In South Korea, cervical cancer is a principal health concern, affecting 1 in 8 women diagnosed with cancer in their 20s or 30s [1]. The United States Food and Drug Administration recommends the human papillomavirus (HPV) vaccination before the first sexual contact to increase its effectiveness in preventing cervical cancer [2].

According to a study of HPV infection among adult women in South Korea [3], approximately 20,000 women (34.2%) were exposed to HPV infection, with the highest overall infection rate (49.9%) among women aged 18 to 30 years (ie, early adulthood). These changes highlight the need for increased awareness among young people regarding the prevention of HPV infection. The average age at the time of the first instance of sexual intercourse among middle and high-school students in Korea has been reported to be 13.2 years, with the rate of sexual experience increasing every year by 5.9% [4]. Thus, educating adolescents about HPV and instilling awareness of early screening is critical.

In 2016, HPV vaccination became mandatory in South Korea, and the Healthy Women First Step Clinic project, providing free vaccines for girls aged ≥12 years, has been promoted [5]. The primary vaccination rate has increased since the implementation of the vaccination program in 2016. In 2017, it was 72.8% for those born in 2004 and 64.4% for the second dose [6], which was lower than the vaccination rates for the other 16 free vaccines mandated for Korean children below the age of 12 years (95% as of 2018) [6]. These suggest that interventions are needed to increase the willingness to be vaccinated against HPV at an age eligible for free vaccination.

Previous research reported a lack of education to alleviate HPV vaccine hesitancy among adolescents and insufficient communication regarding the positive effects of the vaccine [7]. The implementation rates of school health education programs regarding HPV vaccination among adolescents have been reported to be as low as 15% in South Korea, and health teachers’ accurate responses to questions related to HPV vaccine knowledge among health teachers was <77% [7]. These results suggest that school health education programs do not provide sufficient information on the HPV vaccine to middle-school girls who are at an appropriate age for HPV vaccination and that effective education programs and guidance on HPV vaccination for middle-school girls are needed. Thus, additional programs of age-appropriate vaccination initiatives addressing HPV infection prevention should be provided to middle-school girls aged 12 to 13 years, who are physically, cognitively, and psychologically able to absorb diverse knowledge and experience [8].

eHealth communication [9], which is designed to promote health through digital technologies, such as mobile and social media, has been used as an effective intervention strategy for the practice of adolescent health behaviors, particularly preventive health behaviors [10], HPV vaccination [11,12], prevention of binge drinking [13], and prevention of smoking [14]. Recently, developing a social networking service (SNS) platform that facilitates lasting relationships among its members and provides motivation and professional advice has attracted attention among health researchers [15]. Furthermore, the number of health chatbots that can provide interactive services for teenagers has increased [16,17]. It is important to strategically design eHealth communication that complements other health communication channels and facilitates easy-to-use and effective communication between health care professionals and patients [18]. In particular, mothers must be included in the health communication channel for HPV vaccination because mothers influence their daughters' vaccination decisions and serve as key health care decision-makers [19,20].

The rate of smartphone ownership among middle-school students in South Korea is 95.9%, the highest among adolescents [18], and communication apps represent a major health communication channel for disseminating health information to adolescents [19].

Therefore, to increase HPV vaccination rates among middle-school girls, this study aimed to develop an eHealth communication interventions targeting middle-school girls and their mothers.

The eHealth communication intervention to promote HPV vaccination among middle-school girls in this study was developed using intervention mapping (IM) [21], a systematic approach to developing interventions using a detailed needs analysis of the subject. This paper describes the process of development and usability test of an eHealth communication intervention for middle-school girls and their mothers using SNS to improve HPV vaccination among middle-school girls. The intervention was developed in 6 steps: a needs assessment, setting the program outcomes and objectives, selecting a theory-based method and practical strategies, developing an intervention, developing an intervention plan, and testing the intervention.

The second stage of IM specifies the intervention goals that will change because of the intervention. The goals are divided into intervention goals, expected outcomes, and specific performance objectives [21]. The final health outcome for the intervention was “Increase HPV vaccination uptake among middle-school girls.” The specific performance objectives were to identify positive changes in the key variables (ie, HPV vaccination knowledge, HPV vaccination health beliefs and responsibility, and sexual communication) and develop detailed performance objectives leading a final health outcome.

