[Work Log] FIRE - notes: joint meeting
Q: how to align data?
David - align by first treatment
Emily - end point is also interesting - release point
Emily - Three most interesting time-syncs: line up at first, line up at last, line up since diagnosis
Rebecca - Surgery date? (diagnosis date could be a proxy for this)
Question for karen: if surgery, when did it happen?
David: what is being studied? adjustment to diagnonis, or adjustment to treatment?
Rebecca plots
- refacttot over time by treatment group
Ambivalence
This idea of ambivalence seems to be an interesting avenue of study. The main idea is having high "helpfulness" and high "upsetting" indicates an "ambivalent" relationship; there seems to be much variability between how these two variables covary, may cluster well.
- Dataset: SRI
- we see evidence of this in our data
- how do these dimensions track over time? (this would be novel)
Different dynamics of interest:
- healthy: float low vs flat high
- healthy: one doesn’t affect the other much.
- uncertain: okay but wobbly, and connected
- unhealthy: high "upsetting" but low "helpfulness"
Experiment: can ambivalence (sri) model predict "health"?
- health could be measured by physical functioning (fact), or stress (pss)
Next steps
Investigae ambivalence as a two-dimensional model. 2D dynamical states-space model? Cluster on a few different state space models?
…but continue with exploratory. Remember to split up SRI dataset into two columns.
Miscellanous ——————— Might be interesting to fit a 5-dim first order auto-regressive model on the 5 datasets of interest, with iid observations.
Posted by Kyle Simek