Spatio-temporal analysis

Understanding the spatiotemporal evolution of opioid overdose events using a regionalized sequence alignment analysis

The latest paper from the Franklin County Opioid Crisis Activity Level (FOCAL) mapping project, led by my former student Dr. Yuchen Li, in collaboration with Dr. Ayaz Hyder from OSU College of Public Health.

Li, Y., Miller, H.J., Hyder, A. and Jia, P. (2023) “Understanding the spatiotemporal evolution of opioid overdose events using a regionalized sequence alignment analysis.” Social Science & Medicine, p.116188.

Abstract

Background.  Opioid overdose events and deaths have become a serious public health crisis in the United States, and understanding the spatiotemporal evolution of the disease occurrences is crucial for developing effective prevention strategies, informing health systems policy and planning, and guiding local responses. However, current research lacks the capability to observe the dynamics of the opioid crisis at a fine spatial-temporal resolution over a long period, leading to ineffective policies and interventions at the local level.

Methods. This paper proposes a novel regionalized sequential alignment analysis using opioid overdose events data to assess the spatiotemporal similarity of opioid overdose evolutionary trajectories within regions that share similar socioeconomic status. The model synthesizes the shape and correlation of space-time trajectories to assist space-time pattern mining in different neighborhoods, identifying trajectories that exhibit similar spatiotemporal characteristics for further analysis.

Results. By adopting this methodology, we can better understand the spatiotemporal evolution of opioid overdose events and identify regions with similar patterns of evolution. This enables policymakers and health researchers to develop effective interventions and policies to address the opioid crisis at the local level.

Conclusions. The proposed methodology provides a new framework for understanding the spatiotemporal evolution of opioid overdose events, enabling policymakers and health researchers to develop effective interventions and policies to address this growing public health crisis.

Keywords: Opioid overdose epidemic; Sequential analysis; Neighborhood context; Geographic information science; Spatiotemporal pattern mining

Realizable accessibility: Evaluating the reliability of public transit accessibility using high-resolution real-time data

New paper: Liu, L., Porr, A. and Miller, H.J. (2023) “Realizable accessibility: Evaluating the reliability of public transit accessibility using high-resolution real-time data,” Journal of Geographical Systems, 25, 429-451.

Abstract. The widespread availability of high spatial and temporal resolution public transit data is improving the measurement and analysis of public transit-based accessibility to crucial community resources such as jobs and health care. A common approach is leveraging transit route and schedule data published by transit agencies. However, this often results in accessibility overestimations due to endemic delays due to traffic and incidents in bus systems. Retrospective real-time accessibility measures calculated using real-time bus location data attempt to reduce overestimation by capturing the actual performance of the transit system. These measures also overestimate accessibility since they assume that riders had perfect information on systems operations as they occurred. In this paper, we introduce realizable real-time accessibility based on space–time prisms as a more conservative and realistic measure. We, moreover, define accessibility unreliability to measure overestimation of schedule-based and retrospective accessibility measures. Using high-resolution General Transit Feed Specification real-time data, we conduct a case study in the Central Ohio Transit Authority bus system in Columbus, Ohio, USA. Our results prove that realizable accessibility is the most conservative of the three accessibility measures. We also explore the spatial and temporal patterns in the unreliability of both traditional measures. These patterns are consistent with prior findings of the spatial and temporal patterns of bus delays and risk of missing transfers. Realizable accessibility is a more practical, conservative, and robust measure to guide transit planning.

Spatiotemporal patterns of bus operation delays in Columbus, Ohio, USA

New publication: Park, Y., Mount, J., Liu, L., Xiao, N. and Miller, H.J. (2020) “Assessing public transit performance using real-time data: Spatio-temporal patterns of bus operation delays in Columbus, Ohio, USA,” International Journal of Geographical Information Science, 34, 367-392.

ABSTRACT: Public transit vehicles such as buses operate within shared transportation networks subject to dynamic conditions and disruptions such as traffic congestion. The operational delays caused by these conditions can propagate downstream through scheduled transit routes, affecting system performance beyond the initial delay. This paper develops an approach to measuring and assessing vehicle delay propagation in public transit systems. We fuse data on scheduled bus service with real-time vehicle location data to measure the originating, cascading and recovery locations of delay events across space with respect to time. We integrate the resulting patterns to construct stop-specific delay propagation networks. We also analyze the spatiotemporal patterns of propagating delays using parameters such as 1) transit line-based network distance, 2) total propagating delay size, and 3) distance decay. We apply our methodology using publicly available schedule and real-time location data from the Central Ohio Transit Authority (COTA) public bus system in Columbus, Ohio, USA. We find that delay initiation is spatially and temporally uneven, concentrating on specific stops in downtown and specific suburban locations. Core stops play a critical role in propagating delays to a wide range of connected stops, eventually having a disproportional impact on the on-time performance of the bus system.

KEYWORDS: Mobility, urban applications, public bus delay propagation, public transport reliability assessment, spatio-temporal data modelling