Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Machine learning research in med school
#1
Hello, 

I have the opportunity to work on 3 possible research interests for Fall medical school (MS1). Which one is best for neurosurgery residency? Thank you so much!

  1. Machine learning-- My first choice. Here, I will work with a computer science PI (PhD) who develops deep learning models (software) that reads PET/CT images for lung and esophageal cancers. Basically, this software is working to read the images to identify cancer and disease. He hopes to expand to brain tumors but of course that depends on funding etc. I would help with programming and imaging (so I might read the images myself and then teach the software. PI publishes in low-impact journals. 
  2. Bench lab research. PhD, cell signaling. HOWEVER, a GREAT PI. Getting some good funding, and EXCELLENT journals. Think Cell.
  3. Bioinformatics--pretty traditional. Work with statistics PhD to use databases to do stastical analysis on patient outcomes, as well as genomic (using NCBI and other databases to compare protein/DNA sequences and the like). PI publishes in low-impact journals, but I think can churn out posters/crappy papers. 
What would you guys recommend? I really want to go for machine learning research, but I think that most PDs are not family at all with it? How would you guys look at it? 

Thank you so much!
Reply
#2
There's a lot of machine learning in neurosurgery right now, so familiarity with it is not the issue. Main thing is that the machine learning project isn't specific to neurosurgery, but it sounds like most of these aren't. A cell paper would be impressive but how likely are you to actually get such a paper by the time you apply? That route could be high risk but high reward, but it sounds like the other two options may get higher numbers of research items. In the end it sounds like the machine learning is most what you're interested in and more likely to publish than the bench lab so that could be a good choice. If you think option 2 is likely to yield a high impact paper with your name on it and it's more related to neurosurgery somehow, that's also a strong option
Reply
#3
(05-02-2021, 10:32 PM)Guest Wrote: There's a lot of machine learning in neurosurgery right now, so familiarity with it is not the issue. Main thing is that the machine learning project isn't specific to neurosurgery, but it sounds like most of these aren't. A cell paper would be impressive but how likely are you to actually get such a paper by the time you apply? That route could be high risk but high reward, but it sounds like the other two options may get higher numbers of research items. In the end it sounds like the machine learning is most what you're interested in and more likely to publish than the bench lab so that could be a good choice. If you think option 2 is likely to yield a high impact paper with your name on it and it's more related to neurosurgery somehow, that's also a strong option

CELL OR BUST
Reply
#4
(05-02-2021, 10:32 PM)Guest Wrote: There's a lot of machine learning in neurosurgery right now, so familiarity with it is not the issue. Main thing is that the machine learning project isn't specific to neurosurgery, but it sounds like most of these aren't. A cell paper would be impressive but how likely are you to actually get such a paper by the time you apply? That route could be high risk but high reward, but it sounds like the other two options may get higher numbers of research items. In the end it sounds like the machine learning is most what you're interested in and more likely to publish than the bench lab so that could be a good choice. If you think option 2 is likely to yield a high impact paper with your name on it and it's more related to neurosurgery somehow, that's also a strong option


Thank you so much for this advice. Option 2 is only tangentially related to neurosurgery (perhaps a project on cell signaling for oncogenic cells). Mind you, he's an excellent PI with great papers. I have worked on bench research and have gotten several publications. So I enjoy it. 

It was also great to hear about machine learning in neurosurgery. I can only think of diagnostic applications thus far; am I missing anything? 

I am going to be MS1. I'll have 2 solid years to work on this, so I hope to apply his software to brain pathologies. The great thing about machine learning, I'd say, is the convenience of doing programming; you can literally do it anywhere, on your own time.

The bio-stats one would probably lead to the most publications, but it is just so boring to compare studies after studies. It would be a pain and quite frankly I can't imagine how anyone can do that.
Reply
#5
(05-02-2021, 11:55 PM)In Love with a Loli Wrote:
(05-02-2021, 10:32 PM)Guest Wrote: There's a lot of machine learning in neurosurgery right now, so familiarity with it is not the issue. Main thing is that the machine learning project isn't specific to neurosurgery, but it sounds like most of these aren't. A cell paper would be impressive but how likely are you to actually get such a paper by the time you apply? That route could be high risk but high reward, but it sounds like the other two options may get higher numbers of research items. In the end it sounds like the machine learning is most what you're interested in and more likely to publish than the bench lab so that could be a good choice. If you think option 2 is likely to yield a high impact paper with your name on it and it's more related to neurosurgery somehow, that's also a strong option


Thank you so much for this advice. Option 2 is only tangentially related to neurosurgery (perhaps a project on cell signaling for oncogenic cells). Mind you, he's an excellent PI with great papers. I have worked on bench research and have gotten several publications. So I enjoy it. 

It was also great to hear about machine learning in neurosurgery. I can only think of diagnostic applications thus far; am I missing anything? 

I am going to be MS1. I'll have 2 solid years to work on this, so I hope to apply his software to brain pathologies. The great thing about machine learning, I'd say, is the convenience of doing programming; you can literally do it anywhere, on your own time.

