Using Smartphones for Research Outside Clinical Settings: How Operating Systems, App Developers, and Users Determine Geolocation Data Quality in mHealth Studies.

Smartphones that collect user geolocation provid opportunities for mobile Health (mHealth). Although granularity of geolocation data may be high, data completeness depends on the device's operating system, application developer decisions, and user actions. We investigate completeness of geolocation data collected via smartphones of 5601 people that self-reported daily chronic pain symptoms on 349,293 days. On 17% of these days, hourly geolocation data is reported, but days with 0 (16%), 1 (14%) and 2 (13%) geolocations are common. Android phones collect geolocation more often than iPhones (median 17 versus 2 times a day). Factors on operating system level and individual user level influence completeness of geolocation data collected with smartphones. mHealth researchers should be aware of these factors when designing their studies. The mHealth research community should devise standards for reporting geolocation data quality, analysing systematic differences in data quality between participant groups, and methods for data imputation.