GIFT: A tool for generating free text reports from encoded data

Abstract

The benefits of the electronic patient record versus the traditional paper-based clinical chart have been widely illustrated in the medical informatics literature. However, hospital information systems are often used only to administrative purposes (admission/discharge) and most of the clinical information is still paper-based. There is a resistance of healthcare operators to shift from paper to computer, and there are many reasons for this behaviour, that will be discussed in the paper. One of these reasons, that is also the focus of our work, is that while formulating the data model, physicians are often fostered (by computer scientists) to reduce free text data. In other words, computerisation of the medical record is often offered as an opportunity to “encoding” information as much as possible, with the promise that this will facilitate further statistics and data sharing. However, in some cases physicians perceive this encoding as a constraint and a limitation. In this paper we discuss this issue and we illustrate a solution devised for a Stroke Unit.