Generative AI Patient Education Module for Breast Oncology

Accuracy and Effect of Prescribed Generative AI Patient Education Module Within Breast Oncology Care

This study evaluates the safety, accuracy, and impact of an artificial intelligence (AI) tool designed to support patient education in breast cancer care for breast oncology patients under selected physicians care within the University of California, San Francisco breast cancer center and affiliate sites.

Study Overview

• Status: Recruiting
• Conditions: Breast Cancer; Metastatic Breast Cancer; Carcinoma of the Breast
• Intervention / Treatment: Other: Survey (Participant); Other: OpenEvidence CareConnect; Other: Survey (Provider)

Detailed Description

PRIMARY OBJECTIVES:

I. Lead-in period: To evaluate the safety of the OpenEvidence CareConnect a Retrieval-Augmented Generation (RAG) Large Language Model (LLM) tool during the initial lead-in phase in patients with metastatic breast cancer.

II. To evaluate the clinical accuracy and appropriateness of the OpenEvidence CareConnect RAG LLM tool in delivering general health and breast cancer-specific information.

SECONDARY OBJECTIVES:

I. To assess patient perceptions of the AI tool's helpfulness, alignment with other information sources, and its impact on communication with their oncology care team.

II. To evaluate provider perceptions of the tool's usefulness and impact on communication and the quality of patient education.

III. To compare patient experiences and perceptions between immediate access and delayed access to the AI educational tool.

EXPLORATORY OBJECTIVES:

I. To evaluate the changes in inter-visit communication between patient and care team with and without access to the AI tool.

II. To understand patient concerns and educational needs related to breast cancer diagnosis and treatment.

OUTLINE: The first 5 participants will be assigned to the Lead-In Phase. After which, the next 30 participants will be randomized to either an Intervention arm or a Waitlist Control arm. Participants on the Waitlist Control arm will cross over after 3 months to the Intervention arm. Surveys and provider assessments occur throughout, and adherence and safety are monitored during the lead-in. Each participant will be on study for approximately 6 months.

• Study Type: Interventional
• Enrollment (Estimated): 35
• Phase: Not Applicable

Contacts and Locations

• Study Contact: Name: Ivy Ow; Phone Number: 877-827-3222; Email: Ivy.Wong@ucsf.edu

Study Locations

• United States
  • California
    • San Francisco, California, United States, 94143
      • Recruiting
      • University of California, San Francisco
      • Contact: Email: cancertrials@ucsf.edu
      • Contact: Ivy Ow; Phone Number: 877-827-3222; Email: Ivy.Wong@ucsf.edu
      • Principal Investigator: Michelle Melisko, MD

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

Description

Inclusion Criteria:

1. Females or males ages 18 and over.

2. Patients must have either:

   1. Stage IV breast cancer facing a treatment change and are expected to have follow-up visits at least once every three months, or
   2. Stage I-III breast cancer diagnosed within the past 6 months.
3. Basic computer literacy and regular internet access at home.
4. Able to understand study procedures and to comply with them for the entire length of the study.
5. Ability of individual or legal guardian/representative to understand a written informed consent document, and the willingness to sign it.

Exclusion Criteria:

1. Contraindication to any study-related procedure or assessment.
2. Cognitive impairment that would interfere with tool usage or survey completion

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Health Services Research
• Allocation: Randomized
• Interventional Model: Crossover Assignment
• Masking: None (Open Label)