In the third step of IM, a theory and an appropriate method for studying changes in the related factors were developed. For the conceptual framework of this study, the HBM proposed by Champion and Skinner [27], which was modified by dividing the HBM proposed by Rosenstock [28] into modifiable factors and individual beliefs, was used. In addition, in this study, the expanded parallel process model (EPPM) proposed by Witte [29], a prominent theory in health communication research, was applied to construct an eHealth communication intervention. For the messaging to be successful, it is important to emphasize self-efficacy and high awareness of the perceived threat [30] and response efficacy (ie, the ability to avoid a threat if one takes the action recommended by the persuasion message)—in this case, the intention to receive the HPV vaccination. On the basis of a previous study [22] in which response efficacy had a significant effect on HPV vaccination intention rather than disease threat, the concept of EPPM was applied to the composition of eHealth communication. This conceptual framework is shown in Figure 2.

In addition, the themes derived from the FGIs and using the behavior change techniques (BCTs) [31] were included as specific intervention strategies to improve health beliefs about HPV vaccination. An effective BCT was selected by referring to experimental studies [32,33]. BCTs were used to increase vaccination intention and implementation.

The intervention was designed using a theoretical framework based on the results of a literature review and FGIs and had a health communication strategy aiming to increase HPV vaccination rates among middle-school girls.

“Kakao i Open Builder,” an artificial intelligence (AI) platform developed by KakaoTalk, was used to deliver the HPV vaccination health communication through a chatbot (ie, software programs that can respond to text or voice messages).

In the first stage, the researcher had 3 face-to-face consultations with an expert who had the capacity to develop chatbots in the health and education fields using KakaoTalk to supplement the chatbot’s functions. The researcher also received advice from an expert through email communications and Zoom (Zoom Video Communications, Qumu Corporation) meetings during the development process. The researchers and experts discussed the types of chatbots that should be developed to improve HPV vaccination among middle-school girls. In the second step, the researchers designed a scenario consisting of questions based on the answers of the target group. In this step, we prepared questions and answers based on the details related to HPV vaccination (ie, cervical cancer, HPV, vaccine information, etc). Representative and similar words from frequently used terms were arranged. In the third stage, the implementation of the chatbot, written questions and answers were entered as utterance patterns and responses, and “skills” (ie, techniques that expresses a response to a user according to a topic, and it analyzes a question and creates an appropriate response) were developed so that answers matching key terms could be generated, and were connected with Open Builder. “Skill server” means receiving a request from a bot system, analyzing the information contained therein and responding appropriately. Images and videos necessary for health communication were developed based on the needs assessment. In the fourth stage, the chatbot was connected to the KakaoTalk channel for testing. The experts and researchers continuously tested and updated the chatbot. Additional block updates, troubleshooting, and maintenance were performed to address errors before the intervention was delivered.

In the next stage, a plan was established to implement an eHealth communication intervention. Appropriate methods to recruit participants effectively and implement specific interventions during the COVID-19 pandemic were discussed by consulting with participating middle-school health teachers. Monitoring methods were established to control the quality of the intervention so that records, such as chatbot use frequency and commonly searched blocks, were collected through the KakaoTalk channel manager system.

In the final stage, structured questionnaires, namely the Chatbot Usability Questionnaire (CUQ) [34] and Health Communication Message Review Criteria [35], were used to evaluate the validity of the eHealth communication intervention. The CUQ and health message evaluation tool were translated from English to Korean, and the validity based on the expert and participant groups and the feasibility of the participants of the FGIs were evaluated. The CUQ comprises 16 items (eg, clear indication, error handling, user input language recognition, complexity, and convenience) that were evaluated on a 5-point Likert scale.

The health communication message review criteria comprised 12 minimum standards for a persuasive message and evaluated message clarity, evidence, reliability of delivery method, appropriateness of tone, and uniqueness of the message on a 4-point Likert scale. Because the original text of the CUQ and health communication message evaluation was in English, approval was obtained from the original author for translation and use. The committee approach to translation attempts to reduce the cultural bias inherent in the native language by striving for a consensus translation at the beginning of the translation process [36]. This contrasts with back-translation, in which a single individual initially translates, and then a panel of experts translates. In the committee approach, several translators individually translated the original questionnaire [37]. Therefore, in this study, the scales were translated independently by 2 PhD nursing students, including the researcher, who is fluent in both English and South Korean and has experience translating scales. The next step was for the committee of reviewers to review the translations and reach a preliminary consensus. In this study, 2 nursing professors fluent in English and South Korea were assigned as reviewers and attempted to reach a consensus on the translation of all questionnaire items. Finally, the accuracy of the translation into South Korea was verified by a PhD student from the Department of South Korean Language and Literature.

