1
|
Kalaw FGP, Baxter SL. Ethical considerations for large language models in ophthalmology. Curr Opin Ophthalmol 2024:00055735-990000000-00188. [PMID: 39259616 DOI: 10.1097/icu.0000000000001083] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 09/13/2024]
Abstract
PURPOSE OF REVIEW This review aims to summarize and discuss the ethical considerations regarding large language model (LLM) use in the field of ophthalmology. RECENT FINDINGS This review of 47 articles on LLM applications in ophthalmology highlights their diverse potential uses, including education, research, clinical decision support, and surgical assistance (as an aid in operative notes). We also review ethical considerations such as the inability of LLMs to interpret data accurately, the risk of promoting controversial or harmful recommendations, and breaches of data privacy. These concerns imply the need for cautious integration of artificial intelligence in healthcare, emphasizing human oversight, transparency, and accountability to mitigate risks and uphold ethical standards. SUMMARY The integration of LLMs in ophthalmology offers potential advantages such as aiding in clinical decision support and facilitating medical education through their ability to process queries and analyze ophthalmic imaging and clinical cases. However, their utilization also raises ethical concerns regarding data privacy, potential misinformation, and biases inherent in the datasets used. Awareness of these concerns should be addressed in order to optimize its utility in the healthcare setting. More importantly, promoting responsible and careful use by consumers should be practiced.
Collapse
Affiliation(s)
- Fritz Gerald P Kalaw
- Division of Ophthalmology Informatics and Data Science, Viterbi Family Department of Ophthalmology and Shiley Eye Institute
- Department of Biomedical Informatics, University of California San Diego Health System, University of California San Diego, La Jolla, California, USA
| | - Sally L Baxter
- Division of Ophthalmology Informatics and Data Science, Viterbi Family Department of Ophthalmology and Shiley Eye Institute
- Department of Biomedical Informatics, University of California San Diego Health System, University of California San Diego, La Jolla, California, USA
| |
Collapse
|
2
|
Stagg BC, Tullis B, Asare A, Stein JD, Medeiros FA, Weir C, Borbolla D, Hess R, Kawamoto K. Systematic User-centered Design of a Prototype Clinical Decision Support System for Glaucoma. OPHTHALMOLOGY SCIENCE 2023; 3:100279. [PMID: 36970116 PMCID: PMC10033738 DOI: 10.1016/j.xops.2023.100279] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Subscribe] [Scholar Register] [Received: 09/15/2022] [Revised: 12/05/2022] [Accepted: 01/27/2023] [Indexed: 02/05/2023]
Abstract
Purpose To rigorously develop a prototype clinical decision support (CDS) system to help clinicians determine the appropriate timing for follow-up visual field testing for patients with glaucoma and to identify themes regarding the context of use for glaucoma CDS systems, design requirements, and design solutions to meet these requirements. Design Semistructured qualitative interviews and iterative design cycles. Participants Clinicians who care for patients with glaucoma, purposefully sampled to ensure a representation of a range of clinical specialties (glaucoma specialist, general ophthalmologist, optometrist) and years in clinical practice. Methods Using the established User-Centered Design Process framework, we conducted semistructured interviews with 5 clinicians that addressed the context of use and design requirements for a glaucoma CDS system. We analyzed the interviews using inductive thematic analysis and grounded theory to generate themes regarding the context of use and design requirements. We created design solutions to address these requirements and used iterative design cycles with the clinicians to refine the CDS prototype. Main Outcome Measures Themes regarding decision support for determining the timing of visual field testing for patients with glaucoma, CDS design requirements, and CDS design features. Results We identified 9 themes that addressed the context of use for the CDS system, 9 design requirements for the prototype CDS system, and 9 design features intended to address these design requirements. Key design requirements included the preservation of clinician autonomy, incorporation of currently used heuristics, compilation of data, and increasing and communicating the level of certainty regarding the decision. After completing 3 iterative design cycles using this preliminary CDS system design solution, the design was satisfactory to the clinicians and was accepted as our prototype glaucoma CDS system. Conclusions We used a systematic design process based on the established User-Centered Design Process to rigorously develop a prototype glaucoma CDS system, which will be used as a starting point for a future, large-scale iterative refinement and implementation process. Clinicians who care for patients with glaucoma need CDS systems that preserve clinician autonomy, compile and present data, incorporate currently used heuristics, and increase and communicate the level of certainty regarding the decision. Financial Disclosures Proprietary or commercial disclosure may be found after the references.
