Andrew Rutenberg

136 posts

Andrew Rutenberg banner
Andrew Rutenberg

Andrew Rutenberg

@RutenGroup

AC Fales Professor of Theoretical Physics @DalhousieU. Soft-matter and biological physics. Complex aging, collagen fibrils, single-file diffusion.

Halifax Nova Scotia Canada Katılım Aralık 2020
31 Takip Edilen89 Takipçiler
Andrew Rutenberg
Andrew Rutenberg@RutenGroup·
Just published "Efficient representations of binarized health deficit data: the frailty index and beyond" in Geroscience: rdcu.be/c4jzt Glen Pridham used PCA on binarized deficit data and showed how the FI (and more) naturally emerges! #aging #geroscience #frailty
English
0
1
5
478
Andrew Rutenberg
Andrew Rutenberg@RutenGroup·
ncl.ac.uk/futures/ageing… Nice to see Newcastle recruiting at a senior level in #aginghealth. “This will require a multidisciplinary approach, drawing upon areas such as: •data science •artificial intelligence and machine learning •software engineering”
English
0
0
1
0
Andrew Rutenberg
Andrew Rutenberg@RutenGroup·
So in #aging with studies of #frailty, you should use good imputation if you want calibrated (comparable) measures of FI between studies, or between populations, or between individuals — essentially if you want to (as precisely as you can) compare FI.
English
0
0
1
0
Andrew Rutenberg
Andrew Rutenberg@RutenGroup·
The idea is that while ignoring missing data still lets the FI (a summary measure of aging health) predict well, the ignored values are (with ignore) replaced with the individual average. This leads to a bias (or inaccuracy).
English
0
0
1
0
Andrew Rutenberg
Andrew Rutenberg@RutenGroup·
Spencer found doi.org/10.1371/journa… that the interactions between health attributes are well-modelled as linear and time-independent. It will be interesting to see how far that generalizes to other types of data and organisms.
English
0
1
1
0
Andrew Rutenberg
Andrew Rutenberg@RutenGroup·
Joint models of #aging predict both health trajectories and mortality — both are needed to compare to large population studies. It is too easy to just get mortality (#Gompertz) right…
English
0
0
1
0
Andrew Rutenberg
Andrew Rutenberg@RutenGroup·
Glad to say our #ML #aging paper has just been published by PLOS Computational Biology doi.org/10.1371/journa… We call it the DJIN model of aging (dynamic joint interpretable network model) Congratulations Spencer Farrell, who also just received a PhD from Dalhousie!
English
0
1
4
0
Andrew Rutenberg
Andrew Rutenberg@RutenGroup·
Nice paper discussing how DAGs (directed acyclic graphs) push variance around, and how this can distort causal discovery benchmarks but may be exploitable IRL if variables have the same natural units. #APaperADay arxiv.org/abs/2102.13647
English
1
3
6
0