The studies show that reading PP intervals using PPG (optical sensor, where PP intervals are Peak-To-Peak) is quite accurate and HRV (sometimes called PRV due to the method) isn't falling far behind the ECG and RR intervals. In fact, in most cases, PP is enough. My doubts are mostly related to Fitbit's methodology. The HRV is being measured for a longer period of time (how long? it's nowhere specified) and comparing the HRV measured within different time domains won't work. The long-term HRV and short-term HRV isn't the same. Now, I don't know whether Fitbit does long-term (which scientifically is 24h and more) or short-term (which is equivalent of measurement up to 5min). However, Fitbit mentioned using RMSSD which is more suitable for short-term measurement (up to 5min long, there is also ultra-short HRV but not gonna be getting there) while long-term usually uses SDNN (which is commonly used for 24h monitoring). The time-domain of sleep (when HRV measurement takes place) however is unknown (minutes? hours?) What would make sense would be measuring HRV and delivering HRV from the last 1-5min before waking up (then RMSSD makes sense). Most people don't need hours of measuring but only those few minutes before or/and after waking up. Also, long-term measurement usually results in higher HRV which stands opposite to what users report. Otherwise, it is hard to interpret the HRV without knowing what is the actual source (and source means a lot here!). Accuracy of the sensor and detecting PP intervals is another story. How does Fitbit deal with signal artefacts (even using ECG signal artefacts appear)? No one knows (and there is no way to check it) what is the accuracy of detecting PP intervals (because we need to know the exact moment of the peak it is a lot harder than detecting HR). The HRV number has no meaning (hence nobody should really be concerned about it) as there is no telling what it means. HRV isn't the same as RHR, lower/higher doesn't mean worse/better (it may but doesn't have to). There is a lot more to it and the lack of known time domain makes it useless as long-term and short-term HRV - despite even using the same math formulas - the physiological meaning may be totally different. Fitbit usually gives users just a number, some score (sleep score, readiness score etc.) but how often does it really mean anything useful if it's taken out of the complex context? HRV is a lot more complicated and I think, Fitbit did more harm than help by introducing such a half-baked feature without explaining anything or revealing the methodology behind it.
