rethink "phase": preceding duration

[cirorandazzo]

Goal

Rework phase to meaningfully categorize a call's relationship to ongoing respiration.

Problem Statement

Phase seems a natural way of thinking about the respiratory trace, but poses some difficult problems. In particular, deciding the length of a "normal breath" yardstick can be tricky. Using the average breath length neglects changes in baseline breath rate and generally merges distinct phenomena (eg, slightly longer than average breaths are binned into the same phases as very early phase deviations). Conversely, using some exemplar "good, recent" breath is susceptible to noise, ruling out many trials. To compound this, phase can't tell us anything about, for example, abbreviating the pre-call expiration.

So, the question is: how do we meaningfully categorize a call's relationship to ongoing respiration while addressing some of the failures of phase?

Proposed solution

• As a start, it makes sense to consider the duration of several preceding breath segments separately.
• Apply this also to separating cluster/call traces: group traces in 2 dimensions, according to previous exp duration and previous insp duration.
• Can also be used to ask: is local respiratory rate predictive of call duration? (Should be, if call is riding on the normal expiratory motor pattern.)