Abstract
In recent years, there has been increased discussion about the potential of expert systems to support medical decision-making tasks, including applications in clinical laboratory settings. This study provides data regarding the cognitive errors that technologists make on an important problem-solving task: the identification of antibodies in a patient’s blood. It explores alternative designs for expert systems developed to reduce such errors. It also evaluates the effects of these alternative designs on the ability of the users to effectively stay “in the loop,” applying their own expertise and judgment while using the computer as a tool to assist with their analyses. A pilot study was conducted involving 32 certified medical technologists, which compared two alternative roles for the computer: (1) use of the computer to automatically complete subtasks upon request, and (2) use of the computer as a monitoring device to critique technologists as they completed the analyses themselves. The system design that automatically completed subtasks for the technologist induced a 29 percent increase in errors relative to the design that critiqued technologists as they completed the analyses themselves.