Evaluation of Long-Term Diurnal Variability in RMSSD Using a Smartwatch and Pulse Sensor

This paper measures and evaluates the long-term diurnal variability in RMSSD (Root Mean Square of Successive Differences) from a smartwatch and pulse sensor. RMSSD was measured using two methods: Constant measurement of RMSSD by a wearable device such as Fitbit and instantaneous measurement by a fingertip pulse oximeter. Although Fitbit uses a 24-hour average value as a representative daily value, it has shown cosine similarity to the instantaneous data measured by the fingertip pulse oximeter. From these results, the two time-series data are very similar in terms of cosine similarity. Furthermore, the results would be more reliable if the subjects were increased further.