From Interviews to Insight: AI Qualitative Data Analysis

According to the PLoS ONE study published on November 5, 2026, qualitative interviews suggested that social connection and wellbeing were more closely associated with positive behaviour change in the ELLY intervention than a focus on healthy weight alone.

• The ELLY intervention ran for 12 weeks from September 2023 to November 2023 and reached 75 residents across two disadvantaged Scottish communities.
• Researchers interviewed 20 participants, 10 from each community, and used framework analysis to identify themes and construct a conceptual pathway model.
• The study reported themes involving behaviour change, social connectedness, healthier diets, perceived physical and mental health improvements, and new skills and knowledge.
• Participants could receive a £25 shopping card after earning at least 9 loyalty-card stamps through community participation during the 12-week intervention.
• The researchers stated that qualitative findings alone were insufficient to establish definitive causation, showing why interview insights should be combined with longer-term and mixed-methods evaluation.

The ELLY study examined how residents experienced a co-designed healthy weight and wellbeing intervention in two disadvantaged communities in Forth Valley, Scotland. According to the 2026 PLoS ONE article by Julie Cowie and colleagues, the intervention combined free soup twice weekly, community-activity engagement, optional goal setting, information resources, and self-monitoring of weight and wellbeing.

According to the study methods, researchers recruited 35 participants in one community and 40 participants in the other between June 20, 2023, and August 31, 2023. The intervention was delivered by volunteers between September 2023 and November 2023.

The qualitative component used semi-structured, face-to-face interviews lasting approximately 30 minutes. According to the authors, two researchers conducted the interviews at participants' homes or in community settings, and the recordings were stored using encrypted equipment and password-protected systems.

The researchers developed an initial coding framework after reviewing five interviews, refined the framework through team discussions, and independently cross-checked 10% of the data. The research team then used the interview findings to develop a conceptual logic model describing links between intervention components, mechanisms, and reported outcomes.

One participant described the intervention's effect on activity and diet by saying, "Now I walk every night...I continue in this situation, like walking, and soup, and healthy food, that has helped me a lot." The quotation illustrates why qualitative analysis can reveal perceived mechanisms that standard outcome measures may not explain.

The study authors also cautioned against overstating the results. The 2026 PLoS ONE article states that the qualitative data were "insufficient to establish definitive causation, " because stronger causal claims would require more extensive mixed-methods research.

AI qualitative data analysis is most useful when it accelerates organization and comparison while researchers retain responsibility for interpretation, context, and ethical judgement.

For public health researchers, the ELLY study shows why coding should distinguish intervention mechanisms from outcomes. The 2026 PLoS ONE researchers separated mechanisms such as peer recruitment and community connection from outcomes such as social connectedness and behaviour change.

For evaluation teams, cross-segment analysis can reveal whether an intervention works differently across communities, participant profiles, or levels of engagement. The ELLY researchers reported that local context, community readiness, partner participation, and existing assets shaped delivery in different ways across the two communities.

For qualitative methodologists, AI-assisted coding should support rather than conceal the analytic process. Researchers should preserve the interview sample, code definitions, evidence excerpts, disagreement-resolution process, and limits of inference.

For health and policy analysts, participant language can identify implementation barriers that numerical outcomes may miss. One ELLY participant said, "I found some things very specifically orientated to different groups and I didn’t fit into those groups, " highlighting how inclusion and accessibility can affect intervention reach.

Health research involving weight, mental health, or other sensitive topics requires informed consent, privacy protection, careful de-identification, and non-diagnostic interpretation. AI-generated themes should be reviewed by qualified researchers before they inform clinical, policy, or service decisions.

The ELLY study shows how qualitative interviews can explain the pathways connecting a community intervention with reported changes in wellbeing, social connection, diet, confidence, and behaviour. AI qualitative data analysis can make this kind of work more manageable by organizing transcripts, comparing segments, and linking themes to supporting evidence, while researchers preserve interpretive and ethical control. Evidano is designed to help research teams turn interviews and other qualitative sources into transparent thematic and cross-segment findings. Try Evidano for free and explore a faster workflow for qualitative research analysis.