Phenotyping Diagnosis: Identification of Diagnostic Paths

We would like to ask the question, how much can we understand of a physician's diagnostic reasoning process from what is in the electronic record? However, it is well known that a good deal of important information about the diagnostic process is not recorded; it would be impractical to record every detail unless it somehow has a bearing on a treatment decision. What remains is the formal record of "chief complaint", vitals and physical examination, medications prescribed, referrals made, orders written—for tests, imaging, etc.—and any procedures performed. The sequencing of these events (which we have defined as the diagnostic path) and decisions made on that basis, vary considerably from patient to patient and from physician to physician. Equally, it is likely that some diagnostic paths are more efficacious than others. We therefore (a) seek to identify paths; (b) group them by "similarity"; (c) consider which archetypal paths, in different circumstances, prove more effective, or more cost-effective. We report on the extensive methodological effort undertaken at this stage of the project.