Unsupervised Extraction of Epidemic Syndromes from Participatory Influenza Surveillance Self-reported Symptoms

摘要: Seasonal influenza surveillance is usually carried out by sentinel general practitioners who compile weekly reports based on the number of influenza-like illness (ILI) clinical cases observed among visited patients. This practice for generally affected two main issues: i) are released with a lag about one week or more, ii) definition case patients symptoms varies from system to other, i.e. country other. The availability novel data streams disease can alleviate these issues; in this paper, we employed Influenzanet, participatory web-based project which collects directly population real time. We developed an unsupervised probabilistic framework that combines time series analysis counts and performs algorithmic detection groups symptoms, hereafter called, syndrome. Symptoms were collected through platforms consortium called Influenzanet found correlate Influenza-like incidence as detected doctors. Our aim suggest how provide epidemiological signal capable detecting illness9 temporal trends without relying specific definition. We evaluated performance our showing syndromes closely follow ILI reported traditional surveillance, consist combinations compatible definition. The proposed was able predict quite accurately trend forthcoming season only available information previous years. Moreover, assessed generalisability approach evaluating its potentials gastrointestinal syndromes. We against and despite limited amount data, successfully detected. The result real-time flexible prediction tool not constrained any