The bio-stats one would probably lead to the most publications, but it is just so boring to compare studies after studies. It would be a pain and quite frankly I can't imagine how anyone can do that.

Mostly diagnostics, but that is significant in a field like this where patient selection for surgery is important. Think algorithms for determining based on imaging and clinical data whether disease is progressing and requires intervention etc.
Reply
#6
(05-03-2021, 01:51 PM)Guest Wrote: Mostly diagnostics, but that is significant in a field like this where patient selection for surgery is important. Think algorithms for determining based on imaging and clinical data whether disease is progressing and requires intervention etc.

Thank you for that information, it is what I thought. Of course, in the future we will no doubt see machine learning involved in things like understanding protein and cell structure, anesthesiology and robot-assisted medicine. 

It is absolutely AMAZING how fast the field is progressing. Just 5 years ago it was barely beginning! I believe machine learning will play a pivotal role in medicine. It is certainly exciting to be a (very) small part of that!
Reply
#7
(05-03-2021, 07:00 PM)In Love with a Loli Wrote:
(05-03-2021, 01:51 PM)Guest Wrote: Mostly diagnostics, but that is significant in a field like this where patient selection for surgery is important. Think algorithms for determining based on imaging and clinical data whether disease is progressing and requires intervention etc.

Thank you for that information, it is what I thought. Of course, in the future we will no doubt see machine learning involved in things like understanding protein and cell structure, anesthesiology and robot-assisted medicine. 

It is absolutely AMAZING how fast the field is progressing. Just 5 years ago it was barely beginning! I believe machine learning will play a pivotal role in medicine. It is certainly exciting to be a (very) small part of that!

Agreed. I've started into some of that work myself and see a lot of potential.

If this is really your interest, worth pursuing it and learning the ins and outs of the process. I kind of wish I had a computer science background before starting research, but hindsight is always better.
Reply
#8
(05-03-2021, 07:06 PM)Guest Wrote: Agreed. I've started into some of that work myself and see a lot of potential.

If this is really your interest, worth pursuing it and learning the ins and outs of the process. I kind of wish I had a computer science background before starting research, but hindsight is always better.

Oh wow that is great to know! I don't have a computer sci background, but like you I wish I did. It would have made things so much easier. 

I am teaching myself machine learning now actually, I am reading a few books on it. It is not hard to get the basic gist of and I'm beginning to write a few algorithms. Of course, I'll never be a pro, but I hope to use my medical knowledge to contribute to the team. I guess that will be the area I work in, because most computer science guys don't know much about medicine and are actively looking for knowledgeable collaborators. 

So even if you have only a basic idea of programming and such, I am sure you'll be useful to the team. I'm also learning some of the match behind it. 

If I may ask, are you in med school? 

Computer sci research is just great, imo. It is intellectually challenging--unlike those stupid meta-anlysis papers that are churned out daily. Best part, you can do it anytime and their are tons of opportunities if you position yourself right. 

At least this is what I gather, it is certainly a very exciting time.
Reply
#9
The likelihood that you will be the first author on an ML project with only a couple of summers to work and no background as an MS1 is slim-to-none. These projects are big, high-risk undertakings that will involve a decent-sized team of people. The senior grad student or postdoc heading the project will undoubtedly get the first authorship and you'll be buried in the middle.

I would personally go for option number #2 and try to head up a project of your own that you will be able to talk about at length. It really does not take long during an interview to figure out if someone was a hanger-on for a project or if they were integrally involved.

Take this advice or leave it. I'm a co-PI on a project with an NIH grant and a patent involving ML for a neurosurgical application, but I'm also an anonymous internet troll so who knows?
Reply
#10
(05-04-2021, 07:08 PM)drstrange Wrote: The likelihood that you will be the first author on an ML project with only a couple of summers to work and no background as an MS1 is slim-to-none. These projects are big, high-risk undertakings that will involve a decent-sized team of people. The senior grad student or postdoc heading the project will undoubtedly get the first authorship and you'll be buried in the middle.

I would personally go for option number #2 and try to head up a project of your own that you will be able to talk about at length. It really does not take long during an interview to figure out if someone was a hanger-on for a project or if they were integrally involved.

Take this advice or leave it. I'm a co-PI on a project with an NIH grant and a patent involving ML for a neurosurgical application, but I'm also an anonymous internet troll so who knows?


Thank you for this reply. I have first-authored a few basic science publications so I kind of know the hang of it. Even a negative result can be spun to a presentation/poster. This is what I was thinking of doing. 

I did not realize that a first-author publication would make such a big difference. 

I will probably speak to both of them, and inquire about a small side project. I can put in 10 hours/week while in med school, and of course, 50+/week on holidays/summer.  

Thank you again. 

Also, if I may ask, but the aggressive tone in the last sentence? I noticed that such language is common here, is it common amongst neurosurgeons?

Thank you again.
Reply


[-]
Quick Reply
Message
Type your reply to this message here.

Image Verification
Please enter the text contained within the image into the text box below it. This process is used to prevent automated spam bots.
Image Verification
(case insensitive)

Forum Jump:


Users browsing this thread: 1 Guest(s)