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

Abstract

Background Recent years have witnessed a constant increase in the number of people with chronic conditions requiring ongoing medical support in 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 and ubiquitous support. Moreover, different aspects of CAs 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 interaction between patients and physicians. Patient-physician interaction is recognized as a critical success factor for patient satisfaction, treatment adherence, and subsequent treatment outcomes. However, so far, it remains effectively unknown how different interaction styles can be implemented into CA interactions and whether these styles are recognizable by users. Objective The objective of this study was to develop an approach to reproducibly induce 2 specific interaction styles into CA-patient dialogs and subsequently test and validate them in a chronic health care context. Methods On the basis of the Roter Interaction Analysis System and iterative evaluations by scientific experts and medical health care professionals, we identified 10 communication components that characterize the 2 developed interaction styles: deliberative and paternalistic interaction styles. These communication components were used to develop 2 CA variations, each representing one of the 2 interaction styles. We assessed them in a web-based between-subject experiment. The participants were asked to put themselves in the position of a patient with chronic obstructive pulmonary disease. These participants were randomly assigned to interact with one of the 2 CAs and subsequently asked to identify the respective interaction style. Chi-square test was used to assess the correct identification of the CA-patient interaction style. Results A total of 88 individuals (42/88, 48% female; mean age 31.5 years, SD 10.1 years) fulfilled the inclusion criteria and participated in the web-based experiment. The participants in both the paternalistic and deliberative conditions correctly identified the underlying interaction styles of the CAs in more than 80% of the assessments (X21,88=38.2; P<.001; phi coefficient rφ=0.68). The validation of the procedure was hence successful. Conclusions We developed an approach that is tailored for a medical context to induce a paternalistic and deliberative interaction style into a written interaction between a patient and a CA. We successfully tested and validated the procedure in a web-based experiment involving 88 participants. Future research should implement and test this approach among actual patients with chronic diseases and compare the results in different medical conditions. This approach can further be used as a starting point to develop dynamic CAs that adapt their interaction styles to their users.

Personalization of Conversational Agent-Patient Interaction Styles for Chronic Disease Management: Two Consecutive Cross-sectional Questionnaire Studies

“…Given the rising number of chronically diseased patients and the associated financial and personal burdens, CAs represent scalable and ubiquitous digital tools to support chronic patients and relieve human health care professionals. A systematic approach for inducing two specific interaction styles into CAs in a health care setting has previously been developed and validated [ 29 ].…”

“…First, we followed a two-step approach by determining baseline differences of COPD patients’ preferences for their interaction with a CA in Study 1 and subsequently expanding the findings to Study 2, a between-subject online experiment. Second, we deployed a systematic and validated approach for inducing two specific interaction styles into the patient-CA interaction [ 29 ]. Third, we continuously ensured an objective approach by integrating both theoretical knowledge and applied medical expertise into the development of the experiment.…”

“…We designed a between-subject online experiment where patients were randomly assigned to interact with a CA that followed either a deliberative or paternalistic interaction style. We chose these two styles since (1) we have already developed and experimentally tested the implementation of deliberative and paternalistic CA interactions in a recent study [ 29 ]; (2) there is the most information in the literature for these two styles regarding the moderating influences of sociodemographic and health-related variables; and (3) we expected to find significant effects when choosing the most and least preferred interaction styles as determined in Study 1. Both studies did not fall within the scope of Human Research Law, according to the local Swiss ethics authority, and thus did not require any formal authorization.…”

“…The interactions were text-based and followed by a prescripted dialogue based on two predeveloped scripts. These scripts were developed and assessed in a recent study, where they were verified to be perceived as eliciting a deliberative or paternalistic interaction style between a CA and users [ 29 ]. During the conversation with the CA, patients could choose between one to three predefined answer options, which were identical in both conditions (ie, deliberative and paternalistic interaction styles).…”