The prototype chatbot was validated by 3 nursing professors and 4 school health teachers using the translated CUQ and health communication message review criteria. Its feasibility was evaluated by 5 middle-school girls who participated in the FGIs also using the CUQ and health communication message review criteria. According to the method developed by Lynn [38], health message evaluation calculated the item content validity index (I-CVI) for each item, and items with an I-CVI of <0.80 were reviewed and corrected. The item “Message uses an appeal method suitable for the subject” had an I-CVI of 0.71, so it was reviewed and corrected. For the chatbot usability test, the average score was calculated using CUQ’s calculation tool. El Hefny et al [39] used the same tool to evaluate the usefulness of chatbots and set the user acceptance score at 71.1.

This study was approved by the Yonsei Health System Institutional Review Board before data collection (approval number Y-2020-0172). In this study, an electronic signature service called MODUSIGN was used instead of a paper signature. MODUSIGN is an electronic contract service in which a signature has legal effect. Before the study was conducted, all participants were given an explanation and a consent form through MODUSIGN. They signed the consent form on the web and participated in the study.

It was necessary to use messages that girls could understand in a friendly tone, accompanied by images or characters to enhance their understanding. It was also found that interactions could occur through mobile phones. In addition, while emphasizing the need for vaccination and positive awareness of vaccination in the message, intervention strategies, reminders to complete vaccination, and activities that encourage confidence in girls to decide and get vaccinated themselves, such as quizzes, providing incentives, and emotional support, were confirmed. Middle-school girls suggested using mobile phones to highlight the benefits of vaccination, such as free cancer prevention and health care, to middle-school girls and their mothers, and to exchange questions and answers in real time. They also mentioned that promoting vaccination on SNS and having celebrities endorse it would increase their interest.

The expected behavioral outcome of the eHealth communication intervention was that middle-school girls will receive their first dose of the HPV vaccine within 2 months of participation in the intervention. On the basis of the theoretical framework and the HBM, the change determinants were set as health beliefs, self-efficacy, knowledge, sexual communication, and health responsibility. With these changes, the following 2 performance objectives focused on the behavior change of middle-school girls were composed: “I will vaccinate myself to prevent HPV infection” and “I will look up information on HPV vaccination through the chatbot.”

The HBM proposed by Champion and Skinner [27] shows that modifiable factors (eg, age, education level, race, and knowledge) can affect an individual’s health beliefs and that an individual’s health beliefs directly affect preventive health behaviors. Of these modifiable factors, health education is the most necessary before other interventions to promote the prevention behaviors of HPV infection [41] because an increase in knowledge through education increases preventive behavior [28].

In previous studies [42,43], the intention to change health behavior to prevent HPV infection and cervical cancer improved after educating the patients about cervical cancer and HPV infection. Therefore, knowledge appears to be critical for promoting preventive health behaviors. HPV infection prevention can also be included in sexual and reproductive health promotion practices because HPV infection is transmitted through sexual contact. A meta-analysis [44] examining the effects of parent-child sexual communication on adolescents’ safe sexual behavior found that sexual communication with mothers resulted in a safer sexual behavior, particularly among girls. Adolescents face the challenge of transitioning from health care dependent on parents to personal responsibility for health; thus, they need opportunities to develop and express their health care autonomy [45]. Information delivery through outlets such as mass media acts as cues for action urging individuals to engage actively in preventive health behavior and is important in determining health behavior [46]. This study set the eHealth communication intervention as the HPV vaccination uptake “cues for action.” Therefore, the HBM was established as the study’s theoretical framework, and sexual communication between parents and children and health responsibility of middle-school girls were set as additional modifying factors affecting HPV vaccination uptake.

Practical strategies for designing the eHealth communication intervention were constructed based on BCT, following the results of the literature review and needs assessment (Table 1). The contents to improve HPV vaccine knowledge consisted of images, texts, and quizzes.

According to the HBM, the “threat” should be explained in terms of perceived susceptibility and severity. The intervention was designed to explain the severity of the disease caused by HPV infection and that anyone could easily get it, and to check the risk of HPV infection. We provided card news (a new concept news report optimized for mobile environments created by combining images and text) about a woman diagnosed with cervical cancer before marriage because she had not been vaccinated to convey the importance of getting vaccinated and the seriousness of the disease. The perceived benefits were conveyed through images and text that explained that the HPV vaccine is effective in preventing cervical cancer, the benefits of vaccination have been proven, and the HPV vaccine was included in the National Immunization Program worldwide. Perceived barriers were addressed through information that dispelled misunderstandings about vaccinations, side effects, and precautions, was communicated through images and text, and connected to the Korea Centers for Disease Control and Prevention (KDCA) call center to report the side effects of the HPV vaccine. The researcher provided reminders, encouraged the girls to take a pledge to get vaccinated, and developed animation about HPV vaccination support to overcome vaccinations barriers and improve participants’ self-efficacy. The “cues to action” were to increase positive awareness and participants’ intentions through eHealth communication intervention and to provide praise and material incentives, referring to the results of the needs assessment and the BCT of “social reward” and “material incentive.”

