Background and aims: The accurate differentiation between T1a and T1b Barrett’s cancer has both therapeutic and prognostic implications but is challenging even for experienced physicians. We trained an Artificial Intelligence (AI) system on the basis of deep artificial neural networks (deep learning) to differentiate between T1a and T1b Barrett’s cancer white-light images. Methods: Endoscopic images from three tertiary care centres in Germany were collected retrospectively. A deep learning system was trained and tested using the principles of cross-validation. A total of 230 white-light endoscopic images (108 T1a and 122 T1b) was evaluated with the AI-system. For comparison, the images were also classified by experts specialized in endoscopic diagnosis and treatment of Barrett’s cancer. Results: The sensitivity, specificity, F1 and accuracy of the AI-system in the differentiation between T1a and T1b cancer lesions was 0.77, 0.64, 0.73 and 0.71, respectively. There was no statistically significant difference between the performance of the AI-system and that of human experts with sensitivity, specificity, F1 and accuracy of 0.63, 0.78, 0.67 and 0.70 respectively. Conclusion: This pilot study demonstrates the first multicenter application of an AI-based system in the prediction of submucosal invasion in endoscopic images of Barrett’s cancer. AI scored equal to international experts in the field, but more work is necessary to improve the system and apply it to video sequences and in a real-life setting. Nevertheless, the correct prediction of submucosal invasion in Barret´s cancer remains challenging for both experts and AI.

The endoscopic features associated with eosinophilic esophagitis (EoE) may be missed during routine endoscopy. We aimed to develop and evaluate an Artificial Intelligence (AI) algorithm for detecting and quantifying the endoscopic features of EoE in white light images, supplemented by the EoE Endoscopic Reference Score (EREFS). An AI algorithm (AI-EoE) was constructed and trained to differentiate between EoE and normal esophagus using endoscopic white light images extracted from the database of the University Hospital Augsburg. In addition to binary classification, a second algorithm was trained with specific auxiliary branches for each EREFS feature (AI-EoE-EREFS). The AI algorithms were evaluated on an external data set from the University of North Carolina, Chapel Hill (UNC), and compared with the performance of human endoscopists with varying levels of experience. The overall sensitivity, specificity, and accuracy of AI-EoE were 0.93 for all measures, while the AUC was 0.986. With additional auxiliary branches for the EREFS categories, the AI algorithm (AI-EoE-EREFS) performance improved to 0.96, 0.94, 0.95, and 0.992 for sensitivity, specificity, accuracy, and AUC, respectively. AI-EoE and AI-EoE-EREFS performed significantly better than endoscopy beginners and senior fellows on the same set of images. An AI algorithm can be trained to detect and quantify endoscopic features of EoE with excellent performance scores. The addition of the EREFS criteria improved the performance of the AI algorithm, which performed significantly better than endoscopists with a lower or medium experience level.

No abstract

Comprehensive knowledge, specific skills, and data-analysis competences are prerequisites of successful use of continuous glucose monitoring systems (CGM). SPECTRUM is a structured and manufacturer-independent training-program for real time CGM (rtCGM) comprising one web-based introduction and six modules (each 90 minutes) of face-to-face group sessions. SPECTRUM was evaluated longitudinally among adults with type 1 diabetes from 10 diabetes centers. Outcome parameters were rtCGM-knowledge and -skills (rtCGM-Profi-Check), satisfaction with the course, technology acceptance and metabolic control. Initially 120 participants with type 1 diabetes were included (mean age 42.4±13.4 years, diabetes duration 21.6±11.6 years, 56% female, mean HbA1c 7.70±1.34%). Data were collected at study entry, after the final group session, and at 6 months follow-up. The study was completed by 108 patients (10% dropped out, mainly due to scheduling problems). After training rtCGM knowledge (scale 0-40) improved by 43% (from 21.2±7.6 to 30.4±4.5; p<0.001) and persisted until 6 months follow-up (29.6±4.6). After 6 months HbA1c decreased to 7.58±1.32% (p=0.036). On a scale from 0-14, practical skills were 13.1±1.3 after the program. Satisfaction with SPECTRUM was 1.4±0.5 (1 is good - 6 is bad). Satisfaction with the rtCGM system was 4.2±0.5 (scale from 1 (low) to 5 (high)) and acceptance of the rtCGM system was 87.6±8.2 (scale from 14 (low) to 98 (high)) after the training and 87.0±9.7 at follow up. This indicates a high acceptance, positive attitude, and intension to use rtCGM continuously. SPECTRUM was shown to be effective in increasing the knowledge and skills about rtCGM in adults with type 1 diabetes. The effect was sustainable and independent from diabetes center and rtCGM-system used. Training participants showed an improvement in glycemic control and improved satisfaction and acceptance of rtCGM. Disclosure G. Freckmann: Advisory Panel; Self; Abbott, Sensile Medical AG. Consultant; Self; Beurer, iSens, Metronom Health, Pharmasens, Profusa, Inc., Roche Diabetes Care. Speaker’s Bureau; Self; Novo Nordisk A/S. S. Schlueter: Advisory Panel; Self; A. Menarini Diagnostics, Abbott. Consultant; Self; Roche Diabetes Care. Speaker’s Bureau; Self; Ascensia Diabetes Care, AstraZeneca, Berlin-Chemie AG, Dexcom, Inc., Medtronic, Merck Sharp & Dohme Corp., Novo Nordisk A/S, Roche Diabetes Care, Sanofi-Aventis Deutschland GmbH, Ypsomed AG. Other Relationship; Self; Springer Publishing Company, Verlag Kirchheim & Co. GmbH. P. Wintergerst: None. L. Heinemann: Consultant; Self; Becton, Dickinson and Company, LifeCare, Inc., Roche Diabetes Care. Stock/Shareholder; Self; Profil Institute for Clinical Research. T. Rückert: None. A. Hinz: Research Support; Self; Institut for Diabetestechnologie Ulm, Novo Nordisk A/S. Speaker’s Bureau; Self; AstraZeneca, Berlin-Chemie AG, Eli Lilly and Company. M. Wernsing: Advisory Panel; Self; Abbott. Research Support; Self; Roche Diabetes Care. F. Thienel: Research Support; Self; Roche Diabetes Care. Speaker’s Bureau; Self; Abbott. T. Biester: Advisory Panel; Self; AstraZeneca. Speaker’s Bureau; Self; Dexcom, Inc., Medtronic, Roche Diabetes Care, Sanofi, Ypsomed AG. P. von Blanckenburg: None. W. Keuthage: Advisory Panel; Self; Deutschland GmbH, Novo Nordisk A/S, Roche Diabetes Care. Board Member; Self; Abbott, Berlin-Chemie AG. R. Ziegler: Advisory Panel; Self; Abbott, Lilly Diabetes, Novo Nordisk A/S, Roche Diabetes Care. Consultant; Self; Novo Nordisk A/S, Roche Diabetes Care. Speaker’s Bureau; Self; Abbott, Dexcom, Inc., Novo Nordisk A/S, Roche Diabetes Care. E. Martin: None. M. Holder: None. D. Deiss: Advisory Panel; Self; Abbott, Roche Diabetes Care. Consultant; Self; Senseonics. G. Buchal: None. L. van den Boom: Advisory Panel; Self; Medtronic. K. Lange: Consultant; Self; Abbott, Dexcom, Inc., Medtronic. Funding Dexcom, Inc.; Roche Diabetes Care; Sanofi; Menarini Group; Kirchheim Verlag; Medtronic

No abstract