Alpha1 & Respiration Rate

@Luisma_Gallego_Soy_P is doing some cool work with the Garmin IQ Apha1HRV app combining Alpha1 and Respiration Rate. My question is: is it possible to create a custom chart on the activity page which has one metric divided by another i.e. RespRate/Alpha1? @david


Not yet. But I am am working on making extensible like that.

  • Explain the split metric search? Thanks

As per your Twitter posts I download the data from Intervals and put it into Google sheets to create the graph of RR/A1 against time on the ramp test. I was asking if that could be configured in Intervals to save time, as messing around graphing is quite time consuming.

From your and Inigo’s Twitter posts, it looks like you are using breathing frequency (in Hz - so breaths per sec) divided by a1 to get a stronger indicator for the deflection point.
A typical stable state under AeT would return values in the order of 35 breaths per min and a1 around 1.5 resulting in a calculated value around 0.4.
For intensities well above AeT, 60 breaths/min and a1 at 0.5 resulting in 2.
That explains the vertical scaling in the screenshots. And since you have already seen multiple cases where the a1 and resprate deflection points occur at the same time, the combination of both will make it easier to pinpoint the AeT.
Love the work you’re putting in all this :+1:

Exactly, both a1 and RespRate variations seem good indicators for AeT detection. But, each parameter alone can be affected by “noise” that makes difficult to use it in a simple way for that purpose.
Combining both as FR/A1 provides a strong indicator where responsiveness of each one is potentiated. Even, in case that one fails, the other one can provide enough information.
Moreover, AeT detection is being based on slope variations instead of absolute thresholds. That makes the procedure stronger and less dependent on specific values that could vary for different athletes. Slope variation seems to be a general “behaviour”.
Right now, location of the slope variation point is detected by means of a bilinear fitting process, where both lines are automatically optimized, without any necessity of manually defining range convered by each line.
Using Heart rate instead of time for X axis has also shown better results as it makes more clear any trend.
Results are promising.