VOLUME 89 - ISSUE 2

PURPOSE: Natural language models and chatbots, particularly OpenAI’s Generative Pre-Trained Transformer architecture, have transformed human interaction with digital interfaces. The latest versions, including ChatGPT-4o, offer enhanced functionalities compared to their predecessors. This study evaluates the accuracy of ChatGPT-4, ChatGPT-4o, and Claude 3.5 Sonnet in answering questions from the Brazilian Retina and Vitreous Society certification exam.
METHODS: We compiled 200 multiple-choice questions from the Brazilian Retina and Vitreous Society 2018 and 2019 exams. Questions were categorized into three domains: Anatomy and Physiology of the Retina, Retinal Pathology, and Diagnosis and Treatment. Using a standardized prompt developed according to prompt design guidelines, we tested ChatGPT-4, ChatGPT-4o, and Claude 3.5 Sonnet, recording their first responses as final. Three retina specialists performed a qualitative analysis of the answers. Accuracy was determined by comparing responses to the official correct answers. Statistical analysis was conducted using chi-square tests and Cohen’s Kappa.
RESULTS: Claude 3.5 Sonnet achieved the highest overall accuracy (72.5%), followed by ChatGPT-4o (66.0%) and ChatGPT-4 (55.5%). Claude 3.5 Sonnet and ChatGPT-4o significantly outperformed ChatGPT-4 (p<0.01 and p=0.03, respectively), while no significant difference was observed between Claude 3.5 Sonnet and ChatGPT-4o (p=0.16). Model responses agreed 74.5% of the time, with a Cohen’s κ of 0.47. Retinal Pathology was the best-performing domain for all models, whereas Anatomy and Physiology of the Retina and Diagnosis and Treatment were the weakest domains for Claude 3.5 Sonnet and ChatGPT-4, respectively.
CONCLUSIONS: This study is the first to assess Claude 3.5 Sonnet, ChatGPT-4, and ChatGPT-4o in retina specialist certification exams. Claude 3.5 Sonnet and ChatGPT-4o significantly outperformed ChatGPT-4, highlighting their potential as effective tools for studying retina specialist board exams. These findings suggest that the enhanced functionalities of Claude 3.5 Sonnet and ChatGPT-4o offer substantial improvements in medical education contexts.

Keywords: Artificial intelligence; ChatGPT; Retina; Medical education; Ophthalmology, Large language model; Natural language processing

PURPOSE: To assess the one-year impact of prostaglandin analog (travoprost) on periorbital appearance and lower eyelid horizontal tension, compared with the contralateral eye treated with Pascal selective laser trabeculoplasty.
METHODS: This nested case-control study was derived from a non-inferiority randomized clinical trial comparing Pascal selective laser trabeculoplasty efficacy with travoprost in fellow eyes over 12 months. One eye received daily travoprost, while the contralateral eye underwent Pascal selective laser trabeculoplasty. Lower eyelid horizontal tension was measured using a validated digital imaging method, and prostaglandin-associated periorbitopathy signs were graded by masked observers. Statistical analyses included generalized estimating equations and the Mann–Whitney U test for nonparametric data, with p<0.05 considered significant.
RESULTS: Ten patients met the inclusion criteria for this subanalysis. Travoprost-treated eyes had significantly lower mean lower eyelid distension compared with Pascal selective laser trabeculoplasty-treated eyes (4.32 mm vs. 5.02 mm; mean difference: 0.70 mm; p<0.001). Prostaglandin-associated periorbitopathy scores were significantly higher in the travoprost group, reflecting more pronounced changes in ptosis, periorbital hyperpigmentation, and eyelid retraction.
CONCLUSIONS: Chronic use of prostaglandin analogs is associated with notable periorbital changes and increased lower eyelid tension, potentially affecting aesthetics and ocular function. Pascal selective laser trabeculoplasty may offer a safer profile for preserving periorbital anatomy while maintaining effective intraocular pressure control. Laser trabeculoplasty should be considered as an initial treatment when appropriate to minimize cosmetic and functional changes from chronic topical therapy.

