Let’s talk about the Global Burden of Disease study. [The GBD is published on a nearly annual basis by the IHME.] Because this study is an independent and now—with the additional funding your foundation is providing—a regularly updated assessment of global health, it could in principle serve as a gold-standard reference of progress in various parts of the world on various diseases. Have you actually used it in that way to identify programs that are working and those that are not working and need redirection?
The GBD assembles data from lots of different field studies, many of which we are funding. For example, we ran a big study called GEMS[Global Enteric Multicenter Study] to try to figure out all the different causes of diarrheal disease—rotavirus is the biggest cause, but there’s alsoE. coli, shigella, cryptosporidium and others—and how important each one is. We still struggle with large uncertainty about the locations and extent of certain diseases, such as typhoid and cholera, which no country wants to admit they still have. My teams, like others who are very active in fieldwork, usually are looking at the primary papers as soon as they come out in the scientific literature. By the time the information gets aggregated and vetted and incorporated into the GBD database, it should no longer be surprising to us.
The GBD assembles data from lots of different field studies, many of which we are funding. For example, we ran a big study called GEMS[Global Enteric Multicenter Study] to try to figure out all the different causes of diarrheal disease—rotavirus is the biggest cause, but there’s alsoE. coli, shigella, cryptosporidium and others—and how important each one is. We still struggle with large uncertainty about the locations and extent of certain diseases, such as typhoid and cholera, which no country wants to admit they still have. My teams, like others who are very active in fieldwork, usually are looking at the primary papers as soon as they come out in the scientific literature. By the time the information gets aggregated and vetted and incorporated into the GBD database, it should no longer be surprising to us.
But GBD is super helpful when we’re talking to developing countries and saying “Look, here’s what is going on with tuberculosis in your country versus others like yours.” It’s a very important tool to educate people—like the World Development Report was for me. You can see the time progressions and zoom in on any country. It’s one of the better data-visualization sites in the entire Web. It’s super nice. And most people aren’t that up to date on these disease trends—particularly for infectious diseases. So I’ve been taking GBD charts with me when I’ve met with people in Cambodia or Indonesia or even at the French aid agency about trends in francophone Africa. They can reveal when we haven’t set the right priorities—so it’s a very important tool for me. Before I go into strategy meetings, I sometimes look at the GBD to remind myself of the numbers.
We now have enough detailed data to break big illness categories like diarrheal disease apart into separate diseases by the root cause. Even so, the error bars tend to be quite large on these estimates because, unlike in the rich world where disease cases are actually counted and tracked, in the poorer parts of the world we have to rely on sampling and extrapolation. If you happen to sample in places where the condition is unusually prevalent, the extrapolated numbers can be wrong.
That raises an interesting point. One of the potential advantages of having this statistical inference machine that IHME uses to produce the GBD estimates is that you could identify where you would get the most bang for the buck if you did a new study that will improve the empirical input to the system. Has the GBD actually been used to prioritize funding of surveillance in this way?
Oh, yeah. Disease surveillance in the poor world is terrible. While it’s great that we now have this published set of numbers, they have pretty big error bars—and probably some of the error bars should be even bigger than they are shown in the IHME reports. But we’re actively looking at ways to improve the situation. New diagnostics are becoming available, for example, that can check for lots of different diseases by analyzing just a few drops of blood. So rather than running one study after another, each of which has to set up a bunch of different centers just to get data on one kind of disease, we might be able to use clinics that are running all the time and constantly monitoring the prevalence of lots of diseases simultaneously.
Oh, yeah. Disease surveillance in the poor world is terrible. While it’s great that we now have this published set of numbers, they have pretty big error bars—and probably some of the error bars should be even bigger than they are shown in the IHME reports. But we’re actively looking at ways to improve the situation. New diagnostics are becoming available, for example, that can check for lots of different diseases by analyzing just a few drops of blood. So rather than running one study after another, each of which has to set up a bunch of different centers just to get data on one kind of disease, we might be able to use clinics that are running all the time and constantly monitoring the prevalence of lots of diseases simultaneously.
From Wikipedia, the free encyclopedia
Error bars are a graphical representation of the variability of data and are used on graphs to indicate the error, or uncertainty in a reported measurement. They give a general idea of how precise a measurement is, or conversely, how far from the reported value the true (error free) value might be. Error bars often represent one standard deviation of uncertainty, one standard error, or a certain confidence interval (e.g., a 95% interval). These quantities are not the same and so the measure selected should be stated explicitly in the graph or supporting text.
Error bars can be used to compare visually two quantities if various other conditions hold. This can determine whether differences are statistically significant. Error bars can also suggestgoodness of fit of a given function, i.e., how well the function describes the data. Scientific papers in the experimental sciences are expected to include error bars on all graphs, though the practice differs somewhat between sciences, and each journal will have its own house style. It has also been shown that error bars can be used as a direct manipulation interface for controlling probabilistic algorithms for approximate computation.[1] Error bars can also be expressed in a plus-minus sign (±), plus the upper limit of the error and minus the lower limit of the error.[2]
*****
Bill Gates has a well-established knack for sifting through complex data sets to find the right pathways for making progress around the globe in health, education and economic development.
沒有留言:
張貼留言