CoTI: Collaborative Theme Identification Agent

Research applicability

Multi-agent framework supporting qualitative analysis of patient interviews through key-phrase extraction, theme identification and codebook generation.

Getting started

Start with arXiv:2512.16063 and its version history; inspect the full method before attempting a small, consented pilot.

Maintenance

Future maintenance not established by this source review.

License

Licence not independently audited; check source terms before reuse.

Workflow & human oversight

Workflow level unspecified for this resource; researchers define objectives, control input data and validate outputs. Suggested adoption steps were not exercised.

Workflow level is not an evidence score; higher is not better evidence.

Approach

The authors describe Instructor, Thematizer and CodebookGenerator agents and compare CoTI outputs with investigator and NLP baselines on 12 heart-failure interviews.

Limitations

Abstract-level review of the retrieved v1 record only; queued metadata names a later title/version not returned by this extraction. Findings on 12 interviews and marginal collaboration gains are author-reported, not independently reproduced. Sensitive interview data require appropriate consent, privacy safeguards and expert interpretation.

Reproducibility

Stored primary arXiv v1 abstract/metadata retrieved 2026-10-17 reviewed; no full text, software, model outputs or reported results independently reproduced.

Adoption requirements

Review the method and establish model access, approved transcript handling and a domain-led coding protocol; distribution and full setup were not verified.