Skip to main content
. 2019 Nov 1;7(11):e14452. doi: 10.2196/14452

Figure 6.

Figure 6

General scheme of the proposed transfer learning strategy. For patients with fewer observations, we pretrain a model with observations sampled from all the patients in the dataset (based on either no weighting or weighting strategy). The pretrained model is then fine-tuned with the data of the patient of interest.