DELIVERING DISEASE PROGRESSION MODELS TO CLINICAL PRACTICE: A USER STUDY ON OPPORTUNITIES AND BARRIERS FOR TRANSLATION AND ADOPTION

Abstract

Research on biomarkers as early indicators of Alzheimer's Disease (AD) is stimulating the development of novel disease progression models [1]. These models have great potential to define data-driven disease stages and predict patients’ individual progression. However, obstacles arise in adoption by clinicians, suggesting the need to package them in a way that fits clinical routine [2]. For this purpose, the present study aimed to map out current clinical practice to identify key opportunities and barriers in translating such computational innovations into diagnostic/prognostic tools, as envisioned by the EuroPOND project [3]. These can support front-line clinicians in AD management and decision-making. Current clinical procedures for analysis and discussion of AD cases were studied by means of field observations at 6 multidisciplinary team meetings, and one-on-one semi-structured interviews (5 neurologists and 1 psychiatrist), focusing on current diagnostic and prognostic practice. Interviews further explored clinicians’ perceptions of potential impact and applicability of predictive models, using low-fidelity prototypes [Fig.1, a - b] as a prompt to discussion. Data were qualitatively analysed using predefined themes from literature [2]. Analysis of current clinical practice produced a workflow model [Fig. 2] and a set of prototypical interviewee profiles. The workflow helped identify opportunities for facilitating adoption of the tool, the key strategy being to start with early adopters, i.e., clinical professionals working in research-oriented settings, with access to sophisticated data and a focus on early assessment. Barriers might be represented by lack in the tool's use of clinically acknowledged terminology or adaptation to clinical workflow [Tab. 1].