Evaluation of event-based internet biosurveillance for multi-regional detection of seasonal influenza onset - DUMAS - Dépôt Universitaire de Mémoires Après Soutenance Access content directly
Master Thesis Year : 2020

Evaluation of event-based internet biosurveillance for multi-regional detection of seasonal influenza onset

Abstract

Infectious diseases remain a major public health problem worldwide. Hence, event-based surveillance (EBS) was developed to allow more timely detection of infectious disease outbreaks using web-based data, but EBS systems have never been evaluated on a global scale. Thus, the objective of this thesis is to evaluate the ability of EBS to detect epidemic outbreaks of influenza in 24 countries worldwide. Additionally, factors influencing system performance will be identified. Data were obtained from two EBS systems, HealthMap and EIOS. Publicly available weekly virological influenza data were gathered from the FluNet platform as gold standard data. Bayesian change point analysis was used to detect the beginning and end of influenza epidemics. Then, evaluation metrics were calculated, with timely sensitivity, i.e. outbreak detection within the first two weeks after onset, as the main outcome of interest. System performance varied widely between countries and systems, and timely detection of outbreaks was poor in both systems, with HealthMap showing consistently better performance than EIOS. Whereas data abundance influenced the performance of both systems, the human development index (HDI) was influential for HealthMap, and EIOS performance was dependent on a country's geographical location. It could be shown that application of biosurveillance methods to the frequency of online media reports about influenza from two EBS was not able to detect seasonal influenza outbreaks in a timely manner. However, extraction and analysis of additional information from online media and the integration of EBS with other data sources may help to attain this goal.
Fichier principal
Vignette du fichier
M2PHDS_2020_GANSER.pdf (3.46 Mo) Télécharger le fichier

Dates and versions

dumas-03149876 , version 1 (23-02-2021)

Identifiers

  • HAL Id : dumas-03149876 , version 1

Cite

Iris Ganser. Evaluation of event-based internet biosurveillance for multi-regional detection of seasonal influenza onset. Santé publique et épidémiologie. 2020. ⟨dumas-03149876⟩
72 View
137 Download

Share

Gmail Facebook X LinkedIn More