Number of Arms

3

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: Safety Lead-in Phase; Participants will receive access to OpenEvidence CareConnect a Retrieval-Augmented Generation (RAG) Large Language Model (LLM) tool that provides general health and breast cancer-specific information for 3 months.	Other: Survey (Participant); Participants will complete surveys to evaluate their experience using OpenEvidence CareConnect RAG LLM tool.; Other: OpenEvidence CareConnect; Retrieval-Augmented Generation (RAG) Large Language Model (LLM) tool; Other: Survey (Provider); The provider will complete a survey to assess OpenEvidence CareConnect RAG LLM tool's usefulness and the quality of its patient education communication.
Experimental: Interventional (Immediate Access); Participants will be given immediate access to OpenEvidence CareConnect for up to 3 months and complete surveys regarding use of tool. Participants are then followed up for an additional 3 months.	Other: Survey (Participant); Participants will complete surveys to evaluate their experience using OpenEvidence CareConnect RAG LLM tool.; Other: OpenEvidence CareConnect; Retrieval-Augmented Generation (RAG) Large Language Model (LLM) tool; Other: Survey (Provider); The provider will complete a survey to assess OpenEvidence CareConnect RAG LLM tool's usefulness and the quality of its patient education communication.
Experimental: Control (Delayed Access); Participants will be given access to OpenEvidence CareConnect for 3 months 90 days after enrollment	Other: Survey (Participant); Participants will complete surveys to evaluate their experience using OpenEvidence CareConnect RAG LLM tool.; Other: OpenEvidence CareConnect; Retrieval-Augmented Generation (RAG) Large Language Model (LLM) tool; Other: Survey (Provider); The provider will complete a survey to assess OpenEvidence CareConnect RAG LLM tool's usefulness and the quality of its patient education communication.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Mean score of Safety and Risk Assessment (Lead-in and Provider)	Safety and Risk Assessment evaluates safety concerns based on provider responses using a 5-point Likert scale.	Up to 90 months
Mean score of Provider Assessment of AI patient Education Tool Survey (Provider)	AI patient Education Tool Survey evaluates clinical accuracy and appropriateness of the OpenEvidence CareConnect tool based on provider responses using a 5-point Likert scale ranging from 1 (Strongly Disagree) to 5 (Strongly Agree).	Up to 180 days
Mean score of Assessment of Answer Accuracy and Content Appropriateness (Provider)	Provider will evaluate participant questions, answer accuracy and content provided within the OpenEvidence using 3-point scales.	Uup to 180 days

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Mean score of AI Tool Experience survey (Participant)	The AI Tool Experience Survey is a 40-question instrument to assess participants' perceptions of usability, usefulness, clinical integration, and support for shared decision-making. Each item is rated on a 5-point Likert scale ranging from 1 (Strongly Disagree) to 5 (Strongly Agree).	Up to 180 days
Mean score of Medical Information Experience Survey (Participant)	Medical Information Experience Survey is a 13-question instrument to evaluate how participants access and engage with medical information. Questions are measured using a 5-point Likert scale.	up to 180 days
Mean Change of Score of Patient AI Tool Experience Survey (Participant)	The AI Tool Experience Survey is a 40-question instrument to assess participants' perceptions of usability, usefulness, clinical integration, and support for shared decision-making. Each item is rated on a 5-point Likert scale ranging from 1 (Strongly Disagree) to 5 (Strongly Agree). Comparisons of survey responses between the immediate and delayed access groups will be reported.	up to 180 days
Mean Change of Score of Medical Information Experience Survey (Participant)	Medical Information Experience Survey is a 13-question instrument evaluates how participants access and engage with medical information. Questions are measured using a 5-point Likert scale. Comparisons of survey responses between the immediate and delayed access groups will be reported.	Up to 180 days

Collaborators and Investigators

• Sponsor: University of California, San Francisco
• Investigators: Principal Investigator: Michelle Melisko, MD, University of California, San Francisco

Study record dates

Study Major Dates

• Study Start (Actual): July 14, 2026
• Primary Completion (Estimated): January 31, 2027
• Study Completion (Estimated): January 31, 2027

Study Registration Dates

• First Submitted: March 30, 2026
• First Submitted That Met QC Criteria: March 30, 2026
• First Posted (Actual): April 6, 2026

Study Record Updates

• Last Update Posted (Actual): August 6, 2026
• Last Update Submitted That Met QC Criteria: August 4, 2026
• Last Verified: August 1, 2026

More Information

Terms related to this study

• Keywords: Digital Health; Artificial Intelligence (AI)
• Additional Relevant MeSH Terms: Neoplasms by Site; Neoplasms; Skin Diseases; Breast Diseases; Skin and Connective Tissue Diseases; Breast Neoplasms; Health Care Quality, Access, and Evaluation; Investigative Techniques; Epidemiologic Methods; Data Collection; Health Care Evaluation Mechanisms; Quality of Health Care; Public Health; Environment and Public Health; Surveys and Questionnaires
• Other Study ID Numbers: 257517; NCI-2026-02050 (Registry Identifier: NCI Trial ID)

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No