Using decision fusion methods to improve outbreak detection in disease surveillance

Gaëtan Texier,Jean-Baptiste Meynard,Loty Diop,Rodrigue S Allodji,Hervé Chaudet,Liliane Pellegrin

doi:10.1186/s12911-019-0774-3

Abstract

BackgroundWhen outbreak detection algorithms (ODAs) are considered individually, the task of outbreak detection can be seen as a classification problem and the ODA as a sensor providing a binary decision (outbreak yes or no) for each day of surveillance. When they are considered jointly (in cases where several ODAs analyze the same surveillance signal), the outbreak detection problem should be treated as a decision fusion (DF) problem of multiple sensors.MethodsThis study evaluated the benefit for a decisions support system of using DF methods (fusing multiple ODA decisions) compared to using a single method of outbreak detection. For each day, we merged the decisions of six ODAs using 5 DF methods (two voting methods, logistic regression, CART and Bayesian network - BN). Classical metrics of accuracy, prediction and timelines were used during the evaluation steps.ResultsIn our results, we observed the greatest gain (77%) in positive predictive value compared to the best ODA if we used DF methods with a learning step (BN, logistic regression, and CART).ConclusionsTo identify disease outbreaks in systems using several ODAs to analyze surveillance data, we recommend using a DF method based on a Bayesian network. This method is at least equivalent to the best of the algorithms considered, regardless of the situation faced by the system. For those less familiar with this kind of technique, we propose that logistic regression be used when a training dataset is available.

Highlights

When outbreak detection algorithms (ODAs) are considered individually, the task of outbreak detection can be seen as a classification problem and the Outbreak detection algorithm (ODA) as a sensor providing a binary decision for each day of surveillance
With the aim of improving decision making for disease surveillance system users, we propose to evaluate the benefit of using decision fusion (DF) methods fusing multiple ODA decisions versus using a single method of outbreak detection
The implementation of Decision fusion methods (DFM) showed that voting methods provided detection sensitivities per outbreak [78 to 82%], close to those of Cumulative Sum (CUSUM), C3 or Farrington while other DFMs such as logistic regression, Classification and regression trees (CART), or Bayesian networks (BNs), had on average a detection sensitivity per outbreak lower than the range indicated above

Summary

Introduction

When outbreak detection algorithms (ODAs) are considered individually, the task of outbreak detection can be seen as a classification problem and the ODA as a sensor providing a binary decision (outbreak yes or no) for each day of surveillance. ODA performance depends on several characteristics associated with the outbreak curve (shape, duration and size), the baseline (mean, variance) [4, 5] and their relationships (signalto-noise ratio, signal-to-noise difference) [6, 7] In this context, the hope of having a single algorithm that would be efficient enough to detect all outbreaks in all situations faced by a disease surveillance system is probably illusory. All these problems call into question the true benefit of using multiple ODAs for decision-making

Methods

Results

Discussion

Conclusion