Thomas Dratsch

Who they are

His affiliation on the 2023 paper is the Institute of Diagnostic and Interventional Radiology, University of Cologne, Faculty of Medicine and University Hospital Cologne, Germany. The paper was written with colleagues from Cologne, Amsterdam, Kiel, Mainz and Würzburg.

This page records only what the paper and its published record state. Biography beyond the cited work is not given here.

Key papers

Automation bias in mammography: the impact of artificial intelligence BI-RADS suggestions on reader performance (2023), with Chen, Rezazade Mehrizi, Kloeckner, Mähringer-Kunz, Püsken, Baeßler, Sauer, Maintz and Pinto dos Santos. In a prospective experiment, 27 radiologists read 50 mammograms and gave each a BI-RADS category, the standard scale for how suspicious a breast image looks, with help from what they were told was an AI system. After 10 training cases in which the suggestions were right, they read 40 cases in which the suggestion was wrong for 12. Readers at every level of experience did markedly worse when the suggestion was wrong. Very experienced radiologists rated 82.3 per cent of mammograms correctly when the suggestion was right and 45.5 per cent when it was wrong. For inexperienced readers the figures were 79.7 and 19.8 per cent. Inexperienced readers were more likely than experienced ones to follow an incorrect suggestion that a finding was more serious than it was.

Key ideas

The study gives a clean measurement of automation bias, the tendency to favour a suggestion from an automated system, in a clinical task. It matters for any plan that relies on a doctor checking an AI's output, because the check is only as good as the doctor's willingness to overrule the machine. The authors concluded that effects of human and machine interaction must be considered to ensure safe deployment when human readers and AI are combined.

---