Participatory co-creation of the SkinSelfcareQA Dataset with persons affected by skin-related neglected tropical diseases in Nigeria to advance equitable artificial intelligence use: A study protocol (Preprint)

BACKGROUND About half of neglected tropical diseases (NTDs) present with cutaneous manifestations, so-called skin-NTDs thus, share similar selfcare morbidity management approaches enabling cost-effective, integrated care as espoused by the WHO. Yet, despite existing guidelines and technical documents for NTDs self-care, their accessibility and adaptability remain limited. Artificial Intelligence (AI) holds huge potential for expanding access to health information and self-care interventions; however, existing AI models are disproportionately trained on data from high-income, western populations rendering current digital health tools less effective and culturally discordant for under-represented groups like people living with skin-NTDs in resource-limited settings. OBJECTIVE This study aims to develop an evidence-based, clinician-validated skin self-care question and answer (Q&A) dataset through participatory co-creation with persons affected by skin-NTDs. Specifically, we will: i). Elicit real-world questions from people affected by skin-NTDs (LF, Leprosy and Buruli ulcer); ii). Utilize insights from people affected by skin-NTDs, self-care guidelines or patient-facing materials to co-create a structured domain-specific Q&A dataset for AI use; iii). Evaluate the technical usability of the co-created SkinSelfCareQA dataset; and iv). Assess if engaging in the participatory co-creation process affects the digital health literacy, self-esteem and internalized stigma of the affected persons involved. METHODS This study will apply participatory action research with purposively selected people affected by the three skin-NTDs. Primary data collection will be done through interview and interactive workshop sessions to elicit practical day-to-day questions/challenges based on their lived experience. A preliminary question bank will be built from the extracted data and expanded using general purpose AI large language models (LLMs), then matching answer pairs will be generated. We will conduct a scoping review of existing literature and guidelines on skin selfcare as LLMs source documents. The resulting draft candidate questions and answer (Q&A) dataset will then undergo multidisciplinary review involving clinicians, dermatologists and public health experts. Kappa statistics will be used to assess final agreement on clinical safety of each response. Through facilitated workshops, these Q&A pairs will be iteratively reviewed with the persons affected to co-creatively finalize the ‘SkinSelfCareQA dataset’ that captures context and cultural nuances. Additionally, we will evaluate the effect of participatory co-creation on their digital health literacy, self-esteem and internalized stigma pre-and-post the psychosocial intervention. RESULTS Funded in May 2026, participant recruitment started in July 2026 with baseline data collection and questions elicitation. No other form of study intervention nor data analysis has commenced as at time of submission. Study results are expected in February 2027. CONCLUSIONS The primary outcome will be an open-access clinician-validated SkinSelfCareQA dataset primed for immediate use to train AI models and/or integration into AI-powered chatbots and digital health tools to improve self-care management, especially for skin-related NTDs and may have utility for other similar conditions like diabetic ulcers. This study will establish a replicable methodological framework for the participatory co-creation of AI training data which is valuable for other conditions such as rare diseases. Importantly, it also mitigates AI algorithmic bias through its deliberate inclusion and co-creation with under-served populations and advances global health equity. CLINICALTRIAL Pan African Clinical Trial Registry, PACTR202607470386334; pactr.samrc.ac.za.