Investigating cognitive network similarity in breast cancer detection

• Autor(a) principal: Berger, Julian
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.14/37009
• topic: Domínio/Área Científica::Ciências Sociais::Psicologia

Identifying predictors of collective performance in medical decision-making, requiring diagnosticians to independently formulate judgements, is of key importance for effective care. The German Mammography Screening Program poses a prime example of such a situation in which diagnostic decisions are currently being made by at least two individuals. Given this importance, the present study combined insights from research in mental models and cognitive network science to study the cognition and diagnostic performance of experienced radiologists in their ability to correctly identify cancer presence in mammogram images. The study relies on a mixed-methods design, incorporating knowledge gained from interviewing experienced radiologists into a subsequent cross-sectional investigation on cognitive networks of cancer cues relevant in mammogram diagnoses. Cognitive networks were elicited from trained radiologists employed in the national screening program and compared based on their structure, path lengths, degree and clustering. Divergences between radiologists’ networks were revealed through both visual and numerical analyses. However, results of generalized linear mixed modeling indicated divergences and similarities to be almost unequivocally as not being associated with both diagnostic performance and diagnostic similarity in dyadic decision making. The key finding of the present study suggests, however, that more precisely defined clusters within networks, represented via low clustering coefficient, were associated with correct classification of images to diagnostic categories, which warrants future research opportunities. The study concludes with the identification of three limitations present both in this inquiry as well as in prior research and calls for a renewed critical assessment of fundamental assumptions underlying current cognitive network studies.