Implementation of clinical practice guidelines using the Plan-Do-Study-Act framework
The multisite hybrid implementation study, part of the Academy’s Health Informatics Infrastructure Registry Study on gestational diabetes, ran from 2020 to 2023. Ten registered dietitians at five clinical sites completed every activity, and each gave a semi-structured closing interview conducted by the principal investigator. Authors are affiliated with Bowling Green State University, the Academy of Nutrition and Dietetics, and the University of New Mexico.
The article is open access under a CC BY licence, was approved by the Bowling Green State University Institutional Review Board, and follows the Standards for Reporting Qualitative Research.
How Evidano was used
The team analysed the interview transcripts using a constructivist approach and reflexive thematic analysis (Braun and Clarke), with three rounds of manual coding that consolidated 52 initial codes into 28 codes under six themes. To support triangulation, Evidano then generated two independent thematic analyses: one from AI-generated coding, and one using the researchers’ own codebook. Both were compared against the team’s manual coding.
The AI-coded run produced a coarser structure of three themes and 12 subthemes. When the researchers’ codebook was applied, the tool demonstrated continuity with the manual coding and, in the authors’ words, augmented their analysis and analytic reflexivity. The final themes were then mapped onto Normalization Process Theory and the Consolidated Framework for Implementation Research.
How Evidano fit the workflow
- 1Interview transcripts from 10 dietitians
- 2Three rounds of manual reflexive coding: 52 codes refined to 28 under 6 themes
- 3Evidano generates two independent analyses: AI coding plus the researchers’ codebook
- 4AI output compared with the team’s manual coding
- 5Triangulated, researcher-validated themes mapped to implementation frameworks
“To support analytic triangulation, AILYZE, an artificial intelligence (AI)–powered qualitative analysis tool was used to generate two independent thematic analyses: one based on AI‐generated coding and another using the researcher‐developed codebook.”
“When our codebook was applied, the AI tool demonstrated continuity with manual coding and augmented our analysis and analytic reflexivity.”
“...an independent lens to complement but not replace the researcher's analysis.”
An independent lens to cross-check and strengthen your coding.
Apply your codebook or an AI-suggested one, as in the study.
Disclose and cite AI use, keeping the workflow auditable.
AI supports the analysis; the researcher makes the call.