An analysis of order correction factors by using a term extraction method for order entry sequences

Electrical medical information systems have been able to communicate rapidly and steadily among medical staffs. Especially, computer physician order entry systems play an important role in hospital information systems. However, there are still remaining order corrections and deletions, caused by both of changes of patients' condition and operational problems between a system and doctors. In this paper, we analyzed factors of medication order corrections for each patient. We firstly constructed datasets consisting of patients' basic information, numbers of orders, numbers of order corrections/deletions, and sequence of order entries. From the whole of the order entry sequences, we extracted characteristic order entry sub-sequences by using two different approaches. One is the approach by using an automatic term extraction method in natural language processing. The other is the approach based on information criteria developed for finding characteristic sub-sequence of genes. Then, we obtained two rule sets based on the extracted order entry sub-sequences. By comparing their contents and accuracies of the if-then rules, we discuss about usefulness of the characteristic order entry sub-sequences for analyzing the order correction factors.