Crowdsourcing a Training Dataset of Question-and-Answer Pairs for AI-Enabled Health Information Tools on Sexually Transmitted Infections: Protocol for a Cross-Sectional Exploratory Survey Study.

A pilot data collection in Kigali, Rwanda collected 132 questions on sexual health and STIs.

• The pilot was conducted in Kigali, Rwanda as an initial test of the crowdsourcing methodology.
• 132 questions were collected during the pilot phase.
• The pilot preceded the broader multi-country data collection effort across sub-Saharan Africa.

As of August 2025, the study had collected over 5,620 question-and-answer pairs on sexual health and STIs.

• Data collection began on June 12, 2024, and was ongoing as of the report date.
• Questions are being collected via online platforms, paper-based submissions, and in-person interactions at public events.
• Participants must be aged ≥15 years to contribute questions.
• The study targets participants across sub-Saharan Africa.

The study employs a multi-channel crowdsourcing approach to collect sexually transmitted infection questions from a broad population.

• Collection channels include online platforms, paper-based submissions, and in-person interactions at public events.
• The study targets participants aged ≥15 years across sub-Saharan Africa.
• Questions are anonymized before review to protect participant identity.
• Medical professionals review each question and provide accurate, evidence-based answers.
• Health workers also contribute new questions based on their clinical experience.

Collected questions undergo a simultaneous rigorous data processing phase in collaboration with health workers to prepare the dataset for AI training.

• Data processing includes cleaning and tagging for AI training purposes.
• The process adheres to FAIR principles: findability, accessibility, interoperability, and reusability.
• Health workers contribute evidence-based answers and generate new questions based on clinical experience.
• The final dataset is planned for open-access publication in 2025.

The study identifies a lack of contextualized datasets as a key constraint on the effectiveness of AI-enabled health information tools for sexually transmitted infections in sub-Saharan Africa.

• Sexually transmitted infections are described as a significant public health concern, particularly in sub-Saharan Africa, where prevalence remains high.
• Those affected often have limited access to accurate and culturally appropriate health information.
• Existing AI tools such as chatbots are noted as promising but constrained by the absence of contextualized, population-relevant datasets.
• The study frames this dataset as essential for ensuring effectiveness and relevance to diverse populations.

The final dataset will be published as open access to support development of AI-driven health tools and promote public health literacy.

• Open-access publication is planned for 2025.
• The dataset is intended to support development and training of digital and AI-enabled health information tools.
• FAIR principles (findability, accessibility, interoperability, and reusability) guide dataset preparation.
• The dataset is described as contributing to both AI tool development and broader public health literacy goals.