Human mobility and urban dynamics are the keys to understanding diversity and complexity in cities. With advancement of technologies, a significant amount of geospatial data is generated, shared, and analyzed. Big geospatial data analytics highlights the importance of geography in data-driven knowledge discovery. The goal of this dissertation is to progress the fundamental understanding of human mobility and urban dynamics by developing novel methodological frameworks and models that utilize micro-scale spatiotemporal big data and encourage new knowledge creation. To achieve this goal, this dissertation includes three studies that focus on developing new methods to understand human mobility, urban dynamics, and their interactions. In the first study (Chapter 2), I propose a new index measuring similarities between human mobility patterns from different data sources such as social media and traditional survey. The second study (Chapter 3) shifts focus to urban dynamics. Applying a series of spatiotemporal exploratory studies, an efficient method for examining spatiotemporal patterns at a micro-scale in restaurants is proposed. The third study (Chapter 4) investigates the relationships between human mobility and urban dynamics with a novel explainable deep learning approach enhancing predictivity and interpretability of neural network models within geographic context. Throughout this dissertation, a comprehensive framework for understanding complexity of human mobility and urban dynamics is suggested through incorporating detailed spatial data into big data analytic models from geographic perspectives. This will provide individuals and governments with fundamental knowledge for better decision-making associated with economic growth and development.

For vector-borne diseases dynamics in human-vector contact play an importance role, and human behavior including mobility and prevention behavior has been studied as a key contributor to drive the variations in host-vector contact. Human mobility can increase the human risk of exposure to vector-borne pathogens, while human prevention strategies effectively introduced to communities can reduce the risk of disease transmission and outbreak. Various modeling approaches from mathematical to simulation have been applied to examine the impacts of human mobility and prevention behavior on disease transmissions; however, conventional models are often limited by their coarser and aggregated representations in space, time, and human behavior. This study proposes a novel modeling framework (Activity-ABM) that integrate the Activity-Based Model and the Agent-Based Model. The proposed model incorporates individual-scale human mobility and prevention behavior at the level of both individuals and population into an epidemiology simulation, and is capable of representing complex virus transmission dynamics emerged from bottom-up human-vector interactions across space over time. Such individual-scale investigation also allows artificial laboratory experiments to explore scenarios with the intervention strategies associated with human mobility and prevention behavior and to examine the effectiveness of the interventions in the local context. The activity-based model embedded in the Activity-ABM creates heterogeneous daily movement of all individuals in a population from travel surveys and demographic data. The Activity-ABM then generates prevention behavior of all individuals in the population, including the host-based controls, the vector-based controls, and the host-vector contact-based controls, and examines the disease transmissions engaged in the generated movement and prevention behavior. As a case study, this research takes the 2016 Zika outbreak event in Miami-Dade County, Florida. Our modeling results suggest that the vector-based controls are the most effective, as the 2016 Zika outbreak in Miami-Dade County ended after an outdoor spraying to control mosquitoes. With the emphasis on the role of human mobility and prevention behavior to disease transmissions, this dissertation research makes theoretical contributions to the literature on behavioral geography, complex systems, and spatial epidemiology. The proposed Activity-ABM is applicable to simulate other infectious disease transmission dynamics, and can be extended to a general disease.

Background

Methods

Results

Conclusion