One question we receive fairly often from Health Graph (@healthgraphapi) partners is how to validate that fitness activities (runs, walks, bike rides, etc.) read out of the Health Graph platform were GPS-tracked versus manually entered by the user. Rewards partners a la Earndit and GymPact, corporate wellness providers like Virgin HealthMiles, and forward-thinking brands are often keen to differentiate between tracked versus manually entered activities as part of their programs’ anti-fraud efforts.
So how do you tell the difference between GPS and manual activities?
Each item in the Fitness Activity feed has ‘source‘, ‘entry_mode‘, and ‘has_path‘ fields. These let you determine whether the activity was originally submitted as a GPS-tracked activity. For example, a RunKeeper (@runkeeper) mobile app GPS-tracked run should have values of “RunKeeper“, “API“, and “true” for the aforementioned fields, respectively.
If you are interested in including GPS-tracked sources from other Health Graph partners’ activity trackers, you can include them in your ‘source‘ filtering. In addition, if you need to differentiate by type of activity (i.e. running, walking, cycling, etc.) you can use the ‘type‘ field.
Using these fields should let you skip any activities for which the user simply entered statistics, or originally entered the route map (path) via the Web. For more details on these fields and their usage, please refer to the Health Graph fitness activities documentation, especially the array structures section.
Caveat: The only reliable way to verify whether a user has subsequently edited the map associated with a saved GPS-tracked activity is to manually check each point’s ‘type‘ (a value of “manual” means it has been edited). For efficiency’s sake, we don’t save that information anywhere else in the Health Graph platform and we retrieve points only when full data for the activity is requested. That said, we have found that most users do not edit maps after the fact.
Bill Day (@billday) is Platform Evangelist & PM for RunKeeper where he helps developers learn about and use the Health Graph.
