Authors

Xueliang Yang, Xiaoyue Zhang, Weihao Xia, Wei Yu, Jia Dong

Research Question

How can PPGIS, sensor-based technologies, and advanced urban big data analytics and tools be integrated to effectively identify, monitor, and mitigate urban environmental stressors?

Background

Urban stress, intensified by rapid urbanization and environmental degradation, poses a growing threat to mental health. This project designs an experimental pipeline for collecting biosignals and environmental data using wearable sensors and mobile devices at a micro-level, and incorporates Public Participation Geographic Information System (PPGIS) methods to examine the spatial distribution of perceived stress at the macro level.

Methodology

Deep learning techniques were applied to train a model using the WESAD dataset, labeling biosignal data into stress, baseline, and relaxed states. This pre-trained model was then used to label biosignal data collected from an experiment conducted by the authors. These stress predictions were combined with environmental variables to identify contributing urban features and perform spatial analysis. PPGIS responses were categorized into favorite and stressful locations, spatially analyzed and clustered using Neural Information Processing Systems (NIPS) methods, and paired with satellite-derived variables such as air quality and heat exposure to understand stress distribution patterns across Accra.

Deliverables

• Publication on Micro-level Sensor Analysis
• Publication on Macro-level PPGIS with Remote Sensing
• Public Data Pipeline to assist future researchers’ data cleaning and modeling
• Technical Report
• ArcGIS StoryMap with video-based presentations

Datasets