Joel Eliason
@jeliason.bsky.social
Husband and dad of two. PhD student in computational biology, studying the spatial ecology of the tumor-immune microenvironment. Stats/math lover. Slava Ukraini!
Reposted by Joel Eliason
Sorry again! That also seems harsh. What I really mean is that it's great to see these would be masters of the universe finally accept that they should apply the same level of statistical thinking that I might expect to see in a 4th year medical student's research project
November 22, 2024 at 11:10 AM
Sorry again! That also seems harsh. What I really mean is that it's great to see these would be masters of the universe finally accept that they should apply the same level of statistical thinking that I might expect to see in a 4th year medical student's research project
That's not to say taxonomy doesn't have a place, just that we shouldn't stop there.
November 21, 2024 at 12:48 AM
That's not to say taxonomy doesn't have a place, just that we shouldn't stop there.
Here's a clip from an interview with Paul Macklin who shared the same sentiment:
Paul Macklin
YouTube video by Math Oncology Interviews
youtu.be
November 21, 2024 at 12:45 AM
Here's a clip from an interview with Paul Macklin who shared the same sentiment:
I just read Uri Alon's take on this yesterday - he recommends at least 3 months! I have also jumped into some projects way too quickly.
How To Choose a Good Scientific Problem
Choosing good problems is essential for being a good scientist. But what is a good
problem, and how do you choose one? The subject is not usually discussed explicitly
within our profession. Scientists...
www.cell.com
November 17, 2024 at 5:43 PM
I just read Uri Alon's take on this yesterday - he recommends at least 3 months! I have also jumped into some projects way too quickly.
Since they depend on generating a lot of simulations, they can be pretty computationally expensive. There's also work on creating faster surrogates of the mechanistic model to generate training data faster.
November 17, 2024 at 2:30 AM
Since they depend on generating a lot of simulations, they can be pretty computationally expensive. There's also work on creating faster surrogates of the mechanistic model to generate training data faster.
A relatively naive form of it is approximate Bayesian computation.
November 17, 2024 at 2:28 AM
A relatively naive form of it is approximate Bayesian computation.
This is bottom-up mechanistic modeling, whereas most ODE mathematical models of tissues (like the TME) are top-down, so the translation to population dynamics is not exact.
November 16, 2024 at 3:14 PM
This is bottom-up mechanistic modeling, whereas most ODE mathematical models of tissues (like the TME) are top-down, so the translation to population dynamics is not exact.