Collapse
Affiliation(s)
- Brian C. Stagg
- Department of Ophthalmology and Visual Sciences, John Moran Eye Center, University of Utah, Salt Lake City, Utah
- Department of Population Health Sciences, University of Utah, Salt Lake City, Utah
| | - Benton Tullis
- School of Medicine, University of Utah, Salt Lake City, Utah
| | - Afua Asare
- Department of Ophthalmology and Visual Sciences, John Moran Eye Center, University of Utah, Salt Lake City, Utah
| | - Joshua D. Stein
- Department of Ophthalmology and Visual Sciences, Center for Eye Policy & Innovation, Kellogg Eye Center, University of Michigan, Ann Arbor, Michigan
- Institute for Healthcare Policy and Innovation, University of Michigan, Ann Arbor, Michigan
- Department of Health Management and Policy, University of Michigan School of Public Health, Ann Arbor, Michigan
| | | | - Charlene Weir
- Department of Biomedical Informatics, University of Utah, Salt Lake City, Utah
| | - Damian Borbolla
- Department of Biomedical Informatics, University of Utah, Salt Lake City, Utah
- Clinical Effectiveness, Wolters Kluwer Health, Salt Lake City, Utah
| | - Rachel Hess
- Department of Population Health Sciences, University of Utah, Salt Lake City, Utah
- Department of Internal Medicine, University of Utah, Salt Lake City, Utah
| | - Kensaku Kawamoto
- Department of Biomedical Informatics, University of Utah, Salt Lake City, Utah
| |
Collapse
|
3
|
Gu B, Sidhu S, Weinreb RN, Christopher M, Zangwill LM, Baxter SL. Review of Visualization Approaches in Deep Learning Models of Glaucoma. Asia Pac J Ophthalmol (Phila) 2023; 12:392-401. [PMID: 37523431 DOI: 10.1097/apo.0000000000000619] [Citation(s) in RCA: 6] [Impact Index Per Article: 6.0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 03/16/2023] [Accepted: 05/11/2023] [Indexed: 08/02/2023] Open
Abstract
Glaucoma is a major cause of irreversible blindness worldwide. As glaucoma often presents without symptoms, early detection and intervention are important in delaying progression. Deep learning (DL) has emerged as a rapidly advancing tool to help achieve these objectives. In this narrative review, data types and visualization approaches for presenting model predictions, including models based on tabular data, functional data, and/or structural data, are summarized, and the importance of data source diversity for improving the utility and generalizability of DL models is explored. Examples of innovative approaches to understanding predictions of artificial intelligence (AI) models and alignment with clinicians are provided. In addition, methods to enhance the interpretability of clinical features from tabular data used to train AI models are investigated. Examples of published DL models that include interfaces to facilitate end-user engagement and minimize cognitive and time burdens are highlighted. The stages of integrating AI models into existing clinical workflows are reviewed, and challenges are discussed. Reviewing these approaches may help inform the generation of user-friendly interfaces that are successfully integrated into clinical information systems. This review details key principles regarding visualization approaches in DL models of glaucoma. The articles reviewed here focused on usability, explainability, and promotion of clinician trust to encourage wider adoption for clinical use. These studies demonstrate important progress in addressing visualization and explainability issues required for successful real-world implementation of DL models in glaucoma.
