Health Information Framing, Source Characteristics, and Health Behavior Promotion

Participants view an AI-sourced message emphasizing the benefits of vaccination.

Participants view an AI-sourced message emphasizing the risks of not being vaccinated.

Participants view an AI-sourced message emphasizing social or public-health approval of vaccination.

Participants view an AI-sourced message explaining vaccination and booking procedures in a structured format.

Participants view an AI-sourced message attributed to professional or expert authority.

Participants view an official message emphasizing the benefits of vaccination.

Participants view an official message emphasizing the risks of not being vaccinated.

Participants view an official message emphasizing social or public-health approval of vaccination.

Participants view an official message explaining vaccination and booking procedures in a structured format.

Participants view an official message attributed to professional or expert authority.

Participants view a gain-framed message emphasizing the benefits of mental health help-seeking, with peer social cues.

Participants view a gain-framed message emphasizing the benefits of mental health help-seeking, attributed to an expert source.

Participants view a loss-framed message emphasizing the risks of not seeking mental health support, with peer social cues.

Participants view a loss-framed message emphasizing the risks of not seeking mental health support, attributed to an expert source.

Participants view a message presenting help-seeking as a common and socially approved behavior.

In the Hunan population, assessed immediately after exposure with one 5-point item: "I am willing to receive the HPV vaccine in the future." Responses range from

1 (strongly disagree) to 5 (strongly agree). Higher scores indicate stronger intention to receive the HPV vaccine.

In Tongzhou students and primary caregivers, assessed using three 7-point items asking how likely the participant would be to seek help from a psychologist, a psychiatrist, and another doctor if experiencing persistent emotional distress or thoughts of self-harm. Scores range from 1 (definitely would not) to 7 (definitely would). The mean of the three items will be calculated; higher scores indicate stronger professional help-seeking intention.

In the Hunan population, a binary behavioral outcome recorded by the online questionnaire platform after exposure. A positive response is defined as either selecting direct booking at the first decision point or agreeing to provide contact information for HPV vaccination booking at the end of the booking pathway. Viewing information without proceeding to booking or providing contact information is not counted as a positive response.

In the Hunan population, assessed with five HPV-adapted Vaccine Trust Gauge items covering vaccine safety, protection against serious illness, relative health risk, benefits versus risks, and effectiveness. Responses will be scored in the direction of greater trust using the prespecified VTG scoring and missing response rules. The item mean will be rescaled to 0-100; higher scores indicate greater trust.

In the Hunan population, assessed with five Vaccine Trust Gauge-adapted items measuring trust in HPV vaccine recommendations from schools or official institutions, doctors or nurses, family or friends, classmates or peers, and social media. Responses will be scored using the same prespecified VTG rules. The item mean will be rescaled to 0-100; higher scores indicate greater trust.

In the Hunan population, assessed using seven HPV-adapted 7C vaccination- readiness items covering calculation, collective responsibility, confidence in the system, confidence in vaccine safety, social conformity, convenience or constraints, and complacency. Items will be reverse scored where required and averaged so that higher scores indicate greater vaccination readiness.

In the Hunan population, assessed using the mean of three 5-point Health Belief Model items representing perceived susceptibility to HPV infection. Items will be reverse scored as specified in the analysis plan so that higher scores indicate greater perceived susceptibility.

In the Hunan population, assessed using one reverse-scored 5-point Health Belief Model item concerning whether the health effects of HPV infection are not serious. The score will be transformed so that higher scores indicate greater perceived severity.

In the Hunan population, assessed using the mean of six 5-point Health Belief Model items concerning perceived drawbacks, cost, insufficient information, uncertain long-term side effects, sexual stigma, and sensitivity of the topic. Higher scores indicate greater perceived barriers to HPV vaccination.

In the Hunan population, assessed using the mean of two 5-point Health Belief Model items concerning exposure to media reports promoting HPV vaccination and the influence of a doctor's recommendation. Higher scores indicate stronger cues to action.

In Tongzhou students and primary caregivers, assessed using the Help-Seeking dimension of the Universal Mental Health Literacy Scale for Adolescents (UMHL-A). Five applicable items are used in students and four in caregivers. Items are scored on a 5-point scale, averaged, and rescaled to 0-100. Higher scores indicate more favorable help-seeking attitudes.

In Tongzhou students and primary caregivers, assessed using the three-item Stigma dimension of the Universal Mental Health Literacy Scale for Adolescents (UMHL-A). Items are scored on a 5-point scale, averaged, and rescaled to 0-100 in the direction of less stigmatizing and more favorable attitudes.

In the Hunan population, assessed using two post-exposure 5-point items asking whether the participant is willing to use an AI chatbot to obtain further HPV vaccination information and whether generative AI would be used as one source of HPV vaccination information. Both items will be coded in the direction of greater intention and averaged.

In Hunan and Tongzhou participants, assessed using the three-item Benevolence dimension of the Human-Generative AI Trust Scale, adapted to HPV vaccination or mental health information. The items assess whether generative AI acts in the participant's best interest, is willing to understand the participant's needs and preferences, and will help to the best of its ability. Items use a 5-point response scale and are averaged; higher scores indicate greater benevolence-based trust.

In Hunan and Tongzhou participants, assessed using the three-item Competence dimension of the Human-Generative AI Trust Scale. The items assess whether generative AI can effectively support the participant in understanding health information, performs well in providing health information support, and has the expected functions for that role. Items use a 5-point response scale and are averaged; higher scores indicate greater competence-based trust.

In Hunan and Tongzhou participants, assessed using the three-item Reciprocity or Dependency dimension of the Human-Generative AI Trust Scale. The items assess whether participants believe they can rely on generative AI, whether it can continue to support health decisions, and whether they refer to health information provided by generative AI. Items use a 5-point response scale and are averaged; higher scores indicate greater reciprocity-based trust.

In Tongzhou students and primary caregivers, assessed immediately after exposure using the 10-item Connor-Davidson Resilience Scale (CD-RISC-10). Each item is scored from 0 (not true at all) to 4 (true nearly all the time). The 10 item scores are summed to a total score ranging from 0 to 40; higher scores indicate greater psychological resilience. If the online platform stores responses as 1-5, scores will be recoded to 0-4 before summation.