University of Twente Student Theses

Login

Spatial Temporal Dynamics in disease data

Temu, Editha Didas (2010) Spatial Temporal Dynamics in disease data.

[img] PDF
1MB
Abstract:Identification of a disease patterns is an important aspect in epidemiology studies. The study of disease patterns comprises of spatial, temporal and spatial-temporal patterns of a disease. Several methods exist to uncover disease patterns. These methods can be classified into retrospective and prospective studies. In retrospective studies one can study about the history of a disease. This study can be grouped into visualization studies, clustering analysis and correlation studies to find the explanatory variables influencing the detected pattern. Prospective analysis is performed for an earliest detection of a future outbreak particularly for infectious disease. This can be performed by either simulation models such as agent based simulation models or cellular automata. The major challenge of using these models is the validation. This research aims to find whether spatial temporal patterns that exist in empirical data can be used for validation of an agent based simulation model by using pertussis dataset. The spatial layers of interest of this research include, European countries, the Netherlands at municipality level, Twente regions at the municipality and postcode level and Enschede at the postcode level. Temporal extent was year 1993 to year 2004. Methods used in this research are the mathematical formula which was used to define epidemic and endemic years of the European counties. At European level, the analysis of synchrony and travelling waves were performed. Week rank method was used to identify hierarchical pattern of spread of pertussis at in the Netherlands at the municipality level, Twente region at municipality and postcodes and Enschede at postcodes. Space-time scan statistic was used to detect spatial temporal clusters which was further categorised to see if there is hierarchical spread pattern that can be compared to week rank. At the European countries, we observe synchrony and travelling waves of epidemic for some of the countries. No global patterns observed for this. Week rank method identify the hierarchical pattern In the Netherlands at the municipality level but this pattern could not observed in Twente region or Enschede city. Moreover space-time statistic detects statistical significant clusters of an epidemic and endemic years. These clusters were analysed and resulted to 7 clusters which originated in urban areas against 23 clusters which originates in rural areas during epidemic year while during endemic year 6 clusters originates in urban areas and 16 clusters originated in rural areas. This findings fails to validate the agent based simulation models since for validation to take place multiple patterns should be observed in different hierarchical level. The observed pattern should be used as a guidance to validate agent based simulation models. With this patterns in the simulated data should match with patterns in the empirical data. Further research is recommended to use different methods to identify patterns in a disease data. Key words: Synchrony, travelling waves, week rank, space-time scan statistic
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:https://purl.utwente.nl/essays/92400
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page