Evidano is an AI-powered qualitative data analysis platform that helps researchers analyze interviews, open-ended surveys, and documents. This post explains how AI-enabled qualitative analysis can surface actionable insight from the PLOS One community AMR study in Enugu, Nigeria, and how teams can apply thematic and cross-segment methods to design community-informed interventions. The primary keyword for this post is "AI qualitative analysis of antibiotic use" and the discussion below highlights methods, concrete statistics, and direct quotes from the PLOS One study (Maduko et al., published July 23, 2026) to make findings extractable for researchers, program designers, and health communicators.
Key Takeaways
According to the PLOS One article published July 23, 2026, community antibiotic use in Enugu is widespread and shaped by access, culture, and informal drug markets: 1, 281 people were surveyed and 57.7% reported antibiotic use within six months. PLOS One.
- Survey size and timing: the baseline survey included n = 1, 281 respondents collected between July 18 and October 4, 2024, and the paper was published on July 23, 2026.
- Prevalence and access: 57.7% reported antibiotic use in the prior six months and 48% of users self-medicated, with 62% sourcing drugs from chemist shops according to PLOS One (Maduko et al., 2026).
- Awareness and practices: only about 15% had heard of AMR and 49% of antibiotic users stopped treatment once they felt better, data reported in PLOS One on July 23, 2026.
What happened and how the PLOS One study measured it
Answer: The PLOS One study combined a statewide survey (n = 1, 281) with 20 interviews and four focus group discussions to map community antibiotic behaviours in Enugu, Nigeria. The authors report that data collection occurred between July 18 and October 4, 2024, and analysis used descriptive statistics plus Braun and Clarke’s thematic analysis framework (Maduko et al., published July 23, 2026).
The PLOS One team operationalised antibiotic misuse as use without prescription, use for non-bacterial conditions, incomplete courses, or non-medical uses, and they triangulated survey frequencies with verbatim interview and FGD quotes to ground interpretation (Maduko et al., PLOS One, 2026).
Key numeric measures reported by PLOS One include: 57.7% used antibiotics in the prior six months, 48% self-medicated, 62% obtained antibiotics from chemist shops, only 15% had heard of AMR, and 49% stopped when they felt better (Maduko et al., 2026).
Findings snapshot table
| Date (study/publish) | Metric (PLOS One) | Value | Implication |
|---|---|---|---|
| Data collection: Jul 18–Oct 4, 2024 | Baseline survey sample | n = 1, 281 | Broad state coverage across 17 LGAs, both urban and rural (Maduko et al., PLOS One, 2026). |
| Reported in paper: published Jul 23, 2026 | Antibiotic use in prior 6 months | 57.7% | Majority of respondents used antibiotics recently, indicating high community-level exposure (Maduko et al., PLOS One, 2026). |
| Reported in paper: Jul 23, 2026 | Self-medication (no healthcare professional consulted) | 48% | Nearly half of antibiotic consumption occurred without clinical oversight (Maduko et al., PLOS One, 2026). |
| Reported in paper: Jul 23, 2026 | Heard of AMR | ≈15% | Low public awareness of antimicrobial resistance complicates behavior-change messaging (Maduko et al., PLOS One, 2026). |
| Reported in paper: Jul 23, 2026 | Primary source: chemist shops | 62% | Informal drug vendors are primary access points, so interventions must engage these actors (Maduko et al., PLOS One, 2026). |
Implications for public health researchers and qualitative teams
How should qualitative teams prioritise analysis after a mixed-methods AMR study?
Answer: Prioritise inductive thematic coding to surface local explanatory models, then run frequency and cross-segment analyses to see who holds which beliefs. The PLOS One study (Maduko et al., 2026) shows that misconceptions and culturally embedded practices (for example using antibiotics for perceived "cleansing" or pregnancy termination) can only be identified through open-ended methods paired with survey frequencies.
The PLOS One authors used NVivo-assisted thematic coding and a codebook (see their S1 Appendix) to move from raw quotes to policy-relevant themes; replicate this pipeline by combining transcription, thematic coding, and structured frequency reports.
What analytic outputs are most useful for program design?
Answer: Deliverables that program teams value are thematic summaries linked to demographic segments, verbatim exemplar quotes, and simple decision trees of care-seeking pathways. The PLOS One study produced participant quotes such as "Antibiotics help build you... antibodies that fight disease." (Interview Participant-3, M) to show local models of medicine.
Pair those qualitative exemplars with survey counts (Maduko et al., PLOS One, 2026) (for example the finding that 48% self-medicate and 62% buy from chemists) to prioritize where to intervene and which messages to co-design.
How to make AMR messaging culturally resonant?
