Transforming qualitative research: The AQUATIC approach to AI-driven data analysis

Area of Science:

• Public Health
• Qualitative Research
• Artificial Intelligence

Background:

• Qualitative data analysis in public health is often time-consuming.
• There is a need for rapid, decision-oriented analysis to inform timely public health strategies.
• Existing methods may not be scalable or efficient for time-sensitive public health issues.

Purpose of the Study:

• To present a protocol integrating conversational artificial intelligence (AI) into qualitative data analysis for public health.
• To describe the protocol's safeguards, including data familiarization and human verification.
• To demonstrate the protocol's practical application in a real-world public health case.

Main Methods:

• Evaluated manual coding, AI-assisted coding, and conversational AI within ATLAS.ti.
• Developed a protocol requiring pre-analysis transcript familiarization and immersion memos.
• Utilized structured natural language queries with AI, followed by mandatory human verification of all outputs.

Main Results:

• Conversational AI generated rapid descriptive findings linked to verifiable text, facilitating efficient auditing.
• Theme-level comparisons showed conceptual overlap between human and AI outputs, with documented divergences.
• The protocol enabled quick training of local personnel and built sustained in-house analysis capacity.

Conclusions:

• The protocol offers a pragmatic, transparent, and scalable workflow for question-led, top-down descriptive qualitative analysis using conversational AI.
• Mandatory human oversight ensures data integrity and contextual review.
• This approach is suitable for time-sensitive, decision-focused qualitative work in public health settings.