Tag Archives: my research

Machine learning in population health: Opportunities and threats

My colleague Theo Vos and I have a perspective published recently in PLoS Medicine, Machine learning in population health: Opportunities and threats. It is not long, so you can skim it in seconds, or read it all in just minutes.

It is not directly related to a short film that I enjoyed recently.  Maybe indirectly.

Comments Off on Machine learning in population health: Opportunities and threats

Filed under Uncategorized

Life Expectancy and Cause-Specific Mortality by Census Tract in King County, Washington

Some new research that I’m excited about came out last week: Variation in life expectancy and mortality by cause among neighborhoods in King County, WA, USA, 1990–2014: a census tract-level analysis for the Global Burden of Disease Study 2015. http://www.thelancet.com/journals/lanpub/article/PIIS2468-2667(17)30165-2/fulltext

In some ways, it is very specific to Seattle and the surrounding county: https://vizhub.healthdata.org/subnational/usa/wa/king-county

But it is also a demonstration of the “fractal” nature of population health—the variation between life expectancy from country to country around the world is big! But it is around as big as the variation between life expectancy from county to county around the United States. And what this work shows is that even in the county where I live, the life expectancy varies between census tracts almost as much as from county to county or country to country. Inequality is happening at all scales.

Here is the data: http://ghdx.healthdata.org/record/united-states-king-county-washington-life-expectancy-and-cause-specific-mortality-census

Comments Off on Life Expectancy and Cause-Specific Mortality by Census Tract in King County, Washington

Filed under disease modeling

New Paper: New challenges for verbal autopsy: Considering the ethical and social implications of verbal autopsy methods in routine health information systems



Verbal autopsy (VA) methods are designed to collect cause-of-death information from populations where many deaths occur outside of health facilities and where death certification is weak or absent. A VA consists of an interview with a relative or carer of a recently deceased individual in order to gather information on the signs and symptoms the decedent presented with prior to death. These details are then used to determine and assign a likely cause-of-death. At a population level this information can be invaluable to help guide prioritisation and direct health policy and services. To date VAs have largely been restricted to research contexts but many countries are now venturing to incorporate VA methods into routine civil registration and vital statistics (CRVS) systems. Given the sensitive nature of death, however, there are a number of ethical, legal and social issues that should be considered when scaling-up VAs, particularly in the cross-cultural and socio-economically disadvantaged environments in which they are typically applied. Considering each step of the VA process this paper provides a narrative review of the social context of VA methods. Harnessing the experiences of applying and rolling out VAs as part of routine CRVS systems in a number of low and middle income countries, we identify potential issues that countries and implementing institutions need to consider when incorporating VAs into CRVS systems and point to areas that could benefit from further research and deliberation.

Comments Off on New Paper: New challenges for verbal autopsy: Considering the ethical and social implications of verbal autopsy methods in routine health information systems

Filed under Uncategorized

New Publication: Implementing the PHMRC shortened questionnaire: Survey duration of open and closed questions in three sites




More countries are using verbal autopsy as a part of routine mortality surveillance. The length of time required to complete a verbal autopsy interview is a key logistical consideration for planning large-scale surveillance.


We use the PHMRC shortened questionnaire to conduct verbal autopsy interviews at three sites and collect data on the length of time required to complete the interview. This instrument uses a novel checklist of keywords to capture relevant information from the open response. The open response section is timed separately from the section consisting of closed questions.


We found the median time to complete the entire interview was approximately 25 minutes and did not vary substantially by age-specific module. The median time for the open response section was approximately 4 minutes and 60% of interviewees mentioned at least one keyword within the open response section.


The length of time required to complete the interview was short enough for large-scale routine use. The open-response section did not add a substantial amount of time and provided useful information which can be used to increase the accuracy of the predictions of the cause of death. The novel checklist approach further reduces the burden of transcribing and translating a large amount of free text. This makes the PHMRC instrument ideal for national mortality surveillance.

Also with a replication archive on the Global Health Data Exchange (GHDx) [http://ghdx.healthdata.org/node/263527].

