What a Degree in Public Health Taught me about Data Science

I’ve seen numerous articles, weblog posts, and tweets arguing fervently for or towards the thought of pursuing a complicated diploma so as to enter the information science discipline. 

I believe that this debate is way too context-dependent to have a successful aspect, however pursuing a complicated diploma after faculty was the suitable path for me. Typically, the one levels I’ve seen included in these conversations are grasp’s and PhDs in information science, laptop science, and statistics. While all of these levels have apparent worth and relevance to information science, I want to introduce levels in public well being to the dialog.

The discipline of public well being is all about defending and enhancing the well being of the general public, and it has a wide selection of sub-fields. Two of those, epidemiology and biostatistics, are extremely quantitative and have a lot in widespread with information science. Epidemiology is the examine of the distribution and determinants of illness and different health-related states and occasions, whereas biostatistics is, unsurprisingly, largely targeted on statistical strategies related to well being and drugs. I accomplished a two-year Master of Public Health (MPH) in Epidemiology and gained invaluable abilities and conceptual data throughout the course of the diploma.

So what makes this diploma so well-suited for a profession in information science? Well, most significantly, a lot of the accessible coursework is instantly relevant to careers in information science. There are programs in information science, machine studying, SQL, geospatial evaluation, regression, longitudinal analyses, and an array of superior statistical strategies and examine design approaches*. While the accessible coursework probably received’t be enough in the event you’re fascinated about essentially the most laptop science-heavy branches of information science (e.g. information engineering), there’s lots to be discovered by somebody on the lookout for extra analytic positions.

*Note: Schools range considerably in how a lot they prioritize the R and SAS (used extensively in authorities and non-profit settings) programming languages, which is necessary to think about when applications. Some faculties have largely transitioned to R, whereas others stay firmly dedicated to SAS. In many situations, like mine, it’s possible you’ll study each however rely closely on one or the opposite.

It’s value noting that the term “data science” itself has only existed since 2001. Epidemiologists and biostatisticians, nonetheless, have been utilizing lots of the strategies that now fall underneath the information science umbrella for many years. This longevity is necessary to think about when looking for graduate applications. While information science diploma applications have been popping up left and proper over the previous few years, most colleges of public well being have been round for much longer. This implies that they’ve had time to curate their colleges, develop a curriculum over many iterations, and construct a robust alumni community that may be a necessary useful resource afterward in your profession. You received’t be one of many first few courses of scholars whereas professors work out the kinks of their programs — as an alternative, you’ll probably be a part of a well-oiled machine the place professors have been working collectively for years and have developed programs that complement one another properly.

Perhaps the best good thing about epidemiology coursework is the wealthy context it gives for fascinated with information. The final aim of epidemiology is transferring past correlation to determine causation, and there’s a robust concentrate on examine design, potential bias, and correct use of advanced statistical strategies to realize this terribly tough aim. The stakes are additionally excessive — epidemiological work is used to make scientific, coverage, and funding choices with very actual, very human penalties — so questions on information high quality and statistical validity are taken severely. After two years on this diploma program, I completely take a look at information in a different way. My intestine intuition is to suppose critically about what inhabitants we’ve actually captured in our information (in distinction to the inhabitants we meant to seize), in addition to what questions we will reply validly primarily based upon the information assortment course of. While it’s pretty simple to search for a picture classification tutorial and practice a mannequin to detect malignancies on pictures of pores and skin, epidemiologists are educated to be considering critically in regards to the information, equivalent to questioning whether or not or not that mannequin was educated utilizing the total vary of pores and skin tones (they’re often not) and the implications that failure to take action could have for mannequin efficiency in populations which can be already marginalized. These questions have necessary penalties outdoors of healthcare too (suppose the facial recognition software program utilized by police departments), and coaching in epidemiology makes it simpler to acknowledge and confront these damaging pitfalls. Epidemiological considering is an asset in way more benign contexts as properly, equivalent to contemplating whether or not or not your gross sales information precisely represents all your prospects, or whether or not or not you’ll be able to actually attribute that change in gross sales to that advert marketing campaign.

If you have an interest in working in information science within the healthcare sector, a level in public well being is especially related. It is extremely worthwhile to study statistics and algorithm utilization with actual healthcare information within the context of real-world issues, and there are a lot of elective programs you could take to realize a greater scientific understanding of the illnesses you’ll be working with. These programs, equivalent to “Cancer Epidemiology” or “Psychiatric Epidemiology”, additionally cowl the examine designs and analytic strategies most incessantly used to review these illnesses — materials you most likely wouldn’t cowl in a normal statistics or information science diploma. Many information scientists who transfer into healthcare afterward of their careers lack this industry-specific data, and it may well set you aside in an utility course of in addition to making it simpler to do your job properly.

Finally, most respected public well being applications be certain that you graduate with extremely marketable expertise. Again, the truth that these applications have been round for many years implies that they’ve had time to develop relationships with quite a lot of corporations and organizations and may also help you get your foot within the door. My college required a “practicum” expertise (primarily an internship) between the primary and second 12 months of this system, and supplied help find and securing these worthwhile alternatives. Having this requirement and help makes it far simpler to realize information science expertise at a healthcare firm or associated group, which might additionally result in a job provide down the street. Additionally, most applications require a thesis as a commencement requirement. While sometimes far much less daunting than PhD dissertations, epidemiology and biostatistics grasp’s theses are primarily publishable year-long information science tasks that present glorious expertise to characteristic in your resumé and produce up in job interviews.

While my MPH expertise was overwhelmingly optimistic and the suitable alternative for me, I firmly imagine that the choice to pursue any superior diploma is very private and dependent upon your distinctive circumstances. However, I do imagine that in case you are trying round for a level program, particularly in case your long-term aim is to work in well being, then quantitative levels in public well being are greater than value your consideration.


Please enter your comment!
Please enter your name here