JMIR Publications

ABSTRACT

Background:

The intricate relationship between health and the environment is well-established, particularly for global youth. Adolescents experiencing environmental challenges in their communities, such as poor waste management or stagnant water on the streets, are more likely to develop diseases such as dengue fever, Zika, and parasitic diarrhea. However, a substantial gap persists between existing community knowledge about health risk factors and local capacity to effectively transform or eliminate such barriers. Youth-led citizen science strategies—such as the Our Voice Initiative, which aims at enhancing local environments—have demonstrated significant benefits. Successful implementation of these interventions in low-and-middle-income countries requires stronger stakeholder commitment, well-planned solutions, and equitable access to innovative technologies.

Objective:

The application of AI is ready for broad exploration, with several tools providing opportunities for initial testing of potential impacts on enhancing youth and stakeholder engagement and improving ideas in community health improvement programs. Thus, we focused on understanding how Generative Pre-trained Transformer (GPT) language models can provide constructive feedback related to solution-building in a citizen science initiative. Additionally, we sought to investigate how the application of deep learning image processing (specifically, image transformation through AI) can enhance stakeholder engagement.

Methods:

The study consisted of several phases. First, during a Preparation Phase, we evaluated the initial applicability of two AI tools: image transformation and feedback generation using GPTs. We conducted small experiments, transforming images with various prompts and refining solutions with GPTs. The results of these experiments allowed us to reflect on the effectiveness of the AI tools and determine the best ways to adapt them for real-world applications. Second, we applied insights gained from the Preparation Phase to tailor the AI tools for the Our Voice project in El Pozón, Colombia, utilizing these adapted tools during the Our Voice Discuss and Advocacy steps and evaluating their ability to enhance the process.

Results:

The findings demonstrated that AI-based image transformation (e.g., Photoshop AI) effectively produced realistic visualizations of desired environmental improvements, significantly enhancing the citizen scientists' ability to communicate their vision and advocate for change. SecureGPT's structured feedback facilitated deeper reflection on practical considerations, sustainability, and potential unintended consequences of proposed interventions. Additionally, establishing clear guidelines for AI use and facilitating structured reflection sessions among participants improved critical thinking and teamwork.

Conclusions:

The integration of AI tools showed substantial promise in empowering youth-led advocacy, fostering critical thinking, and accelerating social transformation toward healthier community environments. Future research should evaluate the long-term impacts of AI-enhanced citizen science interventions and explore strategies to expand equitable access to these innovative technologies. Clinical Trial: NA