Comments Off on New Publication: Implementing the PHMRC shortened questionnaire: Survey duration of open and closed questions in three sites

Filed under Uncategorized

SIAM Mini-tutorial Recording online

Dear AN16 speakers, this one contains the correct link,

The sessions recorded at the Annual Meeting (AN16) in Boston are now available on SIAM Presents…Featured Lectures from Our Archives


You are welcome to link to your talk from your personal web pages and point to the site. You can find your presentation using the search box on the upper right.

We will post links from the SIAM web site to the presentations and will send out notification to attendees shortly. I expect the recorded sessions from the Conference on the Life Sciences (LS16) to also be available soon.

Please do not hesitate to contact me if you have questions.

Sorry about the error in the first email.



Linda C. Thiel
SIAM Director, Programs and Services

Comments Off on SIAM Mini-tutorial Recording online

Filed under global health

Write-up of my SIAM Tutorial

That was nice of them to do: https://sinews.siam.org/Details-Page/machine-learnings-impact-on-global-public-health

Comments Off on Write-up of my SIAM Tutorial

Filed under global health

U.S. county mortality paper

The U.S. county mortality paper, a study analyzing 21 cause groups of death in every U.S. county from 1980 through 2014, was published in JAMA on December 13th along with a trove of other useful resources on county health including updated county profiles, an updated US Health Map data tool, a new US Data GHDx page, a new animated GIF, and two videos produced by JAMA.

Congratulations to IHME study authors Laura Dwyer-Lindgren, Amelia Bertozzi-Villa, Rebecca Stubbs, Chloe Morozoff, Michael Kutz, Chantal Huynh, Ryan Barber, Katya Shackleford, Abraham Flaxman, Mohsen Naghavi, Ali Mokdad, and Christopher Murray.

Additional congrats to the Global Engagement Team (GET) members and alumni involved in the dissemination of these important findings: Dean Owen, Kevin O’Rourke, Kate Muller, Bill Heisel, Dawn Shepard, Sofia Cababa Wood, Katie Leach-Kemon, Adrienne Chew, Pauline Kim, Rachel Fortunati, and Kayla Albrecht.

