JMIR Publications

ABSTRACT

Background:

Background:

Voice user interfaces are becoming more prevalent in healthcare and are commonly being used for patient engagement. There is a growing interest to identify the potential this form of interface has on patient engagement with digital therapeutics in chronic disease management. Digital therapeutics are most often deployed as smartphone applications however, there is evidence to suggest that specific demographics struggle when using touch screens, and often cannot complete tasks independently. Making digital therapeutics accessible through an alternative interaction model has the potential to better meet the needs of some patients.

Objective:

This research aimed to evaluate how heart failure participants interacted with a voice app version a digital therapeutic, Medly, through a proof-of-concept implementation study design. The objective was to understand whether the voice app would enable participants to successfully interact with the digital therapeutic, with a focus on acceptability and feasibility.

Methods:

A mixed-methods concurrent triangulation design was used to better understand the acceptability and feasibility of the use of the Medly voice app with study participants (n=20) over a four-week period. Quantitative data included engagement levels, accuracy rates, and questionnaires which were analyzed using descriptive statistics. Qualitative data included semi-structured interviews and were analyzed using a qualitative descriptive approach.

Results:

The overall average engagement level was 73%, with a 14% decline between week one and four results. The difference in average engagement levels between the oldest and youngest demographic had the biggest difference, 84% and 43% respectively, but these results were insignificant (Kruskal-Wallis test, H(2) = 3.8, P =.14). The Medly voice app had an overall accuracy rate of 97.8% and was successful in sending data to the clinic. From an acceptability perspective, the voice app was ranked in the 80th percentile and overall users felt the voice app was not a lot of work (average of 2.1 on 7-point Likert scale). However, 13% of users were less inclined to use the voice app at the end of the study. The following themes and subthemes emerged from the interviews: (1) feasibility of clinical integration: (a) users adapted to voice app’s conversational style, (b) device unreliability, and (2) voice app acceptability: (a) device integrated well within household and user lives, (b) users blamed themselves when problems arose with the voice app, (c) voice app missing specific, desirable user features.

Conclusions:

Our findings conclude that participants were largely successful in using the Medly voice app despite some of the barriers faced, proving that an app like this could be feasible to deploy in the clinic. Our data begins to piece together the patient profile this technology may be most suitable for, namely those who: are older, have flexible schedules, are confident using technology, and are experiencing other medical conditions.