Q: What exactly is your job?

Martinez: I work for the Pan American Health Organization (PAHO), the Health Bureau for the Organization of American States (OAS) and the World Health Organization (WHO) Regional Office for the Americas. My position at PAHO is as a specialist and advisor in health metrics, where I work generating evidence about the health situation in the Americas and providing advices on the use of population health metrics, measures, application of methodological approaches and analytic methods to turning health data into information. I have also implemented analytic methods and tools in user-friendly software to be used by public health professionals in the field (national and sub-national levels) to support health situation analysis in limited resource settings (developing countries).

At the beginning of 2009, I was leading the development and implementation of PAHO’s public health information platform, so I evaluated products from Tableau Software (Tableau Desktop, Tableau Reader and Tableau Server) as candidate BI components to put on top of a public health data warehouse. I ended up selecting Tableau as the best solution.

My current focus is non-communicable diseases (NCDs) and their risk factors. I work in a wide range of tasks:

1. Data management and integration: collecting, integrating and maintaining the NCDs-related data repository (data warehouse on MS SQL Server) for analysis, monitoring and evaluation, and information dissemination.
2. Data analysis and generation of information: data analysis and application of methodological approaches to turn data into insights and documenting evidences related to situations and trends of NCDs and their risk factors, including the application of visual analytics to produce information products that better communicate key findings and messages to national health authorities, health professionals and the general public.
3. Dissemination of information: managing the Non-Communicable Diseases and Mental Health Data Portal (NMH Data Portal), including design, information architecture and publication of data and information products. 

Q: How do you use Tableau in your job?

Martinez: I use Tableau as visual analytic tool for a variety of tasks:

1. Data quality verification and validation.
2. Data exploration, transforming data into information and producing compelling information visualizations that better communicate key findings to the intended audience.
3. Elaboration of tools (in form of interactive information visualizations and dashboards) to support decision making in public health within the organization at executive, management and technical levels.
4. As a tool to share population health situations, health issues and insights to a broad audience, including policy-makers and decision-makers in public health, health professionals and the general public.

Q: Your blog, Health Intelligence, is wonderful. How do you decide what to post? What are your goals related to your blogging?  

Martinez: The idea of blogging was circulating in my head for some time, maybe a year before I created Health Intelligence as a data blog in March 2011. I had the curiosity to write and share my ideas and thoughts. I had some papers published in scientific journals, a few book chapters and some technical reports published, but I thought that it was nice to write to a more wide audience, to write and publish quick notes about public health and natural sciences using data and information visualizations. My first blog was about just trying to define to myself what business intelligence is, and the second one visualized the Virginia Earthquake that occurred on Aug. 23, 2011. After that, I added some blog posts featuring information visualizations that I found out surfing on Internet and that are kind of inspiring work for me.

The blog post that I feel defined my beginning as a blogger was “Fungal Meningitis Outbreak in the US, 2012,” where I applied the concept and method of Epidemiological Curve to analyze outbreaks in an interactive visualization. The visualization was selected as “Viz of the Day” on October 31, 2012, by Tableau Public. This was great stimulus from Tableau Public for creating new visualizations and blogging. Some close friends and colleagues also encouraged me to blogging more frequently.

I decide what to post based on current events, news or issues that I learn or meet during any normal day. Sometimes they’re based on something that I read in a journal or in a newspaper, or on a public health issue that is occurring, or in a tweet that I read, such is the case of this blog post. Or, I found a data set that I think would be interesting to visualize, mainly related to population health and natural sciences. For example, when the 7.0 magnitude earthquake struck Haiti on January 12, 2010, the news was in everywhere. So, I decided to visualize that event. I went to the USGS web site looking for data from Haiti earthquake and aftershocks. The visualization I created about the Haiti Earthquake was one of the first published in Tableau Public, a few days after Tableau Public was launched. I included that viz in a blog post celebrating three years of Tableau Public. I maintain a list of topics that interest me, but I normally only blog about a few of them.

My goal with the Health Intelligence data blog is to share public health topics with the general public, provide exploratory tools that allow other people to learn more about a specific population health issue, and demonstrate how we can use data and information visualizations to communicate a message or an idea in a more effective way. 

Q: Do you think the Healthcare industry in general is utilizing Tableau well?

