The Author’s interest is to apply physiological recordings for real-life, long-term and unsupervised monitoring as a complement to telemedicine and home care.
This means the search for parameters that will be used to alert when some kind of support will be needed.
To this end, the Author has recorded and published an appropriate Motionwatch8 dataset and this article continues to explore some of its basic characteristics in preparation for its analysis.
The Author claims that Motionwatch8 data, both acceleration and light, are recorded as a succession of samples, but the zero threshold levels of the device imply a segmentation of the data from 2 events: crossing the zero thresholds upwards and downwards.
The paper describes those segments and uses a simple numerical example to show some some of their features, such as their circadian rhythms.
Finally, we propose a data model that allows the evaluation of Motionwatch8’s light data, which are often co-recorded along with acceleration.
The examples suggest that, at least for some parameters, using the segment lengths or the sum and average of their Count/Lux values provides the same information.
The pairs of segments and intervals used in the examples are easy to calculate, their relationship to device measurements is intuitive and they provide a useful guideline, but the sequence of segments is open to a variety of analysis methodologies to be explored.
The proposed type of data model may allow some use of the light data measured by the Motionwatch8.