Minimum tracking data design

OK. Instead of working, I just set up a little database with what I would think to be the minimum of data to collect in the uberman-tracking-app-to-be.


Just look at the yellow columns – the rest just comes from my favorite data persistance library.

Basically, we would need to have a user (1) who simply has a nick name – nothing else.

A user can start “experiments” (2).

  • He chooses his schedule (“Uberman”, “Everyman2”, “Custom”, …)
  • adds a Comment and starts the experiment.
  • The status of the experiment can simply be “Running” or “Finished/Success” / “Finished/Failed”.

Two kinds of information are tracked during an experiment:

TrackingData (3) (added very frequently): How fit / alert do you feel right now? Here other metrics might be added in the future.

SleepingPeriod (4): For every time the user slept for more then a micro nap (i.e. more than 5 minutes) or tried to sleep due to a scheduled nap time, he enters a “sleeping period”:

  • Start/End are obvious
  • Quality of sleep: “Did not sleep”, “Deep”, “Dreamed”, “…”
  • EffectOfSleep: “Refreshing”, “Exhausting”, “NoEffect”, “…”
  • ReallySleptInMinutes: Even though the user lay down for 20 minutes he estimates he only could sleep for 5 minutes.

This would track both the amount of sleep and the quality of sleep.

In further versions of the app, these periods could be compared to the planned (schedule) sleeping periods in order to automatically identify “oversleep”, “missed nap”, “postponed nap”, “extra nap” and so on.

With this simple data one could already calculate some meaningful metrics.

From the tracking data one could calculate an average level of alertness and through analysing the amplitudes of this value throughout a day one could calculate a current level of “adaption”.

From the sleeping records in combination with the planned schedule, one could calculate a metric of how exact the person sticks to the schedule, perhaps called “accuracy” or such.

Also one could calculate a metric for the amount of sleep a person gets and of how the person judges the quality of this sleep.

This again related to the level of altertness and the level of adaption might give hints of how much adaption is really affected by accuracy…

Just a few thoughts. And now I have to go to work again.

2 thoughts on “Minimum tracking data design

  1. six40winks

    It looks good for a start. I think that’s the essential structure.
    If people have more granularity in their sleep data (with zeo for instance) you may need to have one level below sleep period. (for example inside one sleeping period you may have 10mn of light sleep, 5mn rem and 10mn of deep sleep). Is that at tool to build a mysql database? Will the framework be PHP based?

    1. admin Post author

      I think I will model zeo as a different metric either beneath SleepingPeriod or on the same semantic level. Since I am an old school microsofty, the tool will be ASP / SQL Server based. But essentially the database can be easily switched to MySql and all APIs will be published as SOA/WebServices, so extending this from PHP/Java will be possible.

Leave a Reply