Influenza surveillance and forecast with smartphone sensors

Abstract

In this paper we introduce an influenza surveillance and forecast system (ISFS) that can track the proliferation of influenza and predict potential infections by analyzing smartphone sensor readings. While previous studies investigate social connectivity to deduce proliferation paths, we focus on the physical contacts of each individual that are the dominant cause of influenza infections. To estimate the probability of an infection through each physical contact we measure the surrounding features of each contact including the staying time of a contact, the human density and the openness of the space, and the infection status of each individual. By using a smartphone equipped with various sensors we can estimate the infection status of its owner by analyzing both the envelope of incoming sound and the surrounding features of the contact. A surveillance server, which aggregates the information from multiple smartphones, monitors the infection status of influenza and ranks both high risk persons and influential persons that have to be vaccinated promptly. To evaluate the forecast accuracy of ISFS we have implemented a full ISFS including an Android ISFS client and compare the forecast accuracy of ISFS against that of the traditional forecast system based on social connectivity. Our evaluation results suggest that influenza surveillance and forecast should be performed based on human activity rather than social connectivity. This would not only improve the forecast accuracy but it can also improve the cost efficiency and the suppression effect of vaccinations by finding the most influential persons in the proliferation paths.