A Unifying Framework and Comparative Evaluation of Statistical and Machine Learning Approaches to Non-Specific Syndromic Surveillance

Moritz Kulessa,Eneldo Loza Mencía,Johannes Fürnkranz

doi:10.3390/computers10030032

Abstract

Monitoring the development of infectious diseases is of great importance for the prevention of major outbreaks. Syndromic surveillance aims at developing algorithms which can detect outbreaks as early as possible by monitoring data sources which allow to capture the occurrences of a certain disease. Recent research mainly concentrates on the surveillance of specific, known diseases, putting the focus on the definition of the disease pattern under surveillance. Until now, only little effort has been devoted to what we call non-specific syndromic surveillance, i.e., the use of all available data for detecting any kind of infectious disease outbreaks. In this work, we give an overview of non-specific syndromic surveillance from the perspective of machine learning and propose a unified framework based on global and local modeling techniques. We also present a set of statistical modeling techniques which have not been used in a local modeling context before and can serve as benchmarks for the more elaborate machine learning approaches. In an experimental comparison of different approaches to non-specific syndromic surveillance we found that these simple statistical techniques already achieve competitive results and sometimes even outperform more elaborate approaches. In particular, applying common syndromic surveillance methods in a non-specific setting seems to be promising.

Highlights

Rather than developing highly specialized algorithms which are based on specific indicators and assume particular characteristics of outbreak shapes [4], we argue that the task of outbreak detection should be viewed as a general anomaly detection problem, where an outbreak alarm is triggered if the distribution of the incoming data changes in an unforeseen and unexpected way
(4) We propose a set of benchmarks for non-specific syndromic surveillance relying on simple distributions which have been widely used in syndromic surveillance
We gave an overview about non-specific syndromic surveillance from the perspective of machine learning

Summary

Introduction

The early detection of infectious disease outbreaks enables to apply control measures at an early stage, which can save lives and reduce suffering [1] In this regard, syndromic surveillance has been introduced which aims to identify clusters of infected people before final diagnosis are confirmed and reported to public health agencies [2]. The fundamental concept of syndromic surveillance is to define indicators for a particular infectious disease on the given data, referred to as syndromes, which are monitored over time to be able to detect unexpectedly high numbers of infections which might indicate an outbreak of that disease.

Methods

Results

Conclusion