Things I have learned as a Data Scientist/Analyst - intern in Healthcare in the past 8 weeks.

A Short Back story

I am an MSc mathematics graduate who took lots of applied statistics classes.  I have a BSc Actuarial Science undergraduate degree too. The two courses sort of set me up on this path and that is how I found myself as an intern in the Enterprise Data & Analytics department of a hospital immediately after graduation.

Now onto the list:

  • The healthcare lingo will throw you off!

For someone who has never liked being sick and thankfully has a very limited number of interactions with hospital personnel, there were some words that caught me off-guard from the first day at orientation. I mean "my doctor" worked pretty fine but I have to refer to them as "PCP - primary care provider". All the other physicians I called "the doctor". (My doctor : the doctor). 
The first project requirements were given to me in this lingo too and I did not want to ask a question after each sentence because that would have turned out to be a lengthy conversation. Thankfully, the supervisor was gracious enough to explain the idea until I had some sort of understanding that guided my curiosity on the data exploration journey that was ahead of me.

  •  Engaging in conversations with fellow employees will make it easier
Since they are not your supervisors in most cases, they offer a more relaxed, calm and easy way to catch up to lingo, give you the organization's perspective, correct and nudge you.  This bit is useful in bringing structure to your thoughts, giving you new avenues to explore or help find a link between your views and the companies views -  collaboration.
  • The feeling that you might not be doing things correctly is overwhelming at the start
Those department meetings and conversations you hear might have you questioning yourself on whether you really are doing your best.  In those moments, it is key to remind yourself that they are talking from a point of experience which you will eventually get. Remember too they do this almost every Monday to Friday and it is their area of specialty. It is probably chit chat to them. It might also be the case that they wonder how you are able to perceive subjects the way you do.

  • Data is available in plentiful
This is probably the most interesting bit - the healthcare industry truly has a lot of data. Of course it is protected  and you cannot access it without proper authorization. The industry provides a good stop for learning all about the first principles/steps of data science; the wrangling/cleaning of data for analysis. There is no running away from correctly formatting date, numerical, factor, logical variables..., reducing the sizes of your data sets by removing columns with no variables or those with a one level value in all rows ... I cannot fully exhaust all scenarios you can possibly think of handling before the analyses start. It is a challenge worth getting yourself into.


Note: 
I cannot wait to see what 6 months at the least has in store for me. 
All the best to those who are thinking of getting into the healthcare industry

~NMN

Comments

Post a Comment

Popular posts from this blog

Financial Mathematics CT-1 Finally Paid Off

Image
 How you may ask,  I sat for my financial mathematics paper in 2017. For those who do not know what this is, here is a summary. In the pursuit to be an actuary, there are exams that you are supposed to take and pass. Financial Mathematics (CT-1) was the name of one of it in the Institute and Faculty of Actuaries (IFoA) exam system before they changed the syllabus and consequently the exam paper names. I passed this paper on the first trial with what I would consider an excellent grade. The surprising bit is, as a unit/course, Financial Mathematics 1, was one of my three lowest scores in my entire undergraduate. So, this felt like part vindication that given different circumstances and motivation, I can do pretty well. I studied for the exam for probably more than 3 months, did almost all the past papers leading to that year and checked them against examiners' reports - thanks to the IFoA  repository . I did have a study buddy during this period. I did 2 more actuarial exa...
Read more

Data Scientist Courses (edX vs DatCamp)

Image
I have had the opportunity to try several online learning platforms. Some of these have been LinkedIn Learning, Pluralsight, edX, DataCamp, Microsoft Learn among many more. You may find particular courses in all the platforms. On this post, I am going to outline chapters from the data science courses on edX and DataCamp. I Audited a data science course on edx around the year 2021. This was my first time going through a complete data science course. A lot of the material was new to me especially the coding. The theories on the other hand were more relatable. It was also the first time before my masters that I got to understand hypothesis testing and got to know why it is never "accept the null hypothesis" but always "Reject or fail to reject the null hypothesis". Some of the chapters on it were hosted outside edX and that is how I discovered DataCamp. The link to the full course is  HarvardX's Data Science Professional Certificate Around the second quarter of 202...
Read more

Self Joins in R

Image
Sometime back at work, I was faced with a dilemma.  I am one of those people who might get stuck on something because I have a "feeling" there is a shorter way to do it. Neither am I the biggest fan of writing several lines of code when there is an option to have less lines. I know that my tendency towards less lines of code sometimes is not the best bacause it might hinder readability.  To explain my dilemma, I have created a sample data frame as follows Code: data <-data.frame(   ID = c(12,12,34,36,47,47,58,68,77,77),   Call_Duration  = c(20,20,25,27,40,40,50,55,20,20),   Call_Reason = c("Reason 1", "Reason 2", "Reason 3", "Reason 1","Reason 3",                   "Reason 6", "Reason 4", "Reason 1", "Reason 2", "Reason 3")   ) Running the code above results to the below dataframe The Dilemma How can I easily find and count duplicate IDs so that I can ration call duration to...
Read more