This site is not optimized for Internet Explorer 8 (or older).
Please upgrade to a newer version of Internet Explorer or use an alternate browser such as Chrome or Firefox.
Variable selection methods for multiple regressions influence the parsimony of risk prediction models for cardiac surgery
A surprisingly readable Australasian manuscript on risk prediction models for cardiac surgery. Interesting that the class of dyspnoea was not captured in 5.28% of the operated patients.