Subscribe to DSC Newsletter

From academia to data science in healthcare: Is the learning path any different?

I’m transitioning from almost a decade of academic bench research (B.S. in neuroscience), publications, and advanced statistical models (SPSS, SAS proc mixed). My goal is to do scientific analyses in healthcare/pharma (not much interest in marketing/finance). I’ve read a lot about different paths and perused healthcare DS job postings which almost always require Master’s (MS) or PhD. It seems that someone with my lack of background in CS would greatly benefit from getting an MS in Analytics or Data Science, which itself requires prior Python/R programming knowledge to be admitted. So my goal is to first go through DataQuest’s online courses to learn Python—my understanding is that unlike DataCamp, DataQuest has a more solid intro to Python and introduces subjects in a highly hierarchical manner without course overlap, which should be more efficient. After that, I intend to complete the online MS in Analytics from Georgia Tech and build a portfolio/compete in Kaggle while working as an analyst/statistician full time.

I wanted to ask the experienced data scientists here: given my career goals in healthcare/science, does this seem to be an efficient path to data science or would you personally do something different? I have been also considering the new Data Science Nanodegree from Udacity since they provide mentorship and help with job search and resumes, but I am not sure if that’s a solid path to data science for $2K. Age factor: approaching mid-30s. This is for the USA, but moving to Europe for jobs is a possibility.

Views: 317

Reply to This

Replies to This Discussion

My experience is opposite. Not that I am looking for a job in Academia, but more and more, I am doing academic-level research in theoretical & applied math, on my own, that involves a lot of data science in a setting very similar to "experimental design" in order to gain insights about how to solve some mathematical conjectures. My question is how to get respected by these math gurus from top universities. Initially my articles got little traction from this community, but more and more, over time, I am gaining respect, even from top University professors, who want to continue exchanging ideas with me. It could lead to something interesting...

How did this happen and how can it relate to you? Here is my answer (in my case):

  • Bringing a new perspective, working on off-the-beaten-path topics where it is easier to make a big contribution
  • Being modest, discrete, yet perseverant and very relevant. 
  • Reaching out to top professors, mentioning my research and the fact that I am citing their books / papers
  • Doing better research to find modern references related to my math research, and to not appear as an "amateur reinventing the wheel"
  • Learn new tools/apps to better compete with great mathematicians.

Healthcare + data science is a popular topic today. There are so many issues to be addressed, plenty of low hanging fruits. I even thought about launching a startup on the subject, see here.

Thanks for your response – I have read quite a few of your articles on this blog and have found them insightful. Indeed, I am going into data science precisely because healthcare has untapped potential requiring thoughtful analyses. There are a lot of opportunities not just for saving costs for patients/hospitals as you rightly mention in that article, but also specifically scientific inquiries investigating even simple questions such as surgical outcomes (and hence insightful conclusions and publications). All that great data is quietly buried, and few are looking at them. I have several ideas which would qualify for a startup, but being an academic, I teamed up with a professor to go the grant route instead. Didn’t get funded (of course!). Nevertheless, the opportunities are there. The data in healthcare is just waiting to be probed…  I do have certain moments of doubt because some “inefficiencies” might be in the interest of the hospitals. Cutting out extra services can negatively affect their bottom line and I imagine most hospitals will not take kindly to this business model, especially when they’re paying a data scientist to augment essentially their business. So there are some nuances that have to be worked out.

Regarding the more mundane part of obtaining a degree, it seems you’re suggesting I could do the equivalent of publishing, but in data science (perhaps projects/portfolio?) to enter the data science realm. Let me know if I misunderstood. Presumably, my plan to obtain an online masters in addition to profile building is not misguided. I considered a PhD, but several professors discouraged that route to data science. MD I am still considering, since being a doctor with data skills will allow for most control.

I have a clinical background (BS in Clinical Laboratory Science) and have made the jump to a technical data analyst (not quite data scientist, but a good jumping off point at least in my company). I'm finishing a master's in health informatics at UIC. I didn't find that technical enough so struck out on my own on Kaggle, DataCamp, and CodeCademy.

I'm fortunate enough that my company allows me to develop and have written several customer-facing reports in Python, using both folium and bokeh libraries. Also we use JIRA, Confluence, and GitHub for code storage and version control, plus SSRS. If you want to "publish" your data science chops, make sure you have a public page on GitHub. It's something that I'm just realizing and trying to make sure I improve upon.

Healthcare is a bit of a pain, but rewarding if you're working with EMR data. Right now I'm working more with claims data, which is a headache and twisted especially considering the switching between BCBS/private payers and CMS file formats.

I will say, the MS in Analytics from Georgia Tech looks promising, especially for the price tag ($10k). However, if you're ok with a little more debt OR can get your employer to reimburse a master's or part of it, Northwestern is offering an online MS in Health Analytics starting Fall 2019 that actually looks a little promising and more targeted.

I don't think there's ever really a "correct" path, but more just making sure your passion shows in an interview. I will say that my real-world healthcare experience (worked in both patient-facing roles and in a high-complexity academic hospital diagnostic lab) helped immensely in getting my foot in the door, and allowed me my opportunity. Healthcare experience and knowledge is more appreciated than the data side because it's harder to come by, and domain knowledge always helps improve an analyst's ability. I've been told by multiple recruiters, plus inside my own company that I have a rare skill set. We'll see how long that holds true.

This is all just my experience, and again, just keep persisting and your opportunity will come!

Hi Abel,

First, some info on my background: I have a MD degree and I am currently working as a data scientist.  After I got the MD degree, I realized medicine wasn't for me, and so I went back to school to get a MS in Computational Science and Engineering from Georgia Tech (which doesn't require a BS in Comp Sci).  I am also a BS in Neuroscience,so looks like we have some similarities there.  I worked for two healthcare companies doing data science before switching to a different industry.

I think your plan on learning data science through online courses is good.  IBM has a new multidisciplinary data science specialization in Coursera that looks like it has all the necessary components for a data science job (Python, SQL, machine learning, etc.).  I would focus on getting a formal degree from a university (like Georgia Techs online MS, as you mentioned) as well, as this will differentiate you from the thousands of other self-starting data scientists.  The cost is very reasonable (more than 2K, but the extra qualification will be worth it when you look for positions).  

In a follow up comment you mentioned getting an MD degree.  I would discourage that just for the data science purpose.  Not worth the insane amount of debt that you will go into.  You can just learn some clinical terminology and that should be good.  

Also, not to self-promote, but I did recently write a book on healthcare data science at an introductory level.  Feel free to send me a private message, and I can send you a free copy if you like.  

Hope this helps!

RSS

Videos

  • Add Videos
  • View All

Follow Us

© 2018   Data Science Central ®   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service