8 Lektionen zur Implementierung auf künstlicher Intelligenz basierender Screeningmethoden in die Primärversorgung am Beispiel der diabetischen Retinopathie

Background: The prevalence of diabetes mellitus is increasing worldwide, leading to an increase in diabetes-related complications, including diabetic retinopathy (DR). Despite the availability of treatment options with early diagnosis, early detection of DR remains below the desirable participation rates. Against this challenge, artificial intelligence (AI)-assisted screening methods in primary care are becoming increasingly important as they offer the potential to improve early detection rates and optimize access to specialized care. Aim: This article examines how AI-supported screening methods for DR can be effectively implemented in primary care. Challenges, opportunities, and practical findings from current studies are highlighted. Conclusion: Despite the high sensitivity and specificity of AI-supported screenings, it is necessary to take sector-specific prevalence into account. Financial aspects, user-friendly technical requirements, and the need for interdisciplinary collaboration are determining factors for successful implementation. Despite the challenges, such as concerns about overlooking other eye diseases and lack of confidence in the quality of AI-based diagnoses, the attitude of general practitioners is positive. The article emphasizes the importance of thorough validation of AI systems under real-life conditions and adaptation to the specific needs of primary care.