University of Twente Student Theses

This website will be unavailable due to maintenance December 1st between 8:00 and 12:00 CET.

Integrating a hierarchical SOM-based intervention method in an agent-based pertussis model

Zhou, Quan (2018) Integrating a hierarchical SOM-based intervention method in an agent-based pertussis model.

[img] PDF
Abstract:A resurgence of pertussis has been detected in the Netherlands since 1996.Several intervention methods have been implementedbut a severe outbreak was detected in 2012. Some authors declared that more efficient alternative interventions should be studied. The main problems of planning a disease intervention areto determine the correct moment in time to start the intervention, the spatial locations in which to conduct the intervention and the target population that should get the intervention. However, previous research always focused either on disease aberration of different age groups for the whole country or the disease aberration in specific geographic regions for all age groups. The main goal ofthis research is to construct an intervention model which can draw up intervention planning automatically for different spatial regions based on local disease occurrence to prevent pertussis. Because of time limitation, only vaccination strategies are taken into accountas the intervention method in this research. The generated vaccination methods are referred as condition-based vaccination method. The intervention model is agent-based with two entities: GGDs and RIVM. GGDs are local health units which supervise the ongoing diseasepattern in their service region and implement interventions. RIVM is the Netherlands national public health and environment institution which collects the disease information from all GGDs and detects disease aberrant change at the national level and orders GGDs to implement intervention in their service area. The behaviors of these two agents were simulated using two sub models: early warning model and intervention selection model. Early-warning model is a hierarchical model which detects disease aberration at both national and regional level. The national aberration determines the time when the interventions should be implemented while the regional detection explores the aberration age groups. A hierarchical self-organizing map (SOM) is used as the aberration detection algorithm which has not been proposed before. The approach is tested on a simulated dataset for Pertussis in the Netherlands and the result can be seen as effectiveness. The aberration age groups will be analyzed in the intervention selection model to determine the target population who should get vaccination priority. Meanwhile, the intervention selection model stipulates the vaccination ratiofor each target population groups in different region. The environment of the intervention model is a disease transmission model which simulates pertussis spread in the Netherlands established by previous research. Several validation experiments were performed at first and the result presents that the model run as expected. The impact of the condition-based vaccination method is assessed regardingthe number of infections and the disease diffusion patterns and compared with the current vaccination strategy and maternal vaccination approach. The total number of infections of the condition-based vaccination scenario is obviously less than the number of other two vaccination scenarios and the disease outbreak is restricted in a small area with the condition-based vaccinations.
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:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page