The conversational AI system based on LLM processes user queries with a sophisticated generation, decision, action, and safety framework to provide optimal feedback. The system has several key features. Firstly, it utilizes retrieval-augmented in-context learning to enhance responses generated by the LLM, sourcing information from trustworthy websites. Secondly, it includes Guardrail module that addresses potential harmful content in the LLM-generated responses by clarifying and validating the content before delivery. Furthermore, the AI system is equipped with Self-checking memory module that retains essential clinical characteristics across multi-turn dialogues, ensuring the continuity and consistency of its interactions with users. The educational content about vaccination was derived from the official website of World Health Organization (https://www.who.int/news-room/questions-and-answers/item/vaccines-and-immunization-what-is-vaccination).

Vaccine acceptance was evaluated based on responses to the question, 'Based on your actual situation, please choose the extent of your willingness to get vaccinated.' The response options were: Absolutely refuse, Refuse but uncertain about the future, Have not considered it yet, Accept but still considering, and Completely willing. Vaccine acceptance was considered if the participant selected 'Completely willing.

The Vaccine Hesitancy Scale was developed based on the "3Cs" model (confidence, complacency, and convenience) by the SAGE Working Group on Vaccine Hesitancy, encompassing three aspects: confidence, complacency, and convenience. It utilizes a Likert 5-point scale (Strongly Disagree, Disagree, Neutral, Agree, Strongly Agree) for scoring. For the complacency dimension, aspect, scoring is reversed with "Strongly Agree" to "Strongly Disagree" rated as 5 to 1 points, respectively. In contrast, other items follow the standard scoring, where "Strongly Agree" to "Strongly Disagree" are scored from 1 to 5 points. Confidence is measured by perceived vaccine safety and efficacy. Complacency is assessed through perceived necessity of the vaccine and the severity of the preventable diseases. Convenience is evaluated by the perceived ease of transportation and appointment scheduling.

To assess feelings of coercion related to vaccination, participants were asked a single question: 'When deciding whether to receive the vaccines, do you feel forced to either take or not take the vaccine?' This question utilizes a 5-point Likert scale (very low, low, neutral, high, very high) with lower scores indicating lower feelings of perceived coercion.