Background Large language models (LLMs) are increasingly used in medical education and assessment frameworks. However, their reliability under varying task demands remains unclear. Significant gaps exist in the literature regarding the reasoning stability of these models when faced with fluctuations in question structure, difficulty, and linguistic framing. In specialized fields such as dermatology, where diagnostic precision is critical, addressing these inconsistencies is essential for the reliable integration of AI into educational training environments. Methods This study was designed as an educational benchmarking analysis under controlled assessment conditions. We evaluated 4 LLMs (ChatGPT 5.0, Claude 4.5 Sonnet, Gemini 2.5 Flash, DeepSeek V3) on 136 dermatology board-style multiple-choice questions, each adapted into 4 task methods: (1) original single-best-answer format; (2) correct option replaced with "None of the answers;" (3) four correct options, requiring identification of the single incorrect option; and (4) no correct option (four distractors). Items were classified as Easy, Medium, or Hard and as Positive or Negative stems. The primary outcome was Correct Response (1/0) across 2,176 model-item-method observations. A generalized linear mixed-effects model with random intercepts for Question ID estimated the effects of model, method, difficulty, and polarity, including interactions; Holm adjustment controlled for multiple comparisons. Results Overall accuracy was 52%. Using ChatGPT as the reference, all models showed significantly lower odds of correctness (Claude OR, 0.45; Gemini OR, 0.56; DeepSeek OR, 0.54; all p < 0.001). Task method had the largest effect, with marked reductions from Method 1 to Methods 2-4 (Method 4 OR, 0.026; p < 0.001). Negative stems lowered accuracy (OR, 0.57; p < 0.001). Method & times; Difficulty, Method & times; Polarity, and Difficulty & times; Polarity interactions were significant, as was the three-way interaction (p = 0.006), with the steepest losses for medium-difficulty negative items in Methods 2 and 4. Conclusion Overall, LLM performance exhibited sensitivity to task design, difficulty, and linguistic framing within this educational assessment framework. These findings suggest task fragility and warrant caution in unsupervised educational applications. Practically, this highlights the value of format-aware evaluation when considering LLMs for medical examination preparation or automated question generation, helping to align expectations with their observed educational capabilities.
Objectives Cefuroxime ocular administration is widely used to reduce the incidence of endophthalmitis after ocular surgery. However, due to ocular administration not being the recommended route in the prescribing information for cefuroxime, there is still a lack of systematic research on the adverse events of cefuroxime ocular administration. This study evaluated the adverse events following cefuroxime ocular administration using the FDA Adverse Event Reporting System database, with particular focus on the potential heterogeneity of reporting adverse events across different populations and dosing regimens.Methods The clinical characteristics of cefuroxime ocular administration-associated adverse event reports were analyzed and significant adverse event signals were identified using four disproportionality methods. Then, the potential demographic and dose-dependent differences in reporting adverse events were studied using the reporting odds ratio algorithm and Fisher's exact test.Results 88.7% of the adverse event cases resulted in severe outcomes. Sixteen signals for "eye disorders" and three signals for "injury, poisoning and procedural complications" were identified. There may be potential difference in reporting cefuroxime ocular administration-associated adverse events across continents, sexes and doses.Conclusion This study emphasizes the need for careful attention to eye disorders and stringent quality control measures for off-label use when cefuroxime was administered ocular. The findings underscore the need for future confirmatory studies into the potential demographic and dose differences in adverse events associated with cefuroxime ocular administration. The findings represent hypothesis-generating associations which require further validation.
Purpose Pediatric myopia is associated with scleral extracellular matrix remodeling. Doxycycline hyclate, a tetracycline-class matrix metalloproteinase inhibitor at sub-antimicrobial doses, is being investigated for long-term myopia treatment. This study evaluated the chronic ocular safety and systemic exposure of a doxycycline hyclate ophthalmic solution following repeated topical administration in a juvenile model.Methods A 9-month Good Laboratory Practice (GLP) repeat-dose study was conducted in juvenile Dutch Belted rabbits. Animals received bilateral topical ocular administration of doxycycline hyclate ophthalmic solution at 0.1, 1.0, or 5.0 mg/mL (0.005, 0.05, or 0.25 mg/eye), or vehicle control, twice daily for 270 days, followed by a 14-day recovery. Assessments included clinical signs, body weights, ophthalmic examinations, intraocular pressure (IOP), clinical pathology, toxicokinetic evaluation, and gross/microscopic pathology.Results No treatment-related mortality, adverse clinical findings, or ocular tissue alterations were observed. Ophthalmic examinations and clinical pathology showed no test article-related toxicity. Systemic exposure increased dose-dependently, with Tmax generally observed 1-12 hours post-dose on Days 1, 180, and 270. Exposure demonstrated low accumulation and was independent of sex. No test article-related macroscopic observations, organ weight changes, or microscopic pathological findings were identified.Conclusion Twice-daily topical doxycycline hyclate ophthalmic solution was well tolerated in juvenile rabbits for 9 months. No systemic or ocular toxicity was identified, establishing the No-Observed-Adverse-Effect Level (NOAEL) at 0.25 mg/eye (1 mg/day). These findings support the safety of long-term ocular exposure to doxycycline hyclate for chronic applications.
INTRODUCTION Proparacaine hydrochloride is a topical ocular anesthetic widely used in diagnostic procedures. Although generally safe when properly administered, repeated or unsupervised use has been associated with clinically relevant corneal injury. In contexts where access to topical anesthetics is insufficiently controlled, misuse may result in cumulative epithelial damage and vision-threatening complications. Despite clinical reports and experimental evidence of corneal toxicity, quantitative data linking exposure conditions to structural corneal damage under controlled ex vivo settings remain limited, particularly regarding dose-injury relationships at the tissue level. METHODS Porcine corneas were mounted in Franz diffusion cells to quantify permeation by UV-Vis spectrophotometry and evaluate corneal histopathological changes. Both pure proparacaine and the commercial 0.5% proparacaine hydrochloride ophthalmic solution (Alcaine®, Alcon) were assessed under therapeutic (2.25 mg/mL), abuse (4.5 mg/mL), and infinite-dose (53.152 mg/mL) conditions. Negative (phosphate-buffered saline) and positive (99.5% ethanol) controls were included for histopathological assessment. ANOVA was applied to permeation data, and histopathological scores were analyzed using non-parametric tests. RESULTS Proparacaine showed low permeability (9.33 × 10-7 cm/s), with only a small fraction of the applied dose permeating across the cornea during the experimental period. Permeation increased in a concentration- and time-dependent manner. High-dose and infinite-dose exposure conditions resulted in marked epithelial damage, including spongiosis, chromatin condensation, and desquamation, with severity comparable to that induced by ethanol used as a positive control. Differences between formulations were primarily driven by concentration rather than formulation composition, and a correlation was observed between cumulative permeation and the extent of epithelial injury. CONCLUSION Proparacaine demonstrates dose-dependent permeation and histopathological changes, emphasizing the risks associated with misuse. The ex vivo model provides relevant and ethically aligned evidence to support ocular hazard assessment through the characterizing of histopathological changes.
