Fig. 7From: Visual transformer and deep CNN prediction of high-risk COVID-19 infected patients using fusion of CT images and clinical dataMean ROC curves from the top performing models. Bold line: 3D Swin Transformer on fusion data (CT + 67 labels), dashed-dotted line: 3D Swin Transformer on CTs only, dotted line: Gradient Boosting, dashed line: Terminal 3D-CNN fusion (CT + 30 labels)Back to article page