[COPD-MICCAI 2018] Perform statistical methods to replicate results in Raul's paper

[sumedhasingla]

• Perform Hosmer-Lemeshow goodness of fit test for logistic regression to validate ARD eventa i.e. exacerbation frequency

• Report p-value for the Hosmer-Lemeshow test

• Perform survival analysis for mortality prediction. Specifically Cox regression.

• Report c-indice for Cox regression and p-value for Greenwood-Nam-D’Agnostino test.

[sumedhasingla]

Points about Rauls's paper

• They are not doing cross validation.

• They report their results of ARD event on n = 1000 subjects (test data).

[jrahimik]

To make the comparison of our results with their papers:
Here is the n and g relationships.

• for N<1000 we use g=10 which is the case in Rauls paper

• for N=10000 subjects our g will be 150 which is the case in our paper.
  This makes the calibration comparable.