What do people ask SickleCellPedia, an AI assistant for sickle cell disease? A sample of 1,925 messages from 434 users in 33 countries, June 2025 to October 2026

SickleCellPedia is a free AI assistant, on WhatsApp and the web, that answers questions about sickle cell disease from a curated library of trusted sources and cites where its answers come from. This poster describes what people actually asked it. We analysed 1,925 messages sent by 434 users in 33 countries and territories between June 2025 and October 2026, excluding SCKIN’s own test conversations. A language model classified each message by type and tagged each conversation with the needs it raised. The largest user groups were in India, Niger, Nigeria, the United States and France. Users wrote in eight languages, including English, French, Hindi, Marathi and Swahili, and asked for answers in five more, among them Hausa, Arabic and Urdu. Of the users who said who they were, 66% were patients, 29% family members or caregivers and 6% health professionals. The most frequent needs were complications (26% of conversations), understanding the disease (24%) and medicines (22%), followed by pain and crisis management, inheritance and testing, daily living, curative therapies, and mental health, relationships and work. As conversations went on, questions shifted from facts and medicines toward psychological and existential concerns. These results are preliminary and descriptive: classification was a single pass by a language model with no second coder, and country reflects where a phone number is registered. Future work will record, with consent, each user’s role, country and language so that needs can be mapped by community; follow up to learn whether an answer helped someone seek care, talk to their doctor or manage a crisis; and share anonymised, country-level dashboards of unmet information needs with patient organisations and clinicians.