The number of five is a trade-off between the efficiency and effectiveness of text prediction. Each list offered five possible choices of predictions (as shown in ▶ Figure 1, part C). & Y.S.) as did similar studies and added to narrative fields for four of the 13 MCQs. In the study, the text prediction lists were manually prepared case by case by domain experts (X.L.
Clicking the bottom button (D) would slide one page of new question(s) in, which helps identify transitions between questions in capturing time on questions in the study. The prediction list (C) is activated as the associated commentary field (B) is checked. The child question appears only when the corresponding item in its parent question is checked (A). The layout of the structured data entry interface with a text prediction list. The study can be instructive to the researchers who work on the optimization of prediction accuracy and interface representation to ascertain their efforts associated with text prediction were worthwhile. In this study, we developed a stand-alone prototype with the task to evaluate the impact of text prediction on the task-related structured data entries. Īccording to a preliminary study, the task of using multiple-choice questions (MCQs) to collect details of process-oriented events was the most time consuming and error-prone step in the course of safety event reporting, due to a great number of cognitive problems such as language ambiguities, mental model mismatches, etc. To further deepen the design at the functional level, text prediction functions on both structured and unstructured data entries were proposed to bridge the information gaps induced by work domain complexity and user disparity. Heuristic evaluation, cognitive task analysis and think-aloud user testing were conducted sequentially to address interface representational issues. The design aimed to increase the efficiency and the data quality by using text prediction in the system. Such systems have shown the problems of underreporting and low quality of reports for a decade although patient safety organizations at local and national levels have advocated the systems for years. This study was grounded in a user-centered design for the development of a patient safety event reporting system. In this study, a two-group randomized design was employed to examine the impact of text prediction on data entry quality and efficiency in a clinical setting, and to determine the effects of text prediction on clinician’s overall performance in structured data entry. Second, despite text prediction having proven effective in reducing the motor requirement for text generation, whether this alone translates into an increased efficiency remains unclear. First, there is a scarcity of research regarding the impact of text prediction on the quality of data entry that clinicians value. However, the text prediction technique has two concerns when being applied in healthcare. The advance of natural language processing techniques has brought text prediction into a broad scope of daily computing activities, such as mobile computing and radiography reports. Text prediction, also known as word, sentence or context prediction originated in augmentative and alternative communication (AAC) to increase text generation rates for people with the disabilities of motor or speech impairment. A solution proposed and examined in this study is the use of text prediction in the build-in narrative fields. This remedy often comes along with the increase of physical and mental loads for text completion and may create a challenge for optimizing the overall performance of structured data entry.
Consequently, a built-in field for narrative comments given as the last option of the predefined list becomes a common remedy. As a process of selecting options from a predefined list, however, structured data entry is restrictive and inflexible compared to clinical report narratives, with respect to ambiguity tolerance and argument making. This is the rationale behind the initiative of a structured data capture project for the meaningful use of Electronic Health Records (EHR) and the continued effort to develop and refine the standardized structured forms for patient safety event reporting. Structured data entry plays an indispensable role because of its merits of interoperability and reuse for research purpose. Many attempts have been made to investigate the difficulties with data entry in order to promote the acceptance and quality-in-use of clinical information systems.