Comparison of dynamic updating strategies for clinical prediction models.

Model updating leads to improved BS, with more frequent updating performing better than less frequent updating. Model recalibration strategies appeared to be the least sensitive to the update interval and sliding window length.