Every month I try to read an open-access article. Then, I share the title and associated link with my followers. This is to encourage clinicians to read articles, stay up-to-date, and continue to grow.
The objective of this month’s paper is to ”compare the preferential and differential diagnostic accuracy between the pulmonologist’s view (control) and the pulmonologist’s view assisted with suggestions provided by a machine-learning model (intervention)” p2.
Collaboration between explainable artificial intelligence and pulmonologists improves the accuracy of pulmonary function test interpretation
By Nilakash Das , Sofie Happaerts, Iwein Gyselinck, Michael Staes, Eric Derom , Guy Brusselle , Felip Burgos , Marco Contoli, Anh Tuan Dinh-Xuan , Frits M.E. Franssen , Sherif Gonem, Neil Greening, Christel Haenebalcke, William D-C. Man, Jorge Moisés, Rudi Peché, Vitalii Poberezhets, Jennifer K. Quint, Michael C. Steiner, Eef Vanderhelst, Mustafa Abdo, Marko Topalovic and Wim Janssens.
European Respiratory Journal 2023; DOI: 10.1183/13993003.01720-2022
Link to the article: https://erj.ersjournals.com/content/early/2023/03/15/13993003.01720-2022
“ In this study conducted in two separate phases, we observed that pulmonologists when aided by XAI significantly improved on their individual preferential and differential diagnostic accuracy in interpreting PFTs. Among secondary end-points, we noted a significant increase in diagnostic confidence but no reduction in the number of differential diagnostic choices. Our results support the hypothesis that a pulmonologist aided by XAI improves on the interpretation of PFTs for the differential diagnosis of respiratory diseases when compared with individual pulmonologists with no support. Interestingly, we also observed that pulmonologists when aided by XAI significantly outperformed XAI itself in preferential and differential diagnostic accuracy. “Page 8
As always, take the time to study the article and learn about all the results and findings.
Happy reading and learning,
June 10th, 2023