Collapse
Affiliation(s)
- Byoungyoung Gu
- Division of Ophthalmology Informatics and Data Science and Hamilton Glaucoma Center, Viterbi Family Department of Ophthalmology and Shiley Eye Institute, University of California San Diego, La Jolla, CA, US
- Division of Biomedical Informatics, Department of Medicine, University of California San Diego, La Jolla, CA, US
| | - Sophia Sidhu
- Division of Ophthalmology Informatics and Data Science and Hamilton Glaucoma Center, Viterbi Family Department of Ophthalmology and Shiley Eye Institute, University of California San Diego, La Jolla, CA, US
- Division of Biomedical Informatics, Department of Medicine, University of California San Diego, La Jolla, CA, US
| | - Robert N Weinreb
- Division of Ophthalmology Informatics and Data Science and Hamilton Glaucoma Center, Viterbi Family Department of Ophthalmology and Shiley Eye Institute, University of California San Diego, La Jolla, CA, US
| | - Mark Christopher
- Division of Ophthalmology Informatics and Data Science and Hamilton Glaucoma Center, Viterbi Family Department of Ophthalmology and Shiley Eye Institute, University of California San Diego, La Jolla, CA, US
| | - Linda M Zangwill
- Division of Ophthalmology Informatics and Data Science and Hamilton Glaucoma Center, Viterbi Family Department of Ophthalmology and Shiley Eye Institute, University of California San Diego, La Jolla, CA, US
| | - Sally L Baxter
- Division of Ophthalmology Informatics and Data Science and Hamilton Glaucoma Center, Viterbi Family Department of Ophthalmology and Shiley Eye Institute, University of California San Diego, La Jolla, CA, US
- Division of Biomedical Informatics, Department of Medicine, University of California San Diego, La Jolla, CA, US
| |
Collapse
|
4
|
Guo X, Boland MV, Swenor BK, Goldstein JE. Low Vision Rehabilitation Service Utilization Before and After Implementation of a Clinical Decision Support System in Ophthalmology. JAMA Netw Open 2023; 6:e2254006. [PMID: 36735257 PMCID: PMC9898817 DOI: 10.1001/jamanetworkopen.2022.54006] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Figures] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Indexed: 02/04/2023] Open
Abstract
IMPORTANCE Electronic clinical decision support systems apply clinical guidelines in real time and offer a new approach to improve referral and utilization of low vision rehabilitation (LVR) care. OBJECTIVE To characterize patients and factors associated with LVR service utilization with and without the use of an electronic health record (EHR) clinical decision support system (CDSS) alert. DESIGN, SETTING, AND PARTICIPANTS Quality improvement study using EHR data to compare patients who did and did not utilize LVR service after referral between November 6, 2017, and October 5, 2019, (primary) and to assess overall service utilization rate from September 1, 2016, to April 2, 2021, regardless of referral status (secondary). Participants in the primary analysis were patients at a large ophthalmology department in an academic medical center in the US who received an LVR referral recommendation from their ophthalmologist according to the CDSS alert. The secondary analysis included patients with best documented visual acuity (BDVA) worse than 20/40 before, during, and after the CDSS implementation. Data were analyzed from August 2021 to April 2022. EXPOSURES Number and locations of referral recommendations for LVR service according to the CDSS alert in the primary analysis; active CDSS implementation in the secondary analysis. MAIN OUTCOMES AND MEASURES LVR service utilization rate was defined as the number of patients who accessed service among those who were referred (primary) and among those with BDVA worse than 20/40 (secondary). EHR data on patient demographics (age, sex, race, ethnicity) and ophthalmology encounter characteristics (numbers of referral recommendations, encounter location, and BDVA) were extracted. RESULTS Of the 429 patients (median [IQR] age, 71 [53 to 83] years; 233 female [54%]) who received a CDSS-based referral recommendation, 184 (42.9%) utilized LVR service. Compared with nonusers of LVR, users were more likely to have received at least 2 referral recommendations (12.5% vs 6.1%; χ21 = 5.29; P = .02) and at an ophthalmology location with onsite LVR service (87.5% vs 78.0%; χ21 = 6.50; P = .01). Onsite LVR service (odds ratio, 2.06; 95% CI, 1.18-3.61) persisted as the only statistically significant factor after adjusting for patient demographics and other referral characteristics. Among patients whose BDVA was worse than 20/40 before, during, and after the CDSS implementation regardless of referral status, the LVR service utilization rate was 6.1%, 13.8%, and 7.5%, respectively. CONCLUSIONS AND RELEVANCE In this quality improvement study, ophthalmologist referral recommendations and onsite LVR services at the location where patients receive other ophthalmic care were significantly associated with service utilization. Ophthalmology CDSSs are promising tools to apply clinical guidelines in real time to improve connection to care.
