Additional file 1 of Differences in technical and clinical perspectives on AI validation in cancer imaging: mind the gap!

Additional file 1: Table C 1. The correspondence of ‘Roles’ Combinations to Clinical and Technical group. Table C 2. Age-working place combinations of the whole group. Figure C 1. Responses regarding user roles (model training, model validation (technical), model validation (clinical), clinical expert, data provider and data curation). Multiple responses were possible. Figure C 2. Stacked bar plots on some ‘Technical validation’ topics comparing technical and clinical profiles points of view. Figure C 3. Stacked bar plots on some ‘Bias and fairness’ topics comparing technical and clinical profiles points of view. Figure D 2. Bar plot about the use of permutation-based techniques to assess model stability by TG respondents. The x axis indicates the proportion of responses. Figure D 3. Bar plot about the use of nested cross-validation by TG respondents. The x axis indicates the proportion of responses. Figure D 4. Bar plot about the strategy of reporting regression models adopted by TG re-spondents. The x axis indicates the proportion of responses. Figure D 5. Dataset Splits. Figure D 6. Use of Permutation Techniques. Figure D 7. Segmentation evaluation metrics. Figure D 8. Regression evaluation metrics. Figure D 9 Reporting on regression models. Figure E 1. Stacked bar plots on the ‘Overall strategy for model building and validation’ top-ics where clinicians would benefit from further exposure. Figure E 2. Stacked bar plots on the ‘Technical Validation’ topics where clinicians would ben-efit from further exposure. Figure E 3. Stacked bar plots on the ‘Technical Validation’ topics where clinicians are some-how limited. Figure E 4. Bar plots on the multiple-choice question regarding permutation-based tech-niques. Figure E 5. Stacked bar plots on the ‘Statistical analysis and evaluation metrics’ topics where clinicians would benefit from further exposure. Figure E 6. Stacked bar plots on the ‘Bias and Fairness’ topics where clinicians are some-how limited. Figure E 7. Word cloud from common terms emerged from open questions on AI validation strategies. Table E 1 List of themes that emerged from the data analysis and concepts per theme. Table E 2 List of themes that emerged from the data analysis and concepts per theme. In the verbatim quotes, ‘P’ stands for participant