Martinez: I think that Healthcare, compared to other sectors, is always behind in adopting new technologies, and Tableau is not an exception. In my area of influence, that is epidemiology and public health institutions in the countries of the Americas, I don’t see that Tableau is extensively included in the toolbox. Some health institutions, mainly from the Ministry of Health, at a national level are utilizing Tableau but just partially, with one or two analysts using Tableau Desktop and sharing packaged workbooks with peers that are using Tableau Reader. There is a lot to do in this regard, including training, development of a culture of information and decision making based on evidence. Implementation cost and competing priorities can become barriers for adopting new technologies in public health and healthcare services.   

Q: What kinds of challenges do you deal with in your role in regards to data? How do you overcome them?

Martinez: I would say that the most frequent challenge is related to data quality and availability, including the format/media in which the data is structured and shared. Many times, there are no data available for talk about a specific health issue. In other cases, the data is published as part of a report in PDF format. Frequently, the data require transformation and cleaning.

Just to put a recent example, data from Ebola virus disease outbreak in West Africa. At the beginning of the outbreak, World Health Organization (WHO) shared the reported number of cases in plain text just as part of a report, so this viz required spending a lot of time preparing and updating the data set. The frustration came when WHO stopped reporting data at the sub-national level, Then, I did an alternative viz.

To avoid quality issues, I prefer to use official data produced by recognized institutions. You can see my list of preferred data sources here.   

Standardization is another common challenge. I mean standardization of terms and codes used in dimensions and use of the same definition of the metrics and measures. This scenario usually happens when multiple data sources (that usually come from different institutions) are needed for comprehensive analysis.  

I overcome challenges and issues with the data by learning about the data source, including data collection and estimation methods (if applicable) as well as using additional tools to transform data formats and structures. In the last year, I started using Alteryx, which allowed me to be more efficient in terms of data blending, transformation and integration.  

Q: How did you learn proper data visualization techniques/best practices?

Martinez: It is interesting because I never received any formal training or course on design-related disciplines. I started feeling the need to apply aesthetic and perception principles in data visualization when I began working at PAHO, developing SIGEpi, a Geographic Information System (GIS) computer application for epidemiology and public health. At that time, the team I was part of discussed user interactions, computer-human interfaces, use of colors and other design considerations. I remember my presentations and charts were ugly at that time.

I’ve learned about proper data visualization techniques and best practices by doing, testing and assessing results, and by trying to replicate good practices done by experts from information design and visualization fields. Of course, I’ve spent a lot of time reading about data visualizations.   

Q: You do a lot of discovery and analysis with Tableau. How do you approach discovery work using Tableau?

Martinez: I apply these steps:

1. Getting familiar with the data set, including its dimensions and measures, structure and granularity.
2. Creating basic frequency distribution by dimensions; I learn about every dimension member.
3. Looking at every measure by simple dimension first and then combining dimensions.
4. Two specific dimensions that I always pay special attention: time and space. Visual representations of time series and geographic distribution provide an enlightening perspective of the data.
5. Applying techniques that allow me to learn about the distribution of the data, including trends, dispersions and specific identification of outliers. Using multiple data classification methods and visual representations such as box plots and scatter plots.     
6. Finally, I apply more sophisticated methods to expose more stories in the data.

Q: What are the blogs you read … the books, magazines? How do you monitor Tableau, the data world and health issues?

Martinez: I follow a lot of blogs related to: public health and global health, statistics and operational research, business intelligence, data and information visualization, analytics, visual perception, GIS, machine learning and storytelling. I also follow web sites from United Nations agencies, national health institutions, scientific journals and newspapers.

I use Feedly to keep track of new blog posts and articles from sites I follow.

I keep a curated list of blogs, blog posts and web sites about topics of my interest. These are some of them: Tableau training,  Data Tools, Data and Visualization sites, reading for data visualization, Best of the Web, Online Courses and some readings for Health Analysis. As you may see, I also use Health Intelligence as my public notebook.  

I read books mainly related to my professional interests, and I like books about psychology, sociology and social sciences.  

Q: What public health data sources do you think people might be interested in? What are your favorites?

Martinez: As I mentioned before, I prefer official data sources, such as data from World Health Organization (WHO), United Nations agencies, internationally recognized institutions like the Institute of Health Metrics and Evaluation (IHME) and national institutions.

I have made available my list of data sources on the Health Intelligence website precisely because I believe that those data sources could be of interest to other people.