Assessing and enhancing the utility of low-cost activity and location sensors for exposure studies

作者：Asimina Stamatelopoulou; Chapizanis D; Karakitsios S; Kontoroupis P; Asimakopoulos D N; Maggos T; Sarigiannis D

来源：Environmental Monitoring and Assessment, 2018, 190(3): 155.

DOI：10.1007/s10661-018-6537-2

• Exposure
• Activity sensors
• Location
• Artificial Neural Network (ANN) model

Nowadays, the advancement of mobile technology in conjunction with the introduction of the concept of exposome has provided new dynamics to the exposure studies. Since the addressing of health outcomes related to environmental stressors is crucial, the improvement of exposure assessment methodology is of paramount importance. Towards this aim, a pilot study was carried out in the two major cities of Greece (Athens, Thessaloniki), investigating the applicability of commercially available fitness monitors and the Moves App for tracking people's location and activities, as well as for predicting the type of the encountered location, using advanced modeling techniques. Within the frame of the study, 21 individuals were using the Fitbit Flex activity tracker, a temperature logger, and the application Moves App on their smartphones. For the validation of the above equipment, participants were also carrying an Actigraph (activity sensor) and a GPS device. The data collected from Fitbit Flex, the temperature logger, and the GPS (speed) were used as input parameters in an Artificial Neural Network (ANN) model for predicting the type of location. Analysis of the data showed that the Moves App tends to underestimate the daily steps counts in comparison with Fitbit Flex and Actigraph, respectively, while Moves App predicted the movement trajectory of an individual with reasonable accuracy, compared to a dedicated GPS. Finally, the encountered location was successfully predicted by the ANN in most of the cases.

• 出版日期2018-3