Towards unambiguous representation of patient data

In the early eighties, the goal set for the development of computer-based patient records was the creation of patient records that were analogous to the paper record. In the Netherlands, where the number of physicians using computer-based patient records is steadily increasing, this strategy has been proven successful. Although these "paper-like" computer-based patient records were suitable for patient care, they were much less suited for other purposes. Experiments showed that the use of data for other purposes than those for which they were recorded, can only be performed reliably when these data permit unambiguous interpretation. Due to a physician's limited time there is a constant tension between benefit and effort. Therefore, we should not expect the physician to provide the large amount of additional information, required for unambiguous interpretation of his record. Many of the inferences made by physicians are based on general knowledge and do not require specific, patient related information. We have focused our research on the potential of using knowledge about concepts in the patient record, to infer information, that is implicit in the patient data. The paper discusses considerations with respect to possible strategies to elicit a maximum of information with a minimum of effort from the physician.