From Innovation to Impact: The CREATE Framework as a Blueprint for Large Language Model Adoption in Opioid Treatment Programs

Unlabelled Large language model (LLM)-based systems have tremendous potential to improve patient-centered health care, especially for medically underserved populations. However, realizing this potential requires careful consideration of the socio-technical contexts in which LLMs are used. The design of such systems should consider the vulnerabilities of any medically underserved group that the system intends to support, and provide trustworthy evidence for its use. This viewpoint reports the lessons learned from…

By the Truvace deskOctober 17, 20262 min read1 sourcePeer-reviewedTRV-2026-1751· cite & verify

In brief

This viewpoint draws on the authors' experience and research with the CREATE framework, which stands for Culture, Respect, Education, Advancement, Trust, and Expertise. It is meant to engage multiple stakeholders in guiding how LLM-based systems are brought into health care delivery. The authors also extend the traditional hierarchy of evidence model and point to junction points in clinical workflows where such systems might help.

The paper matters because people with opioid use disorder often distrust the health care system after stigmatizing experiences, so tools built without that context could fail them. The piece is a proposal, not a trial. It names research directions that must be addressed before LLM systems can be usefully integrated into patient care, so real-world benefit is not yet shown.

Main points

CREATE stands for Culture, Respect, Education, Advancement, Trust, and Expertise and is used to engage multiple stakeholders.
The authors extend the traditional hierarchy of evidence model to help evaluate output generated by AI systems.
Users need to understand how data quality affects AI outputs and the evidence generated from them.
Systems for people with opioid use disorder must account for a general lack of trust and stigmatizing experiences with health care.
The paper proposes concrete approaches for validating and implementing the framework and lists research directions.

The gain

The authors propose that LLM-based systems, guided by the CREATE stakeholder framework, could improve patient-centered care for medically underserved groups such as people with opioid use disorder.

The rundown

The viewpoint argues that LLM design should account for the socio-technical contexts of use and the vulnerabilities of the medically underserved group the system is meant to support. It also says such systems should provide trustworthy evidence for their use.

The authors say users of LLM-based tools must know the strengths, limitations, and effects of these technologies on care workflows and on the quality of evidence behind health care recommendations. They treat the evidence pyramid as a way to evaluate AI-generated output.

For opioid treatment, the paper discusses aspects of the evaluation process needed to build trust, and it identifies junction points in clinical workflows where LLM-based systems could potentially improve patient care.