Why the standard view is standard: people, not machines, understand patients' problems.

The 'Standard View' regarding computer-based medical diagnostic decision support programs is that, while such systems may be useful adjuncts to human decision-making, they cannot replace human diagnosticians. Mazoué (1990) disputes this viewpoint. He notes that human diagnosis is prone to a variety of errors, and claims that the processes of data collection for diagnosis and the intellectual task of making a diagnosis are independent. Mazoué believes that recent progress in computer-based diagnosis has been encouraging enough to consider the concept of "human-assisted computer diagnosis". This commentary explains why the Standard View should remain standard. Diagnosis is a complex process more involved than producing a nosological label for a set of patient descriptors. Efficient and ethical diagnostic evaluation requires a broad knowledge of people and of disease states. The state of the art in computer-based medical diagnosis does not support the optimistic claim that people can now be replaced by more reliable diagnostic programs.