Answer: Co-produce messages with trusted local actors like religious leaders and chemists, and use local channels such as radio, as participants recommended in the PLOS One study (Maduko et al., 2026).
The PLOS One participants suggested "Radio jingles will work... some people in the villages won’t see billboards." (Interview 18), indicating channel choices matter as much as message content.
How Evidano helps: mapping AMR research needs to AI-enabled features
What is Evidano and why mention it here?
Evidano is an AI-powered qualitative data analysis platform that helps researchers analyze interviews, open-ended surveys, and documents.
This definition clarifies the product baseline before mapping problems to features; use Evidano to accelerate coding, extract frequencies, and produce segment-linked thematic reports with illustrative quotes.
Problem: slow synthesis of large mixed datasets → Solution: automated thematic + frequency analysis
Answer: Mixed-methods AMR projects need fast, reliable synthesis to move from transcripts to policy-ready insights. Evidano automates transcript ingestion, runs inductive topic extraction, and outputs code frequency matrices so teams can quantify how many participants expressed a theme (for example, the PLOS One finding that only 15% had heard of AMR).
Evidano feature reference: see Evidano features for thematic, content, frequency, and cross-segment analyses.
Problem: noisy transcripts and PII risk → Solution: transcription with redaction
Answer: Accurate, secure transcription is essential for trustworthy qualitative analysis. Evidano provides transcription with custom dictionaries and PII redaction to preserve participant confidentiality while keeping local terms intact for thematic coding.
Technical fit: teams replicating the PLOS One workflow can ingest audio, apply redaction, and export coded excerpts for stakeholder-facing reports; see Evidano speech-to-text.
Problem: community co-design needs chat over docs → Solution: AI chat over your dataset
Answer: Researchers and program designers need to interrogate datasets quickly and generate plain-language summaries for community partners. Evidano offers an AI chat interface that answers questions over uploaded transcripts and survey tables, producing extractable quotes, counts, and suggested next steps.
Operational benefit: use the AI chat to generate stakeholder-ready briefings that pair participant quotes (verbatim) with survey percentages, matching the evidence style used in the PLOS One article.
FAQ: AI qualitative analysis of antibiotic use
How can AI help identify surprising community practices like the use of antibiotics for pregnancy termination?
Answer: AI-assisted inductive coding flags co-occurrence patterns and unusual phrase clusters that human analysts can then validate. The PLOS One study (Maduko et al., 2026) found emergent practices such as using antibiotics for "cleansing" or attempted pregnancy termination through interview and FGD narratives; AI can surface similar low-frequency but high-impact patterns from large transcript sets.
Follow-up: validate flagged excerpts manually and include community review before communicating sensitive findings.
What data and metadata must I collect to replicate a PLOS One–style mixed-methods analysis?
Answer: Collect structured survey responses (with dates and demographic fields), verbatim audio transcripts, and fieldnotes, and retain a codebook. Maduko et al. combined a survey (n = 1, 281) with 20 in-depth interviews and 4 FGDs and used a published codebook (see S1 Appendix) to ensure reproducibility.
Practical tip: include collection dates (the PLOS One study collected data Jul 18–Oct 4, 2024) to analyse temporal patterns and ensure transparency in reporting.
Is AI analysis ethical for sensitive AMR topics in low-resource settings?
Answer: AI can be ethical when data are anonymised, participants consent to analysis, and outputs are vetted with communities. The PLOS One study obtained ethical approvals and written consent; mirror those safeguards in your workflow and use platforms that support PII redaction and encrypted storage.
Note: this post is research-focused and not clinical advice; follow local ethical guidelines for human-subjects research.
How fast can my team go from raw audio to segment-linked themes using AI tools?
Answer: With modern AI pipelines, teams can move from raw audio to draft themes in days rather than months. Evidano’s integrated transcription, dictionary support, and thematic extraction cut manual coding time and produce exportable frequency tables and quote packs for stakeholder validation.
Operational caveat: speed should not replace rigorous validation; the PLOS One study combined automated tools with manual NVivo coding and community-informed interpretation.
Conclusion & Next Steps
The PLOS One study (Maduko et al., published July 23, 2026) shows that community antibiotic use in Enugu is widespread, socially embedded, and driven by structural access issues as well as local beliefs. AI-enabled qualitative analysis helps teams move rapidly from transcripts and surveys to prioritized, segment-linked recommendations that respect local context.
Actionable next step: replicate the PLOS One mixed-methods approach by pairing representative survey counts (for example, n = 1, 281 and 57.7% recent antibiotic use) with inductive thematic analysis and community co-design.
Get started: to trial a pipeline that combines transcription, thematic extraction, frequency analysis, and an AI chat over your dataset, Try Evidano for free.