Diagnosis by the bit: a method for evaluating the diagnostic process

The extent to which a diagnostic observation or test contributes to a diagnosis is usually represented by a conditional, or Bayesian, probability. According to information theory, the contribution of a diagnostic observation can also be measured in bits of information. This offers a representation of the Bayes rule which may be useful in comparing and evaluating diagnostic sequences, since the diagnostic contributions of the various tests, expressed as bits of information, are additive, providing a simple and graphic representation of the diagnostic process.