Overarching research question: How can we use the abundance of existing digital data, aka big data, (e.g. government administrative data, electronic health records) to support accurate evidence based decisions for policy, management, legislation, evaluation, and research while protecting the confidentiality of individual subjects of the data? This question focuses on the data science of using massive secondary datasets, a step before the traditional statistical methods can be applied to the data to address specific questions to improve population health (public health).

Preferred approaches: Data Science – To build efficient and effective human computer hybrid processes and systems to clean, integrate, and extract valuable information from raw chaotic data and deliver the information in a timely manner to decision makers (e.g. researchers, policy makers, clinicians) to improve public health.

Balancing Personal Privacy Risk vs Social Utility Benefit

Balancing the benefits for social good with the risk to personal privacy in Population Infromatics is one of the most challenging issues. Both are very important, and we must find a way to find the sweet spot that is acceptable in both dimensions. Click here to read my vision of how we might do this based on the Minimum Necessary Standard behind most privacy ethical and legal frameworks such as GDPR and HIPAA, using the Privacy-By-Design approach.

In sum, keeping personal data safe from the threats of the Internet it very challenging. Maintaining privacy and confidentiality of personal data while having sufficient information for meaningful use requires a well-orchestrated system that was designed with privacy in mind from the get go. The appropriate design lies in the art of balancing usability and privacy through privacy-by-design. Usable and affordable privacy requires all stakeholders, including the public, to do their due diligence that is enforced through transparency and accountability.

What is Data Science

Some quotes I like

• Telling your story with data using the scientific method (Hye-Chung Kum)

• Data Science is a mash-up of several different disciplines (Data Engineering, Scientific Methods, Math, Statistics, Advanced Computing, Visualization, Hacker Mindset, Domain Expertise). We also noted that an individual data scientist is most likely an expert in one or two of these disciplines and proficient in another two or three. There is probably no living person who is expert in all these disciplines, and an extremely rare person would be proficient in 5 or 6 of these disciplines. This means that data science must be practiced as a team where, across the membership of the team, there is expertise and proficiency across all the disciplines (Wikibooks: Data Science Intro).

• Knowledge of R , Python or other tools is secondary to knowing how to approach the data, how to ask right questions, and good intuition about what works and what not (Gregory Piatetsky-Shapiro, Analytics/Data Mining Expert, KDnuggets President).

• as one long-time “data scientist” I love the new term for what I do. I think it aptly describes what I do, what I’ve always done, with data (Daniel Dean Gutierrez, Data Scientist at AMULET Analytics).

• And, of course, a natural curiosity on how things work and the ability to hire and manage other folks who also have a passion for information. (Jim Lola, Entrepreneur, Sr. Manager, Technologist, Architect, & Author )

Some useful Links:

• What is Population Informatics

• What is Data Science

• How to become a data scientist

• Actionable information from data: Always ground yourself in the real world problem

• Data science in the health care industry

Required Skill Set: Many have written on the essential skill set for Data Science. The following are common across most and those that I agree with in the order of importance.

• Curiosity: a desire to go beneath the surface and discover and distill a problem down into a very clear set of hypotheses that can be tested with data

• Communication: The ability to communicate and translate between data, models, and real world problems

• Critical Thinking: The ability to reason and think through a real problem, understand the data at hand with little documentation, the limitations, and get the best answer possible for the given problem and data. The ability to know how much accuracy is required in the given problem and an intuition about the accuracy of the data at hand.

• Cleverness: The ability to look at a problem in different, creative ways

• Programming (software engineering): The ability to use the tools (programs like R, SAS, Phython, Excel etc) to manipulate data easily into any shape, and develop efficient processes of converting data into information that can be iteratively updated and adjusted easily. Developing tractable and agile processes with sufficient but minimal documentation is key.

• Statistics/Math: The ability to understand how to take data points and build models

• Domain Knowledge: The ability to either know the domain knowledge (i.e., healthcare, employment, education, welfare, etc.) or learn it quickly or work (communicate well) with a domain expert

Interested to become a Data Scientist? You can try some data science courses at Coursera.

There are a lot of resources online to learn data science. Currently, there are 3 primary languages used in the data science community - R, python and SAS. The first two are open source and are hence gaining rapid adoption but the last one is still widely used in medical systems. If you are conflicted between choosing R and Python here is a good resource that can help you choose.

Basics of data science:

1. r4ds or R for data science is a book that teaches the basics of data science using R and particularly, a set of libraries in R called as the "tidyverse". An extremely accessible and incredibly handy resource that any beginner will benefit from. Though it teaches R libraries, the fundamentals apply to any language and data science in general! It also has a community if community driven learning and a place to discuss your learning sounds good.

2. If you like learning by doing and what is called as interactive learning, datacamp has a huge catalogue of bite sized courses and learning tracks that can help you navigate data science. Datacamp is taught by experienced professionals and content creators and can be really beneficial if a few minutes of video and lessons followed by actual coding to accomplish a task to reinforce the learning, sounds like something that you might want to do. Typically, a susbsription is needed for personal use but a classroom subscription is entirely free if you can get it. Has courses for both R and python and a few courses on SQL, Git and Shell. Recommend learning just the basics of data science - data wrangling, tidying, etc. here and using other resources as you level up.

Machine learning/Deep learning

1. Fastai is hands down the best resource on the internet to learn machine learning and deep learning from. A heavily application focussed teaching method that gets you started on problems from the get go, you can learn cutting edge techniques to achieve state of the art results for any of your problems.

2. Andrew Ng's Coursera courses for machine learning and deep learning are tried and tested and you cannot go wrong with them.

Data Science Videos

• Hans Rosling’s TED Talk

• Data Sharing and Management Snafu in 3 Short Acts

• Wealth Inequality in America

  • income disparity

  • wealth disparity

  • total net worth

  • total financial net worth (total net worth minus housing

• Income Mobility (cross sectional vs longitudinal)

• Learning…(not a data video)

• Texas State of Mind – “Perceptions. Reality. Stigma.” The Mental Health Landscape