Hi there. I'm still a bit of a novice, and I'm wondering if anyone could advice me if this is empirically "responsible"?

So essentially I've generated 100+ summaries of individual patient journeys through univariate linear regression. So each individual model was calculated using time to predict the outcome variable.

So initially I wanted to test the rate of change between two patient subgroups, and I did this by inputting the grouping factor and the estimate coefficients (from each individual model) into an ANOVA. Presto - no significance. But that's not my issue here. Now I want to use this same method to look at within-subject differences across each patient subgroup (using these same summary measures).

I'm wondering if it's okay to, say, run a paired samples t-test between the intercept and the slope of a particular subgroup of patients? (Probably a dumb question).

I think I heard someone mention a method like this a while ago, but I feel that I may have taken them up wrong. I'm sort of making this thread out of doubt, really.

Thanks in advance,

-Michael