Nt1310 Unit 5 Assignment 1

Decent Essays

The knowledge base for DXplain contains more than 65,000 relationships between diseases and patient descriptors (or clinical manifestations). The format for these relationships and the data structures are similar to the format and data structures used in the INTERNEST/QMR system except that in DXplain terms cannot only support a given diagnosis, but can also contradict other diagnoses. The relationships in DXplain are concerned with three different elements: 1) term importance, 2) term frequency and 3) term-evoking power. Term importance is used to express how significant the particular term is in indicating the presence of disease. A high term importance is given to findings that can be identified with high reliability or are rarely found …show more content…

The diseases are presented to the user in two lists: “common diseases” and “rare disease”; in addition, a serious disease (one that may require relatively immediate action) is so indicated on the list by an asterisk. The user can request DXplain to explain why any specific diagnosis was included. DXplain will present the clinical findings entered by the user that support the selection of that disease, the clinical findings that would not be expected in that disease, and additional clinical findings that would be expected if that disease was present. In this way, DXplain assists the user in understanding the logic used by the program and facilitates pattern matching by the user in comparing DXplain’s disease description with his/her knowledge about the patient. The intent is to present sufficient information so that the user can always use his/her own clinical judgment as to the appropriateness of DXplain’s …show more content…

This mode is particularly useful in helping the user select the clinical manifestations that are important without forcing them to enter a large number of less relevant findings. At any time in this mode, the user can interrupt DXplain to ask “Why?” i.e. to ask DXplain to justify why this particular clinical manifestation is important. DXplain will respond by displaying the name of the disease that is being considered at that point in the interaction and the reason the particular finding might be important in confirming the presence of that disease. DXplain’s ability to explain and justify are key elements of the system. It is critical that this system not be perceived as a magic black box that can somehow provide the “answer” to a complex diagnostic problem. We believe that physicians will not accept DXplain as a useful diagnostic assistant unless the clinical interpretations seem reasonable and unless the system can offer explanations that are understandable and persuasive. (Barnett et al.,

Get Access