Utilization of Self-Diagnosis Health Chatbots in Real-World Settings: Case Study

Area of Science:

• Digital Health
• Human-Computer Interaction
• Health Informatics

Background:

• Artificial intelligence (AI)-driven chatbots are increasingly adopted in healthcare for various applications.
• Existing research primarily focuses on controlled settings, with limited understanding of real-world health consumer usage for self-diagnosis.

Observation:

• A large-scale analysis of a self-diagnosis chatbot in China involved over 16,000 users and 47,000 sessions.
• Users across all age groups consulted the chatbot for diverse conditions, including those with privacy and stigma concerns.

Findings:

• Significant user drop-off during consultations and non-therapeutic use were identified as key issues.
• Users reported concerns regarding insufficient actionable information and perceived inaccuracies in diagnostic suggestions.

Implications:

• User-centered design is crucial to address identified issues and enhance chatbot utility in healthcare.
• Future chatbot development should prioritize improved informativeness, usability, trustworthiness, and onboarding to boost user engagement and optimize care.