From leadership research to district action: Large Language Models in Cardiovascular Prevention: A Narrative Review and Governance Framework

AI-assisted editorial implementation guide. Proposed district actions are separate from the original authors’ findings and recommendations. Independent peer review of this guide is not asserted.

Read source Search paper agent CC BY 4.0

Evidence and transfer

Education-related source: verify which participants, institutions and outcomes were actually studied before district adoption. Source publication type: review-article. License and identity checks do not establish study quality or causality.

Original authors’ abstract, reproduced under the source license:

Abstract Background: Large language models (LLMs) are becoming progressively integrated into clinical practice; however, their role in cardiovascular (CV) prevention remains unclear. This review synthesizes current evidence on LLM applications in preventive cardiology and proposes a governance framework for their safe translation into practice. Methods: We conducted a comprehensive narrative review of literature published between January 2015 and November 2025. Evidence was synthesized across three functional domains: (1) patient applications for health literacy and behavior change; (2) clinician applications for decision support and workflow efficiency; and (3) system applications for automated data extraction, registry construction, and quality surveillance. Results: Evidence suggests that while LLMs generate empathetic, guideline-concordant patient education, they lack the nuance required for unsupervised, personalized advice. For clinicians, LLMs effectively summarize clinical notes and draft documentation but remain unreliable for deterministic risk calculations and autonomous decision-making. System-facing applications demonstrate potential for automated phenotyping and multimodal risk prediction. However, safe deployment is constrained by hallucinations, temporal obsolescence, automation bias, and data privacy concerns. Conclusions: LLMs could help mitigate structural barriers in CV prevention but should presently be deployed only as supervised “reasoning engines” that augment, rather than replace, clinician judgment. To guide the transition from in silico performance to bedside practice, we propose the C.A.R.D.I.O. framework (Clinical validation, Auditability, Risk stratification, Data privacy, Integration, and Ongoing vigilance) as a roadmap for responsible integration. Keywords: large language models, artificial intelligence, cardiovascular prevention, risk stratification, clinical decision support

Study methods

This work was conducted as a comprehensive narrative review, with data extraction and synthesis organized around the three core application domains (patient, clinician, and system) outlined in the introduction.

Limitations

This narrative review has inherent methodological characteristics that should be considered when interpreting its findings. Evidence selection relied on author expertise rather than systematic screening protocols, which may introduce selection bias. Formal quality appraisal was not performed given the heterogeneity of included evidence (technical benchmarks, observational studies, position papers, and conceptual frameworks), for which no single appraisal tool is appropriate. The restriction to English-language publications may limit generalizability to non-English-speaking contexts.

A 90-day district inquiry

Days 1–15 · Appraise. Read the source methods, recruitment, comparator, outcomes and limitations. Record how the study setting differs from your district and which claim the design can support.
Days 16–30 · Design. Select one recurring district decision. Publish the decision owner, evidence requirements, consultation process and appeal route; test the process on a bounded low-risk decision. Agree participation, privacy protections, baseline and stopping rules before starting.
Days 31–60 · Observe. Collect the pre-agreed measures, keep a log of implementation changes and invite confidential voluntary feedback. Preserve existing support for staff and learners.
Days 61–90 · Decide. Track decision time, unresolved objections, documented reasons, appeals and access differences. Review whether the process changes who can participate and who carries the burden. Make a dated decision to stop, revise or extend the inquiry.

Board decision brief

Present the local problem, source setting, evidence limits, intervention, responsible owner, total cost, measured changes and uncertainty. Separate staff perceptions, implementation measures and student learning outcomes.

Attribution and reuse

José Ferreira Santos; Hélder Dores. Large Language Models in Cardiovascular Prevention: A Narrative Review and Governance Framework. Diagnostics (2026-01-26). https://doi.org/10.3390/diagnostics16030390

Prepared 2026-11-07. Source authorship and DOI are preserved. The abstract and source passages are attributed extracts; this editorial guide is offered under the same CC BY 4.0 license.