Early-warning signals derived from actigraphy and electrocardiogram time series data

Abstract

In my research we have investigated whether early warning signals could help in foreseeing transitions in (depressive) mood symptoms. These early warning signals have been applied successfully in other scientific fields such as biology, ecology, and economy. We expected to find these early warning signals in physiological data, specifically in movement data and heart rate data. To investigate this we asked participants to monitor their own movement and heart rate data at home for long periods of time. To ensure a correct data collection, we have tested and validated various movement and heart rate monitors. With the data we gathered we were able to investigate whether early warning signals can indeed help in foreseeing transitions in (depressive) mood symptoms. We found that in movement data from some participants there were indeed early warning signals visible, up till four weeks before a transition in (depressive) mood symptoms. In the heart rate data we found early warning signals in some participants, but we were unable to find these in other participants. In summary we found some evidence that early warning signals can help in foresee transitions in (depressive) mood symptoms. However, more research is needed to ascertain this. Future studies could also investigate using early warning signals in a more efficient manner, by for example, combining single early warning signals in aggregate predictors so they will perform better.