JMIR Publications

ABSTRACT

Background:

Attitude-targeted interventions are important approaches for promoting vaccination. Educational approaches alone cannot effectively cultivate positive vaccine attitudes. Artificial intelligence (AI)-driven chatbots and motivational interviewing skills show high promise in improving vaccine attitudes and facilitating readiness for vaccination.

Objective:

This study aimed to evaluate the effectiveness of a theory and evidence-based, motivational interviewing-oriented AI digital assistant in improving COVD-19 vaccine attitudes among adults in Hong Kong.

Methods:

This two-parallel-armed randomized controlled trial was conducted from October 2022 to June 2024. Hong Kong adults (N=177) who were vaccine-hesitant were randomly assigned into two study groups. The intervention group (n=91) interacted with the AI digital assistant over five weeks, including receiving a web-based education program comprising five online educational modules and communicating with an AI-driven chatbot equipped with motivational interviewing skills. The control group (n=86) received online learning materials on COVID-19 vaccine information and knowledge, with the same dosage as the intervention group. Primary outcome included vaccine hesitancy. Secondary outcomes included vaccine readiness, confidence, trust in government, and health literacy. Outcomes were measured at baseline, post-intervention, 3-month, and 6-month follow-up. Focus-group interviews were conducted post-intervention. Intervention effects were analyzed using the generalized estimating equation model. Interview data were content analyzed.

Results:

Decreases in vaccine hesitancy were observed while no statistically significant time-by-group interaction effects were found. The intervention showed significant time-by-group interaction effects on vaccine readiness (p = 0.042), confidence (p = 0.017), and trust in government (p = 0.036). Significant between-group differences with medium effect sizes were identified for vaccine readiness (Cohen’s d = 0.523) and trust in government (Cohen’s d = 0.539) post-intervention, respectively. Increases in vaccine-related health literacy were observed and a significant time effect was found (p = 0.012). Three categories were summarized from interview data: improved vaccine literacy, confidence, and trust in government; hesitancy varied while readiness improved; and facilitators, barriers, and recommendations of modifications on the intervention.

Conclusions:

The intervention indicated promising yet significant effects on vaccine readiness while the effects on vaccine hesitancy require further confirmation. The qualitative findings however further consolidate the significant effects on participants’ vaccine attitudes. The findings provide novel evidence to encourage the adoption and refinement of a motivational interviewing-oriented AI digital assistant in vaccine promotion. Clinical Trial: ClinicalTrials.gov (NCT05531058).