JMIR Publications

ABSTRACT

Background:

In recent years, the number of people suffering from chronic conditions that require ongoing medical support reaching into the everyday lives of patients is constantly increasing. Global health systems are, however, not adequately equipped for this extraordinarily time-consuming and cost-intensive development. Here, conversational agents (CAs) can offer easily scalable and ubiquitous support. However, different aspects have not yet been sufficiently investigated to fully exploit their potential. One such trait is the interaction style between patients and CAs. In human-to-human settings, the interaction style is an imperative part of the relationship between patients and physicians. The patient-physician interaction is recognized as critical success factor for patient's satisfaction, treatment adherence, and subsequent treatment outcome. However, it remains so far effectively unknown how different interaction styles can be implemented into CA relationships and if these are recognizable by users.

Objective:

The objective of this paper is to develop an operationalization scheme to induce two specific interaction styles into CA-patient dialogues and subsequently test and validate them in a chronic healthcare context.

Methods:

Based on the Roter Interaction Analysis System (RIAS) and iterative evaluations by scientific experts and medical healthcare professionals, we identified 15 communication components that characterize the two operationalized interaction styles, i.e., the deliberative and paternalistic interaction style. These communication components were used to develop two CA variations, each representing one of the two interaction styles. We assessed them in an online between-subject experiment. Here, participants were asked to put themselves in the position of a patient suffering from chronic obstructive pulmonary disease (COPD). Participants were randomly assigned to interact with one of the two CAs and subsequently asked to identify the respective interaction style. A chi-square test was used to assess the correct identification of the CA-patient interaction style.

Results:

88 individuals (48% female, mean age= 31.5 years) fulfilled the inclusion criteria and participated in the online experiment. Participants in both the paternalistic and deliberative condition correctly identified the underlying interaction style of the CAs in more than 80% of the assessments (X2 (1, 88) = 38.23, P = .000; rφ = .68). The validation of the operationalization scheme was hence successful.

Conclusions:

We developed an operationalization scheme tailored for a medical context to induce a paternalistic, respectively, deliberative interaction style into a written interaction between a patient and a CA. We successfully tested and validated the operationalization scheme in an online experiment with 88 participants. Future research should implement and test the operationalization scheme with actual chronic patients and compare results between medical conditions. The operationalization scheme can further be used as a starting point to develop dynamic CAs that adapt their interaction style to their users.