Keywords: Prostaglandin analogs; Selective laser trabeculoplasty; Periorbitopathy; Eyelid tension; Glaucoma treatment

PURPOSE: This study aimed to identify the strategies adopted by Brazilian ophthalmologists to control myopia in clinical practice.
METHODS: This was a prospective cross-sectional study. Data were collected using an online questionnaire.
RESULTS: Responses from 148 participants were collected between March and May 2024. The majority of respondents were general ophthalmologists (51%) and pediatric ophthalmologists (43%). They came from all regions of Brazil, but more than half (52%) were from the Southeast region. Most participants (30%) had over 20 years of clinical practice experience. A significant proportion (89.2%) treated progressive myopia. The most requested complementary exams were optical biometry (83.78%) and corneal topography or tomography (69.59%). Behavioral measures were considered the most effective myopia treatment strategies by 41.2% of the respondents, followed by optical (33.8%) and pharmacological interventions (25%). Most recommended spending more time outdoors (94.59%) and reducing screen time (93.92%). Spectacle lenses for myopia (83.11%) and 0.025% atropine eye drops (54.73%) were the most prescribed treatments after the recommendation of environmental and behavioral changes.
CONCLUSION: This study presents a novel analysis of the clinical strategies for myopia control among Brazilian ophthalmologists. Understanding current clinical practices and identifying possible improvements are essential steps toward developing evidence-based guidelines and professional education aimed at improving patient care.

Keywords: Myopia/epidemiology; Refractive errors; Contact lenses; Myopia/drug therapy; Atropine/therapeutic use; Ophthalmologists; Practice patterns, physicians’; Surveys and questionnaires; Brazil/epidemiology

In Part II, this paper addresses ocular motions, their causes (forces), and the governing laws, beginning with the fundamental question: Why do the eyes move? Ocular rotations and different types of translations (ocular, orbital, and corporeal) are reviewed. The discussion then turns to how the eyes move, where concepts such as the plane of muscular action, torque, and arc of contact provide possible explanations for the anatomical arrangement of the ocular muscles within the orbit. Sherrington’s law of reciprocal innervation is used to explain the distribution of muscular active forces in a conservative mechanical system, but in combination with Hering’s law, it may prevent eye rotation (e.g., isometric contractions of antagonist muscles of an asymmetrical convergence). Normally, however, the limitation of an eye rotation is determined by passive forces, evoked by muscular activity itself, particularly natural muscular elasticity. Thus, elongation of an antagonist muscle may passively restrict the active contraction of an agonist. In addition to mechanisms for initiating rotation (active forces) and stopping it (passive forces), the oculomotor system also requires a means of dissipating energy (dissipative forces) to initiate subsequent movements. Hence, it cannot function as a perfectly conservative system of forces. The paper concludes with a review of “selective” effects of muscle function (due to the sparse distribution of fibers), the role of intermuscular membranes (and pulleys), and mechanical considerations of surgical procedures, such as muscular transpositions to alter or abolish actions (e.g., bifid reinsertions). Part III will address the diagnostic complexities of the oculomotor system, general treatment principles, and ocular fixation (eye and head positions). Although the basic concept of the primary position is relativized, the absolute need for referential conditions in defining, qualifying, and measuring strabismus is emphasized. The prim-diopter is challenged due to its lack of “linearity” relative to angular units, and an alternative is proposed. Methods of examining oculomotor disturbances are outlined, including monocular rotations (ductions), and tests to differentiate between muscular deficiencies and opposing forces. Techniques for identifying the site of a rotational restriction are described, followed by approaches to measuring ocular deviations in diagnostic positions. The concepts of muscular overactions and underactions are analyzed before introducing the concept of diagnostic muscle pairs. Classical knowledge about deviations caused by deficient or restricted muscle actions reinforces the theory of distribution of rotational ocular muscles by diagnostic pairs. For vertical deviations, “underactive” muscle pairs must be separately matched (e.g., RSR with LIR, RIO with LSO). Since vertical recti exert stronger vertical actions than oblique muscles, head tilts are recommended to enhance stress on both pairs, mainly by additional stimulation of oblique muscles. Classical diagnostic directions then align with the objective horizontal plane. The article concludes with peropertative oculomotor testing and a broad protocol for evaluating oculomotor imbalance.