Double reading reduces miss errors in low prevalence search.

Area of Science:

• Cognitive psychology
• Medical imaging analysis
• Human factors in diagnostics

Background:

• Rarely occurring targets are frequently missed in visual search tasks.
• This is a critical issue in medical screening, such as mammography, for detecting infrequent cancers.
• Current single-reader methods may not be optimal for low-prevalence diagnostic tasks.

Purpose of the Study:

• To investigate if a "double reading" procedure, involving paired search, reduces miss errors compared to single-person search.
• To determine the impact of shared versus independent reading environments on the effectiveness of double reading.
• To explore how reading order influences miss errors in paired search scenarios.

Main Methods:

• Experiment 1: Participants searched for masses in laboratory mammograms, either together in the same room or independently in separate rooms.
• Experiment 2: Manipulated the order in which paired participants reviewed mammograms (same or different orders).
• Compared miss error rates between single-reading and double-reading conditions across different environmental setups.

Main Results:

• Double reading significantly reduced miss errors compared to single reading.
• No significant effect of reading order on miss errors was observed.
• The mechanism of improvement in double reading varied by environment: increased sensitivity in shared rooms, altered response bias in separate rooms.

Conclusions:

• Paired "double reading" is an effective strategy to reduce miss errors in low-prevalence visual search tasks, like mammography.
• The benefits of double reading are influenced by the reading environment, affecting sensitivity and response bias differently.
• This has direct implications for improving diagnostic accuracy in medical screening programs.