Biomedical Engineering PhD, data science, and Employment

[joshthekid]

I have a B.S. in physics, I am currently in a non-thesis Master of Engineering in Mechanical/Biomedical engineering and set to graduate in the spring. I have been working in a lab to pick up some extra cash and skills over the past 7 months. Mainly what I have been doing is applying data science and Machine learning techniques to predict dose response to chemotherapeutics based on genetic analysis. My P.I. has just asked me to stay on and switch to the PH.d where I would mainly be working on Machine learning techniques for personalized medical applications and drug discovery. I have always wanted to get a PH.d but did not get into any of the programs I wanted when a applied a year ago and decided that I would get a Master's and move on. Now I have an offer for full financial support to get a PH.d in Biomedical Engineering. Honestly, I enjoy research and my skill set is more suitable to research than more traditional engineering roles so a PH.d makes sense from that perspective. This is my concern, tenured faculty positions are on decline, not that I necessarily that I want to go or do not want to go into academia, but it simply means that what you get your PH.d in better be marketable and valuable to non-academic positions. Given that I will be focusing on data science and its application to medicine I was wondering if anybody new how in demand that is for PH.d level positions.

 

[Ben Espen]

That is a pretty interesting specialty. To be perfectly honest, I'm not sure I can give you any kind of forecast on demand for something so specialized. I think with a little legwork, you might be able to answer your own question, by investigating what people do when they leave your program, assuming it has been in existence for a while. Find some recent graduates or soon to be recent graduates and buy them a beer in exchange for information.

Here is the big secret of finding a job with a PhD: you don't need to stay in your specialty. Lots of people with PhDs think this. Lots of people looking to hire PhDs think this too, but they are wrong. Drug discovery and personalized medicine is something useful, but if all else fails, that kind of analysis technique should be applicable in lots of places.

 

[Locrian]

This is my concern, tenured faculty positions are on decline, not that I necessarily that I want to go or do not want to go into academia, but it simply means that what you get your PH.d in better be marketable and valuable to non-academic positions. Given that I will be focusing on data science and its application to medicine I was wondering if anybody new how in demand that is for PH.d level positions.

A PhD with advanced statistical analysis will be valuable in a wide variety of industries, including many far from bio engineering.

Mainly what I have been doing is applying data science and Machine learning techniques to predict dose response to chemotherapeutics based on genetic analysis.

So you're doing a regression where patient response is the response variable and there are a number of variables, including dose amount and several factors based on genetic analysis? Maybe a logistic regression or other generalized linear model? Do you understand how the model works? What software package do you use to do the testing?

This is partly curiosity and partly to help answer your question.

 

[joshthekid]

So you're doing a regression where patient response is the response variable and there are a number of variables, including dose amount and several factors based on genetic analysis? Maybe a logistic regression or other generalized linear model? Do you understand how the model works? What software package do you use to do the testing?

We are actually predicting Inhibition Concentrations, (i.e how much drug does it take to inhibit growth). So far we have used Principle Component Regression, Partial Least Squares Regression, ridge regression, and Neural Networks. If I continue I would want to explore Support Vector Machines and develop more concrete methods for feature selection. I understand it to the point I could generate my own code if needed for all these methods, but most of these methods already have good packages in R or Python that are fast and efficient. At the moment I am proficient, meaning I can write code to get the job done not necessarily up to the standards of software engineer, in C, C++, MATLAB, R, and Python. I also am a second author on a paper where I generated the software in MATLAB for image analysis and maybe more since my code is still being used by the previous lab I worked in.

 

[Locrian]

Sounds like a list right out of Elements of Statistical Learning.

You have skills lots of employers want.

Do those people have jobs you want to do? Not sure, but I bet you'd find them interesting, even if at first you didn't think you would.

Will those jobs still be there when you're done with your PhD? I don't know, but it would blow my mind if they werent.

IMHO concern over future employability should not be a reason to turn down the PhD program. There are other considerations (is the PhD worth getting, instead of going straight to the job? is it work you enjoy doing? etc.) that you should still consider. It sounds like a neat opportunity to me.

Best of luck and I'd love updates as you progress.