Collapse
Affiliation(s)
- Xinxing Guo
- Wilmer Eye Institute, Johns Hopkins University School of Medicine, Baltimore, Maryland
| | - Michael V. Boland
- Department of Ophthalmology, Massachusetts Eye and Ear and Harvard Medical School, Boston
| | - Bonnie K. Swenor
- Wilmer Eye Institute, Johns Hopkins University School of Medicine, Baltimore, Maryland
- Cochlear Center for Hearing and Public Health, Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland
- Disability Health Research Center, Johns Hopkins University, Baltimore, Maryland
- Johns Hopkins University School of Nursing, Baltimore, Maryland
| | - Judith E. Goldstein
- Wilmer Eye Institute, Johns Hopkins University School of Medicine, Baltimore, Maryland
| |
Collapse
|
5
|
Stagg BC, Stein JD, Medeiros FA, Horns J, Hartnett ME, Kawamoto K, Hess R. The Frequency of Visual Field Testing in a US Nationwide Cohort of Individuals with Open-Angle Glaucoma. Ophthalmol Glaucoma 2022; 5:587-593. [PMID: 35605937 PMCID: PMC9675879 DOI: 10.1016/j.ogla.2022.05.002] [Citation(s) in RCA: 12] [Impact Index Per Article: 6.0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 03/21/2022] [Revised: 05/04/2022] [Accepted: 05/13/2022] [Indexed: 05/16/2023]
Abstract
PURPOSE Visual field testing that is not frequent enough results in delayed identification of open-angle glaucoma (OAG) progression. Guidelines recommend at least annual testing. It is not known how frequently patients with OAG across the United States receive visual field testing and how patient characteristics and circumstances influence this frequency. If US patients with OAG do not receive visual field tests frequently enough, interventions to increase this frequency or to develop other forms of testing visual function may reduce unidentified OAG vision loss. DESIGN Retrospective cohort study. PARTICIPANTS The TruvenHealth MarketScan Commercial Claims Database (IBM) contains demographic and claims data for > 160 million individuals across the United States from 2008 to 2017. We identified enrollees in the database with a recorded diagnosis of OAG (International Classification of Diseases, Ninth Revision, Clinical Modification and International Classification of Diseases, Tenth Revision, Clinical Modification codes 356.1x and H40.1x, respectively). We excluded those aged < 40 years at the time of their first OAG diagnosis, those without at least 1 confirmatory OAG diagnosis at a subsequent visit, and those with < 4 years of follow-up data after OAG diagnosis. METHODS We calculated the number of visual field tests that each enrollee with OAG underwent per year and categorized the enrollees based on that number (0, > 0 to < 0.9, ≥ 0.9 to ≤ 1.1, > 1.1 to ≤ 2.1, and > 2.1). We used negative binomial regression to investigate the demographic or health variables that were associated with the frequency of visual field tests that enrollees with OAG received. MAIN OUTCOME MEASURES Frequency of visual field testing among enrollees with OAG. RESULTS Of the 380 029 enrollees included in the study, 33 267 (8.8%) did not receive a visual field test during the study period, 259 349 (68.2%) underwent > 0 to < 0.9 visual field tests per year, 42 129 (11.1%) underwent ≥ 0.9 to ≤ 1.1 visual field tests per year, 42 301 (11.1%) underwent > 1.1 to ≤ 2.1 visual field tests per year, and 2983 (0.8%) underwent ≥ 2.1 visual field tests per year. The median number of visual field tests per year was 0.63 (interquartile range, 0.33-0.88; mean, 0.65). CONCLUSIONS More than 75% of enrollees with OAG received < 1 visual field test per year and, thus, did not receive guideline-adherent glaucoma monitoring.
Collapse
Affiliation(s)
- Brian C Stagg
- John Moran Eye Center, Department of Ophthalmology and Visual Sciences, University of Utah, Salt Lake City, Utah; Department of Population Health Sciences, University of Utah, Salt Lake City, Utah.
| | - Joshua D Stein
- Center for Eye Policy & Innovation, Kellogg Eye Center, Department of Ophthalmology and Visual Sciences, University of Michigan, Ann Arbor, Michigan; Institute for Healthcare Policy and Innovation, University of Michigan, Ann Arbor, Michigan; Department of Health Management and Policy, University of Michigan School of Public Health, Ann Arbor, Michigan
| | | | - Joshua Horns
- Department of Surgery, Surgical Population Analysis Research Core, University of Utah Health Science Center, Salt Lake City, Utah
| | - M Elizabeth Hartnett
- John Moran Eye Center, Department of Ophthalmology and Visual Sciences, University of Utah, Salt Lake City, Utah
| | - Kensaku Kawamoto
- Department of Biomedical Informatics, University of Utah, Salt Lake City, Utah
| | - Rachel Hess
- Department of Population Health Sciences, University of Utah, Salt Lake City, Utah; Department of Internal Medicine, University of Utah, Salt Lake City, Utah
| |
Collapse
|