The researcher visited the experimental and control groups and explained the study to the school health teachers, the principal and vice principal, and received approval. Due to the COVID-19 pandemic, first-grade middle students were attending school irregularly at time of data collection, so it was decided that using a chatbot was an appropriate intervention method. The researcher asked the school health teachers to encourage participation and sought cooperation to prevent dropouts. A log recording system was built and monitored by the researcher with the help of a chatbot development expert that checked the frequency of block use and the content of the frequently used blocks for each participant to manage the quality of the intervention.

The validity of the eHealth communication intervention was evaluated by experts, and its feasibility was tested in a sample of middle-school girls. Three nursing professors with experience in developing interventions and 4 school health teachers currently managing HPV vaccination at the middle-school level were the experts that evaluated the usability of the chatbot and the content of the health communication strategy. The chatbot usability was evaluated on a 5-point Likert scale with 16 items with the scores being canulated according to the original measurement. The experts’ chatbot usability scores ranged from 67.19 to 90.63, with a mean of 80.13 (SD 8.15), which was higher than the minimum score (71.1) for chatbot acceptance of the study [39]. The supplemented content included “there are terms that are difficult for middle-school students to understand” and “there is a need to motivate adolescents to participate until the end.” A block explaining difficult terms was added to the chatbot to address these points, and biweekly notifications were provided to encourage continuous participation. From April 19 to May 18, 2021, if participants used any block more than once in the JUMP chatbot, it was counted as attendance. In this study, the implementation standard for intervention participation was set to at least 3 times a week of chatbot use based on a previous intervention study using a health chatbot [49]. The content validity of the health communication messages was evaluated by experts using a 4-point scale comprising 12 items. The average I-CVI was 0.94. Of the 12 items, 1 item had a score <0.8, indicating that “The message uses an appealing tone suitable for the target (ie, rational or emotion).” The tone of the messages was made more friendly by mixing praises and positive images to address this point and ensure that self-efficacy would increase instead of perceived threat.

Five students who participated in the FGIs and used the chatbot for 1 week immediately after joining the channel evaluated the chatbot intervention feasibility using the chatbot usability and health communication message measures. Their usability scores ranged from 73.44 to 92.19, with a mean score of 84.06 (SD 7.61), which was higher than the previously identified minimum score for chatbot acceptance [39]. According to the middle-school students, the advantages of the chatbot were “It is very intuitive and easy to use,” “It answers kindly and quickly,” and “It is easy to get an answer right away when you press a button.” The average I-CVI of the health communication message was 0.98. All items scored ≥0.8. The students shared that “The pictures are friendly,” “It’s good because it uses emoticons,” and “It’s good that the health information is detailed.” There were no additional corrections or supplements from the open-ended opinions on the usefulness of chatbots or the evaluation of health communication messages.

The eHealth communication intervention was developed by applying IM, a theoretical framework for developing programs aimed at changing health behaviors. The systematic process has been shown to be effective in integrating theoretical and empirical evidence and planning theory-based interventions. Several studies [50-52] have highlighted its application as an intervention development strategy for HPV vaccination among adolescents. In this study, IM guided the development of systematic interventions, such as designing health communication messages appropriate for middle-school girls and their mothers and deriving motivational strategies to participate in the intervention. This study is significant in that it was attempted as a youth-friendly intervention that connects AI and data in the era of the 4th industrial revolution and uses chatbot to deliver health communication messages.

Most existing interventions for adolescents or their parents have focused on providing information about cervical cancer, HPV infection, the effects and safety of vaccinations, and vaccination methods [53,54]. In this study, BCT was applied both to provide knowledge and strengthen the adolescents’ self-efficacy regarding vaccination and vaccination intention. Marshall et al [55] presented potential interventions using BCT to reduce HPV vaccine hesitancy, which was partially consistent with the BCT applied in this study. Examples of effective BCTs for HPV vaccination included “information about health consequences,” “salience of consequences,” “anticipated regret,” “credible source,” and “pros and cons” [55]. In addition, among the BCTs presented by Marshall et al [55], “identification of the self as a role model,” which encourages the person who has been vaccinated against HPV to set an example for others, should be used in future research. Because BCT has been used as a strategy to enhance both HPV vaccination rates and cervical cancer screening [56], it could be used as a practical strategy in future intervention studies to encourage cervical cancer screening among women in their twenties.

