Visualize Global Health | 

It’s a new dawn for global health data borne of necessity, mind-numbing numbers, Netflix and a desire to avoid insanity.

“For our own sanity, we needed to create a new way to look at this stuff,” said Peter Speyer.

Speyer, head of data development at Seattle’s Institute for Health Metrics and Evaluation, explained why he and his colleagues are transforming a massive collection of health data known as the Global Burden of Disease (GBD) into a stunning collection of powerful online and interactive visual tools. Go to the link; below is just a screen grab. Seriously, go there and try these out. You’ll have fun even if you don’t know yet what you’re doing.

GBD visualizations

Today, Bill Gates and Speyer’s boss, IHME director Chris Murray, officially unveiled some of those tools aimed at allowing anyone (even you) to dig deeper into these global estimates arrived at by some 500 researchers working in collaboration worldwide for five years on more than 200 million results tracking the impact of nearly 300 causes of death and disability in 187 countries.

Phew. It makes your head hurt just to read that sentence. Imagine trying to compile a complete report including all of the numbers, statistics and charts.

“That’s one of the most exciting things about this phase of the project,” said Murray, who with his long-time partner in death-and-disability number crunching, Alan Lopez of the University of Queensland in Australia, has been trying for decades to create a reliable yardstick for measuring what’s going on in global health.

Chris Murray
Chris Murray
IHME

“I think these visual tools represent a significant paradigm change for global health,” Murray said. “They engage even the most data-resistant people. It’s an incredible revelation to me how profoundly influential it can be to present the data in this way. I think data visualization will be revolutionary for global health.”

You’d have thought that these global health number-crunchers wouldn’t have been so surprised. After all, our brains are wired for vision more than words or numbers. “A picture is worth a thousand words.” And the online TV/movie service Netflix already played a big role in helping them with their data. I’ll get back to Netflix in a bit.

First, here’s a video I took of Speyer demonstrating one tool the GBD arrow diagram for examining causes of death in Afghanistan.

Speyer also did his own video, of another very cool tool called GBD Compare in which you can do amazing things like explore and compare health data by gender or age within a country, or compare disease burdens or trends between two countries. It’s more professional and certainly better as an instructional video than mine, but also longer and wonkier. Still, it will help to watch before you go try to play with GBD Compare cold.

The massive study, the GBD, was released late last year by the IHME, a branch of the University of Washington created in 2007 with a $105 million donation from the Bill & Melinda Gates Foundation. It is the culmination (to date anyway) of the work by Murray and Lopez that they began in 1990 when they were both at the World Health Organization aimed at finding a better measure of health. Back then, health policies were based just on fairly crude (often wrong) death rates.

Murray and Lopez sought to create a new, more accurate way of measuring health that improved death data but also factored in disability. They created a unit of measurement known as the DALY, or disability adusted life year. Earlier versions of the Global Burden of Disease were based largely on these DALYs, which revealed some surprising results – such as the huge amount of ‘life years lost’ due to chronic diseases and mental illness.

But it should be noted that these data revelations didn’t always shift policy.

Mental illness remains off the radar when it comes to most global health policy discussions. And despite all the rhetoric these days about chronic or non-communicable diseases, most of the global health funding still goes to the traditional target of infectious disease partly because these are easier to target, with vaccines and the like. (Cost-effectiveness of the intervention is as much, or sometimes more, a driver of policy as is disease burden.)

Everybody claims they want evidence-based policy. Bill Gates, as evidenced by his annual letter if not his entire life, is a big fan of measurment and evaluation. But what’s to prevent policy makers from again ignoring the data they don’t like and just doing what they want?

“I think the combination of having much more powerful estimates combined with these new tools will make all of this information more accessible while also shifting the discussion more into the mainstream,” Murray said. In short, he thinks by making data more fun and democratic policy can’t help but be influenced.

Visualization was not considered that important when this version of the Global Burden of Disease was launched.

“It just emerged, first by just dealing with internal needs,” said Speyer. As the amount of data piled up, he said, it became increasingly difficult for researchers to manage it. “We were starting to lose the forest for the trees.”

But as the GBD progressed, Speyer said, it became clear that visualization of the data would also address a number of needs, beginning with transparency and accountability. Murray and Lopez have often ruffled feathers in global health because changing health estimates can also threaten programs, agendas and funding. As the analyses have grown more complex, many complained the Seattle metricians were becoming more like wizards carrying around an inscrutable formula few could study or challenge.

“Visualization not only makes the finding more comprehensible, it allows them to explore it and often review the original data,” Speyer said. “It can be as fine-grained as you want it to be.”

Oh, and how did Netflix help?

NetflixYears ago, as Murray and his hundreds of collaborators continued collecting all their hundreds of millions of study results on the nearly 300 causes of death and disability in nearly 200 countries (ouch! brain!) they recognized that some of the findings and trends didn’t quite match up right. Some studies contradicted the majority of other studies. Certain kinds of data were distinctly different — and not easy to combine — with other data.

It was becoming sort of a mess, like asking someone else to predict the next movie you’re most likely to watch based on all the other movies and TV shows you’ve ever watched.

“We decided to apply the Netflix approach to death modeling,” said Murray. Or more precisely, the Netflix algorithm.

You may not be aware of it, but some of the world’s best mathematicians helped Netflix figure out how to take the apparently sporadic selections we all make when we decide which TV show or movie to watch and transform this into a fairly successful prediction of future behavior. Netflix paid some geeks a million dollars to do this.

“What they found was that if you combine a number of mediocre statistical models together as opposed to trying to focus in on the single most predictive model, you get a better result,” Murray said. This approach allowed the UW number-crunchers to combine a lot of imperfect and incomplete data regarding any particular disease from different sources and end up with a more reliable picture.

“It’s a complete change in the way we model cause of death and disability,” Murray said.

So let the revolution begin. Visualize Global Health Now!