89% TIR, well done! Cool to see the research, I’m surprised there aren’t more in-depth n=1 studies out there from statistics/data-savvy diabetics. I built a small dashboard exactly collating carb, insulin, sleep, and movement data and trying to coax out useful correlations to help manage. Some interesting insights, but clearly I should eat less and move more. Keep up the good work, and thanks for sharing!
0xdeadf1sh•3d ago
Yeah I'm thinking of adding more (especially recent) models and compare their published metrics with those of mine to see how they hold up. Then I'll likely post about it here again. See you then!
Bigchrome•3d ago
I achieved 100% TIR with my model. I'm not T1D, but same same.
JK, as the spouse of someone with T1D I really appreciate seeing this!
zenpositive•3d ago
Very cool very nice work!
orasul•3d ago
awesome
srdjanb•2d ago
This is cool. The accuracy looks good too, except for a few time windows. Do these models take into account dietary info, and if yes, how?
0xdeadf1sh•2d ago
Carbohydrates & insulin are both accounted for during inference. I'm not sure if that's what you meant by dietary info though.
nuthje•3d ago
0xdeadf1sh•3d ago