Networks
  Shanghai Jiao Tong University - Complex Network and Control Lab
  
  • Developed two innovative models for studying information diffusion in social networks. One model incorporated neighborhood influence and user activity to infer message-forwarding behavior, and the other utilized a genetic-algorithm-based approach to simulate gradual changes in message content and popularity over time
  • Led the code implementation, ensuring they aligned with our research objectives and effectively captured information propagation mechanisms in social networks
  • Validated the models' effectiveness by applying them to real-world Enron email datasets, achieving promising results and published two papers with a Ph.D. student