Decreasing barriers to STI and HIV testing with Advies.chat
High quality, anonymous and free
Advies.chat is the first conversation-styled, automated online advice tool in the Netherlands that is based on clinical guidelines and refers users to private testing services for STI and HIV at home. Tailoring health information has been proven effective in STI testing interventions, by increasing personal relevance and elaboration of the message. The output of Advies.chat includes detailed testing recommendations tailored to the user, concise information and calls-to-action to both clinical providers (general practitioners and public STI clinics) and private testing services that were selected based on quality indicators, price and consumer-friendly communication. Advies.chat is free of charge and anonymous: users do not need to register to receive advice. No IP addresses or other identifiable personal data are collected.
In total, the questionnaire included around 60 questions and 200 answers connecting to over 1200 different advice pages. The advice questionnaire was manually tailored to ensure that individuals only answer questions that are relevant for them. Users answer around 15-25 predefined questions about their personal characteristics and sexual behaviors before they receive advice. Based on a needs assessment with 12 potential users, additional questions (for instance about symptoms and sex techniques) and information (for example about testing procedures) were added to Advies.chat.
Advice chat = advies.chat
Users can ask their own questions by shifting from the Advice-mode to the Chat-mode. For the Chat-mode, over 20 years of experience in one-on-one counseling via telephone, e-mail and direct message chat about STIs, HIV, testing and prevention were manually translated into structured responses to common questions. If users cannot find the advice or answers they are looking for, Advies.chat refers to an array of online information sources and an anonymous online helpline.
The chatbot software anonymously registers ‘non responses’ to unidentifiable questions, which are continuously monitored to further improve the tool. Response optimalisation is not yet automated with learning algorithms, because changes in responses first have to validated by a health professional to ensure quality and correctness of the information. Employing machine learning or artificial intelligence that supports Dutch language will be a future step in the iterative improvement of Advies.chat.
Full abstract? Download here.
Enquires and comments can be e-mail to: firstname.lastname@example.org