medium10 pts · first answer countscross-validationdata-leakage

Several scans per patient

A medical dataset has several scans per patient. How should you split it for cross-validation?

A.A random row-level split, since scans are different imagesB.Group k-fold by patient, so no patient is in both train and validationC.Shuffle the labels first, then split randomlyD.Put each patient's most recent scans in training

Discussion 0

Join free to post in the discussion.

Several scans per patient

Discussion 0

Related questions