Consequences and mitigation of cognitive bias in the radiological interpretation of breast cancer imaging using Artificial Intelligence (Preprint)

Abstract

<sec> <title>UNSTRUCTURED</title> The interpretation of radiological images is a critical component in the diagnostic pathway of breast cancer, directly impacting therapeutic decisions and patient outcomes. However, system-related, perceptual, and cognitive factors may lead to the detection of diagnostic errors, which undermine the accuracy. This paper provides a comprehensive overview of the sources and implications of diagnostic inaccuracies in breast imaging, focusing on the growing role of artificial intelligence (AI) as both a supportive and potentially bias-reducing tool. Recent prospective studies demonstrate the clinical safety and efficiency of AI-assisted mammography screening, showing improved cancer detection rates and reduced workload. Nonetheless, integrating AI without an appropriate knowledge of the consequences might introduce new cognitive biases, such as anchoring, automation, and confirmation bias, that may influence radiologists' decision-making and counteract the intended benefits. The paper also explores strategies to mitigate these errors, including targeted training, improved interdisciplinary communication, and reflective practice. A balanced and cautious integration of AI, supported by continuous education and robust validation studies, seems essential to enhance diagnostic accuracy while preserving radiologists’ critical reasoning. </sec>

Ähnliche Arbeiten