Machine learning assists the classification of reports by citizens on disease-carrying mosquitoes

摘要: Mosquito Alert (www.mosquitoalert.com/en) is an expert-validated citizen science platform for tracking and controlling disease-carrying mosquitoes. Citizens download a free app use their phones to send reports of presumed sightings two world-wide disease vector mosquito species (the Asian Tiger the Yellow Fever mosquito). These are then supervised by team entomologists and, once validated, added database. As prepares scale much larger geographical areas user bases, expert validation becomes main bottleneck. In this paper we describe machine learning on automatically validate fraction them, therefore allowing either deal with report streams or concentrate those that more strategic, such as from new (so early warning protocols activated) high epidemiological risks control actions reduce mosquito populations activated). The current prototype flags third “almost certainly positive” confidence. It currently being integrated into workflow platform.