Designing a chatbot system tailored to address a target audience requires an understanding of the individual target users’ background, including their demographic characteristics, living environment, and personality [57]. Interventions tailored to suit these characteristics are effective because the behavior change strategy and persuasive messages can be adjusted according to the target group’s specific needs [51]. According to the behavior change model for health-related chatbots [57], the 4 necessary components included design of chatbot characteristics with a comprehensive understanding of the users’ backgrounds, the relational capabilities of the chatbot (eg, social dialogue and empathetic messages), persuasive conversation skills, and an evaluation method for both the user experience and health outcomes. This study’s chatbot was designed to display text-based health information on the HPV vaccine. In addition, based on the results of FGIs, the chatbot’s persona was designed to be youth-friendly and trustworthy to build the relational competency of the chatbot. Thus, this study can be used as a scientific basis for effectively designed chatbots. In the chatbot behavior change model, the components for persuasive dialogue capacity included rhetorical appeal, credibility (evidence of data), and specific persuasive messages (eg, self-efficacy enhancement narrative and case interviews). In this study, scientific evidence and credible data were presented to provide persuasive health communications about HPV vaccination. Rhetorical expressions (“Isn’t HPV vaccination very important to protect my priceless health?” “Vaccination is only two doses!”) and the case study of an unvaccinated woman diagnosed with cervical cancer were presented. Thus, the chatbot can be seen as an appropriate tool to motivate participants to change their behavior.

The credibility of a health communication intervention is important, particularly when it aims to ensure user acceptance. This study’s chatbot was designed to be youth-friendly, reliable, and professional. The credibility, professionalism, and authenticity of the messages are crucial, as they can influence the persuasiveness of health communication strategies [58,59], and misleading health information can lead to negative health outcomes or the failure to implement health behaviors [60]. Reliability can increase when the intention and expertise of the research team managing the chatbot are introduced to the participants [57]; therefore, in this study, the researcher’s affiliation and the JUMP logo were included in the notification messages, and the sources for all health communication information were provided. Previous research has reported that systematic health communication delivery through health professionals is important to provide accurate information about HPV vaccination in adolescents [61,62]; thus, the HPV vaccine communication strategy involved providing active recommendations for HPV vaccination by health care professionals. Therefore, in this study, the researcher, as a health communication channel, was expected to have a positive effect on the participants’ intention for HPV vaccination.

In this study, health communication was delivered through a chatbot, a cost-effective method of providing health information to adolescents [63]. It was reported that chatbots are useful for life-skill coaching among youth, are easy to use, and are interesting enough to encourage participation from peers [16]. Stephens et al [64] reported that using chatbots to promote weight management and care among +adolescent patients who were prediabetic was effective in treatment adherence, behavior change, and overall health promotion. Therefore, in this study, a Kakao Talk chatbot was developed as a delivery method for HPV vaccination health communication. This study can be used as an empirical support for nursing interventions involving social media and an AI platform considering the positive function of SNS with the advantages of interaction and convenience.

As chatbots improve their ability to understand user intent and respond to input, the amount of personal information shared in chatbots may increase. Therefore, future research should emphasize privacy protection from unauthorized users of chatbots [65]; identify privacy and data safeguards within the system; and provide specific explanations to audiences about privacy, data anonymization, and so on before deploying chatbots.

This study has several limitations. While developing the eHealth communication intervention for HPV vaccination, FGIs were conducted only with middle-school girls, the main targets of the intervention. Follow-up research should conduct FGIs with mothers and develop an intervention program suitable for mothers based on its results to supplement this study’s eHealth communication. In addition, the number of daily chatbot visits was not counted. Several visits within a single day using the same KakaoTalk profile were counted as one visit. It will be necessary to supplement the system to enable the calculation of repeated chatbot participation in the research design.

This study developed and evaluated the usability of a health communication intervention to increase HPV vaccination rates among middle-school girls that was based on EPPM and the HBM and delivered via chatbot. The intervention targeted middle-school girls and their mothers who made decisions concerning adolescent health care. To boost the intention to continue using the JUMP chatbot, a periodic question and answer process should be updated to satisfy ongoing user intent and questions by monitoring participants’ responses. This study confirmed the usability of the health communication strategy to encourage HPV vaccination and is expected to contribute toward establishing a foundation for future youth-friendly health communication strategies.