Skip to main content
. 2020 Jun 23;296(3):584–593. doi: 10.1148/radiol.2020192091

Figure 2:

Receiver operating characteristic curves for image-based binary outcome prediction models. Outcome is defined as undergoing a total knee replacement (TKR) within 9 years or not. Area under the receiver operating characteristic curve (AUC) values are for the outcome of predicting TKR. Kellgren-Lawrence (KL) model provides a baseline model for assessing the significant differences between prediction models by using Delong test (24) with Holm correction (25) for multiple comparisons. DL = deep learning, MT = multitask, OARSI = Osteoarthritis Research Society International, TL = transfer learning.

Receiver operating characteristic curves for image-based binary outcome prediction models. Outcome is defined as undergoing a total knee replacement (TKR) within 9 years or not. Area under the receiver operating characteristic curve (AUC) values are for the outcome of predicting TKR. Kellgren-Lawrence (KL) model provides a baseline model for assessing the significant differences between prediction models by using Delong test (24) with Holm correction (25) for multiple comparisons. DL = deep learning, MT = multitask, OARSI = Osteoarthritis Research Society International, TL = transfer learning.