Deliberative and Paternalistic Interaction Styles for Conversational Agents in Digital Health: Procedure and Validation Through a Web-Based Experiment

Background Recent years have witnessed a constant increase in the number of people with chronic conditions requiring ongoing medical support their everyday lives. However, global health systems are not adequately equipped for this extraordinarily time-consuming and cost-intensive development. Here, conversational agents (CAs) can offer easily scalable ubiquitous support. Moreover, different aspects CAs yet been sufficiently investigated to fully exploit potential. One such trait is interaction style between patients CAs. In human-to-human settings, an imperative part physicians. Patient-physician recognized as critical success factor patient satisfaction, treatment adherence, subsequent outcomes. so far, it remains effectively unknown how styles be implemented into CA interactions whether these recognizable by users. Objective The objective study was develop approach reproducibly induce 2 specific CA-patient dialogs subsequently test validate them care context. Methods On basis Roter Interaction Analysis System iterative evaluations scientific experts professionals, we identified 10 communication components that characterize developed styles: deliberative paternalistic styles. These were used variations, each representing one We assessed web-based between-subject experiment. participants asked put themselves position obstructive pulmonary disease. randomly assigned interact identify respective style. Chi-square assess correct identification Results A total 88 individuals (42/88, 48% female; mean age 31.5 years, SD 10.1 years) fulfilled inclusion criteria participated both correctly underlying more than 80% assessments (X21,88=38.2; P<.001; phi coefficient rφ=0.68). validation procedure hence successful. Conclusions tailored context written CA. successfully tested validated experiment involving participants. Future research should implement among actual diseases compare results conditions. This further starting point dynamic adapt