Smartwatches that learn your every move
Current smartwatches can recognize a limited number of activities that were programmed in advance. There is a new algorithm that can detect new activities as they happen in real time. It is no longer limited to only a few, predefined set of activities.
The current way recognizing activities is to bundle together bursts of activity to estimate what the person was doing and for how long. An example is when a someone takes a bunch of steps, it is clustered into a walk. The new algorithm tracts ongoing activity, especially to transitioning as well as the activity itself. Up comming watches will be better able to track our activities and learn new activities as we engage in them. This will provide a better picture of our daily lives and can be used in healthcare and behavior research.