Development and Validation of a Large Language Model-based Myopia Assistant System

Myopia is a rapidly growing global health concern, and there is an urgent need for advanced tools that can facilitate personalized healthcare strategies. Artificial intelligence (AI)-based solutions, such as large language models, offer robust tools for ophthalmic healthcare. In this study, investigators aim to validate a patient-centered Large Language Model (LLM)-based Myopia Assistant System with the following key objectives: 1) evaluate the ability of the LLM models to generate high-level reports and help self-evaluation of myopia for patients in primary care; 2) evaluate its performance in answering evidence-based medicine-oriented questions and improving overall satisfaction within clinics for myopic patients.

Study Overview

• Status: Completed
• Conditions: Myopia; Large Language Model
• Intervention / Treatment: Device: A patient-centered assistant system based on Large-Language Model (LLM)

Detailed Description

Myopia is a rapidly growing global health concern particularly affecting children and adolescents. The progression of myopia can lead to severe complications such as myopic macular degeneration, significantly impacting visual acuity and quality of life. With the rising prevalence of myopia, there is an urgent need for advanced tools that can facilitate personalized healthcare strategies. Artificial intelligence (AI)-based solutions, such as large language models, offer robust tools for ophthalmic healthcare. Nevertheless, their effectiveness and safety in real clinical environments have not been fully explored.

In this study, investigators aim to validate a patient-centered Large Language Model (LLM)-based Myopia Assistant System with the following key objectives: 1) evaluate the ability of the LLM models to generate high-level reports and help self-evaluation of myopia for patients in primary care; 2) evaluate its performance in answering evidence-based medicine-oriented questions and improving overall satisfaction within clinics for myopic patients. The findings of this study will provide valuable insights for the application of the GPT model in the healthcare field, making a significant contribution to improving the accessibility and quality of medical services.

• Study Type: Interventional
• Enrollment (Actual): 70
• Phase: Not Applicable

Contacts and Locations

Study Locations

• China
  • Hong Kong
    • Hong Kong, Hong Kong, China, 000
      • The Hong Kong Polytechnic University

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Child; Adult; Older Adult
• Accepts Healthy Volunteers: Yes

Description

Inclusion Criteria:

1. Outpatient participants aged 6 to 75.
2. Participants who undergo ophthalmic examinations for medical purposes.
3. Participants who can produce clear ophthalmic images in both eyes.
4. No prior experience in research involving digital medicine
5. Agree to participate in this study with written informed consent

Exclusion Criteria:

1. Participants who are reluctant to participate in this study
2. Participants who are unable to understand the study.
3. Participants who have recently undergone ocular surgery or those with severe ocular conditions that may affect the interpretation of imaging results related to myopia evaluation (e.g., severe vitreous hemorrhage, cataracts, corneal leukoma, etc.) will be excluded from the study.
4. Participants with poor quality of ophthalmic images, including blurriness, artifacts, underexposure, or overexposure.
5. Other unsuitable reasons determined by the evaluators.

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Other
• Allocation: Randomized
• Interventional Model: Parallel Assignment
• Masking: Single

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: Patient-centered assistant system; Participants engaged in the outpatient clinic visit procedure with a patient-centered assistant system based on Large-Language Model (LLM) for 10 minutes.	Device: A patient-centered assistant system based on Large-Language Model (LLM); Participants will engage in a 10-minute medical consultation using LLM model interface embedded in a tablet device before their regular face-to-face consulation with physicians. During the trials, participants could engage in free conversations covering aspects including risk factors, symptoms, diagnosis, examinations, treatment, advice and caution, etc. Participants who have completed the ophthalmic imaging examination will be asked to input results into the assistant model to generate structured reports.
No Intervention: Control group; Participants engaged in the outpatient clinic visit procedure without the support of patient-centered assistant system based on Large-Language Model (LLM) or any similar artificial intelligence assistance for 10 minutes.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Satisfaction level	Participants satisfaction level of the clinical experience with or without the use of a patient-centered assistant system based on a large language model (LLM) was assessed. The total satisfaction score was reported using the questionnaire (Patient User Satisfaction Scale), which evaluated the participant satisfaction with the clinical experience and the effectiveness of resolving their own issues. The questionnaire was measured on a 5-point Likert scale, where 1 represents strongly disagree; and 5 represents strongly agree; with higher scores indicating greater satisfaction.	Immediately after the outpatient clinic visit procedure

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Whether participants adopt the myopia management advice from the physician	It is a binary outcome that assesses whether participants follow the recommendations from the physician for myopia management. It focuses on whether participants implement the prescribed treatments or interventions provided to control or manage their myopia.	Immediately after the outpatient clinic visit procedure

Collaborators and Investigators

• Sponsor: The Hong Kong Polytechnic University
• Investigators: Principal Investigator: Mingguang He, M.D, Ph.D, The Hong Kong Polytechnic University

Study record dates

Study Major Dates

• Study Start (Actual): September 21, 2024
• Primary Completion (Actual): October 26, 2024
• Study Completion (Actual): October 26, 2024

Study Registration Dates

• First Submitted: September 11, 2024
• First Submitted That Met QC Criteria: September 19, 2024
• First Posted (Actual): September 23, 2024

Study Record Updates

• Last Update Posted (Actual): March 25, 2025
• Last Update Submitted That Met QC Criteria: March 11, 2025
• Last Verified: February 1, 2025

More Information

Terms related to this study

• Keywords: Myopia; Assistance; Large language model
• Additional Relevant MeSH Terms: Eye Diseases; Refractive Errors; Myopia
• Other Study ID Numbers: HSEARS20240229009

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: UNDECIDED

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No
• product manufactured in and exported from the U.S.: No