Thursday, May 21, 2009

Wolfram Alpha: "computational knowledge engine"

Wolfram|Alpha has been in the news/blogosphere lately.

Search Principle Blog
Mr.W's lecture at Berkman Center, Harvard
The NYT
The Register's not-so-glowing review
(for more see 1st item in this list)

I've spent a bit of time fiddling with it to see what it can do and whether it might be useful for either myself, health consumers or the health professionals I work with.

So far I have found one potential use which is to look up disease mortality rates. If you type in lung cancer for example, you will get Canadian & world mortality figures (number of deaths & rate of death, per year). If you type lung cancer Italy, you get the results for that country & the world.

More interesting is the list1 of background sources and references that pops up when you click on source information for the original query. In this case:

I guess potentially this is a very quick way to find sources that offer statistics for a particular disease, though you still have to search the resources to find the specific data. Unfortunately Canadian resources are not included. Also not sure which diseases are in there. Only one way to find out...

...even more interesting:

Type in heart attack and you get a risk calculator (based on Framingham) that lets you input age, gender, cholesterol levels, blood pressure, and risk factors, and which then computes the likelihood that the person whose data is inputted will die from a heart attack in the next ten years.2 You are also shown the impact of cholesterol and blood pressure on risk.

I look forward to testing it out further over time. My biggest concern is how reliable and authoritative the information we are getting is. So far my confidence has not been inspired,1 and I'm not ready to recommend it except as something with potential that's fun to play with. Try asking it how old it is.

Cheers,
FF


1 Here's what W|A has to say about the list:
This list is intended as a guide to sources of further information. The inclusion of an item in this list does not necessarily mean that its content was used as the basis for any specific Wolfram|Alpha result.
Hmm. Interesting way to cite your sources. Not sure this would wash in a term paper. I have a feeling the issue will come up again (it was also mentioned in the question period of Mr.W's lecture at Berkman Center, Harvard.)

2 same result if you input myocardial infarction. No link is suggested between the two terms by W|A.

2 comments:

Dean Giustini - blogger said...

Francesca
It was really good to see you at CHLA/ABSC in Winnipeg. I hope next time you and I can get to talk more about what you are doing.

In the meantime, I will be coming back to read the Info.RX Cases blog.

Dean

FF said...

Yes it would have been nice to talk more at the conference. Am still thinking about how learning theories can be applied in an instructional (one-shot) rather than teaching (building over time) environment. May write about it at some point.

I guess I'll have to make sure my posts are really interesting from now on!

Cheers,
FF