Stories by CNN, HealthDay, NBC, and Reuters were picked up by hundreds of local news stations and papers across the nation, totaling nearly 500 media mentions since 8:00am Tuesday. Here are a few of the top news stories covering the paper; many include their own graphics using IHME county mortality data:
• Janet Adamy with the Wall Street Journal wrote What kills Americans varies widely by region. ““It’s much more complicated than saying ‘Everything’s bad in Mississippi and Alabama, and everything’s good in places with high life expectancy,’” said Christopher J. L. Murray, director of the Institute for Health Metrics and Evaluation at the University of Washington and an author of the study.”
• Olga Khazan with the Atlantic wrote Why are so many Americans dying young? “’A place like Colorado, there’s an incredibly low death rate for heart disease, one of the lowest in the world, and low rate for diabetes,’ Murray said. ‘If you look at places like West Virginia, things are getting worse, and it’s not just opioids.’”
• Jacqueline Howard with CNN wrote What’s the most common cause of death in your county? “’We know that unequal access and quality of care create health disparities in the US for many causes of death, while other causes are linked to risk factors or policies. The results of this study prompt future research to further identify what drives health disparities in our country,’ said Dr. Christopher Murray, a professor and director of the Institute for Health Metrics and Evaluation at the University of Washington, who was a co-author of the new study.”
• Anna Maria Barry-Jester with FiveThirtyEight wrote How Americans die may depend on where they live. “Lead author Laura Dwyer-Lindgren, a researcher at the Institute for Health Metrics and Evaluation at the University of Washington, says she hopes the data can be useful to local health workers and the public. ‘If you go to any state health coordinator, they probably know what was recorded on the death certificates. But it can be really difficult to interpret them,’ she said. She hopes that collapsing the various causes of death down to 21, rather than looking at everything that can kill a person, will make it easier to target regional problems.”
• Maggie Fox with NBC News wrote Where you live determines what kills you. “’Heart disease is particularly high in the southeast of the United States,’ said Murray, who has pioneered many different ways to crunch health statistics. Experts know lifestyle — poor diet, a lack of exercise and less access to good medical care — are mostly to blame.”
• Andrew Seaman with Reuters wrote U.S. death rates vary drastically by county. “’Within any individual county, knowing how big of a problem a condition is’ can help counties know which conditions need attention, resources and policies, said the study’s lead author Laura Dwyer-Lindgren, of the Institute for Health Metrics and Evaluation at the University of Washington in Seattle.”
• Dennis Thompson with HealthDay wrote Where you live may determine how you die, which was picked up by U.S. News and World Report. “Armed with this sort of information, county and city health departments can focus their efforts on the specific problems affecting their communities, said lead researcher Ali Mokdad. He is a professor with the department of global health at the University of Washington, in Seattle.”
• Julia Belluz and Sarah Frostenson wrote These maps show how Americans are dying younger. It’s not just the opioid epidemic. “Different geographic regions are experiencing extreme variations in despair-related outcomes like suicides, drug overdoses, and heart disease, said Abraham Flaxman of the University of Washington, one of the authors of the new JAMA paper. ‘If you look at geographic patterns, you can say it’s despair that’s leading people to drink and do drugs. But then why wouldn’t that apply to leading people to overeat and become obese and diabetic? These trends are happening in different places.’”
• Agata Blaszczak-Boxe with Live Science wrote Leading causes of death in US vary greatly by region. “The reasons why higher death rates vary across geographic areas are not completely clear, but the authors suggested some ideas. For example, the higher death rates from cardiovascular diseases might have something to do with higher rates of obesity in these areas, said study co-author Christopher J. L. Murray.
• Carolyn Gregoire with the Huffington Post wrote This GIF sums up the impact of addiction and mental illness on America. “In a cluster of counties in Kentucky, West Virginia and Ohio, researchers uncovered striking death toll increases of 1,000 percent or more. Topping the list were Clermont County, Ohio (the site of one of the worst heroin epidemics in the state), which saw a 2,206 percent spike, and opioid-stricken Boone County, West Virginia, with a 2,030 percent increase.”
• (UK) Mia De Graff with the Daily Mail wrote What is the typical cause of death in YOUR county? Incredible maps show leading killers in each region of America. “Where you live determines how you die. That is the conclusion of a new study that lays bare the most common causes of death county-by-county across the United States, and how it has changed since 1980.”
• (UK) Celine Gounder with the Guardian wrote How long will you live? That depends on your zip code. “In an analysis of 80 million deaths in the United States between 1980 and 2014, a study published on Tuesday finds dramatic differences not only in life expectancy, but also in cause of death from county to county. ‘We’re not narrowing the gap. The gap is widening,’ said Christopher JL Murray, one of the authors of the study.”

Comments Off on U.S. county mortality paper

Filed under global health

CCA for diverticulitis

A short paper from my work on predicting who will get elective surgery for diverticulitis is on arXiv: https://arxiv.org/abs/1612.00516 [tag: my-research]

Comments Off on CCA for diverticulitis

Filed under machine learning

Learning in Surgeons

New paper: Assessing surgeon behavior change after anastomotic leak in colorectal surgery
Vlad V Simianu, Anirban Basu, Rafael Alfonso-Cristancho, Richard C Thirlby, Abraham D Flaxman, David R Flum
Publication date
Journal of Surgical Research

Comments Off on Learning in Surgeons

Filed under global health, health communication

Diabetes by County in USA

Diagnosed and Undiagnosed Diabetes Prevalence by County in the US, 1999–2012
Laura Dwyer-Lindgren, Johan P Mackenbach, Frank J van Lenthe, Abraham D Flaxman, Ali H Mokdad
Publication date
Diabetes Care


Cool maps:

Comments Off on Diabetes by County in USA

Filed under disease modeling, global health