Transcript The Promise and Peril of Data Mining

The Promise and Peril of Data
Mining
Including a Trip to Iceland
Steve Smela, CSCI 5707
October 8, 2013
Connections to CSCI 5707
• Text Chapter 28 (Weeks 13 & 14)—Data
Mining
Iceland’s Flag
Iceland’s Coat of Arms
Iceland’s King
Map of Iceland
What Does Iceland Have?
Icelandic Sheep
Glaciers
Volcanoes and Stuff
But Most Importantly….
Icelanders!
(324,000 of them)
Ingolfur Arnarson, 1st Permanent
Norse Settler in Iceland
One of Ingolfur’s Descendants
Used “Cognitive Performance Scale” to
assess performance of those with
mutation vs. those without mutation
Only possible because ratings
are done frequently in nursing
homes in Iceland
Cognition measured by CPS as a function of age.
T Jonsson et al. Nature 000, 1-4 (2012) doi:10.1038/nature11283
The Promise of Data Mining
• Possible insights into Alzheimer’s Disease,
leading to new treatments
The Perils of Data Mining-I
• Correlation does not equal causality
• Just because A673T is associated with better
scores on the Cognitive Performance Scale, it
doesn’t mean that the mutation protects
against Alzheimer’s
• Independent confirmation needed
Ways to Verify Insights Gained from
Data Mining
• In “Big Data” context, it’s common practice to
split data into at least 2 (sometimes more)
sets
– Training set
– Validation set
• In Icelandic study, what did they do?
Independent Testing of Hypothesis
Perils of Data Mining-II
What Does this Mean?
Take-Home Messages
• Data mining holds great potential, especially
in ideal conditions like Iceland
• Associations found through data mining must
be independently verified
• Data mining must take into account ethical
issues such as the informed consent of the
research subjects
Sources
•
•
•
•
•
•
•
http://news.sciencemag.org/people-events/2012/12/purchase-amgen-wont-affect-decode-genetics-research-founder-says
http://en.wikipedia.org/wiki/Iceland
Jocelyn Kaiser. 2013. “Agency Nixes deCODE's New Data-Mining Plan.” Science 340, 1388-1389 (21 June 2013).
http://www.decode.com/ and http://www.decode.com/news-events/
http://www.interrai.org/assets/files/Scales/Cognitive%20Performance%20Scale.pdf
Thorlakur Jonsson, Jasvinder K. Atwal, Stacy Steinberg, Jon Snaedal, Palmi V. Jonsson, et al. 2012. “A Mutation in APP Protects against Alzheimer’s
Disease and Age-Related Cognitive Decline.” Nature 488, 96–99 (02 August 2012).
Jeff Gulcher, Agnar Helgason, and Kári Stefánsson. “Genetic Homogeneity of Icelanders.” Nature Genetics 26, 395 (2000).
Questions?