Improving Clinical Trial Participant Tracking Tools Using Knowledge-anchored Design Methodologies

Abstract

OBJECTIVE: Rigorous human-computer interaction (HCI) design methodologies have not traditionally been applied to the development of clinical trial participant tracking (CTPT) tools. Given the frequent us of iconic HCI models in CTPTs, and prior evidence of usability problems associated with the use of ambiguous icons in complex interfaces, such approaches may be problematic. Presentation Discovery (PD), a knowledge-anchored HCI design method, has been previously demonstrated to improve the design of iconic HCI models. In this study, we compare the usability of a CTPT HCI model designed using PD and an intuitively designed CTPT HCI model. METHODS: An iconic CPTP HCI model was created using PD. The PD-generated and an existing iconic CTPT HCI model were subjected to usability testing, with an emphasis on task accuracy and completion times. Study participants also completed a qualitative survey instrument to evaluate subjective satisfaction with the two models. RESULTS: CTPT end-users reliably and reproducibly agreed on the visual manifestation and semantics of prototype graphics generated using PD. The performance of the PD-generated iconic HCI model was equivalent to an existing HCI model for tasks at multiple levels of complexity, and in some cases superior. This difference was particularly notable when tasks required an understanding of the semantic meanings of multiple icons. CONCLUSION: The use of PD to design an iconic CTPT HCI model generated beneficial results and improved end-user subjective satisfaction, while reducing task completion time. Such results are desirable in information and time intensive domains, such as clinical trials management.

Fig. 1

Generic layout of a clinical trial protocol schema, composed of atomic temporal constraints. Event instances are shown as Time Point (T) – Event (E), using the notation: T_xE_y, where x is the Time Point descriptor, and y is the Event descriptor. In some instances, a transposed version of this grid is used.

Fig. 2

Overview of study methods, including input and output research products.

Fig. 3

Annotated heat-map visualization of “consensus clusters” generated during phase three. Tightly clustered sections of the heat-map (as indicated by increased density) correspond to “consensus clusters” that have a strong correlation between the visual similarity and semantic meaning of a group of prototype icons, and can therefore serve as the basis for informing the design of icons intended to serve as HCI-model metaphors such concepts.

Fig. 4

Illustrative example of prototype CTPT HCI models. Notable features include: A) the ability to filter the end-user view of records by visit, protocol or participant; B) a tabbed interface model for navigating between visit days as defined by the study protocol; C) optional textual labels to accompany icons; D) result-set rows corresponding to a specific trial participant; E) prototype icons designed using PD which indicated from left-to-right: medical history, physical exam, case report form, lymph node measurement, bone marrow biopsy, radiology, blood specimen collection and laboratory procedures; and F) conventional icons adapted from commonly available CTPT software.