Breast cancer is a significant health issue affecting women worldwide, and accurately detecting lymph node metastasis is critical in determining treatment and prognosis. While traditional diagnostic methods have limitations and complications, artificial intelligence (AI) techniques such as machine learning (ML) and deep learning (DL) offer promising solutions for improving and supplementing diagnostic procedures. Current research has explored state-of-the-art DL models for breast cancer lymph node classification from radiological images, achieving high performances (AUC: 0.71–0.99). AI models trained on clinicopathological features also show promise in predicting metastasis status (AUC: 0.74–0.77), whereas multimodal (radiomics + clinicopathological features) models combine the best from both approaches and also achieve good results (AUC: 0.82–0.94). Once properly validated, such models could greatly improve cancer care, especially in areas with limited medical resources. This comprehensive review aims to compile knowledge about state-of-the-art AI models used for breast cancer lymph node metastasis detection, discusses proper validation techniques and potential pitfalls and limitations, and presents future directions and best practices to achieve high usability in real-world clinical settings.

Accurate, comprehensive, and consistent reporting of adverse events is of great importance for treatment decisions in clinical practice and patient safety. Aiming to evaluate the completeness and transparency of reported adverse events we conducted a retrospective analysis of completed clinical trials on glaucoma interventions registered in ClinicalTrials.gov from September 27, 2009, and updated and with results on or before November 1, 2023, as well as in corresponding journal publications. Any difference in completeness, number, or terminology/description of adverse events and all-cause mortality between ClinicalTrials.gov and the publication was categorized as inconsistent reporting of adverse events. All 79 trials with results both in the registry and a journal publication exhibited at least one inconsistency in reporting adverse events. In 19 publications (24%), the number of serious adverse events was smaller than in the registry. 69 (87%) trials reported more other adverse events in the registry than in the publication. Trials completed after the FDAA mandate for summary reporting of all-cause mortality more often reported this item in the registry but not in the publication. Trials on glaucoma interventions do not consistently report adverse events and thus introduce concerns about study credibility and potential harms of the interventions. Journals and other stakeholders in trial reporting must address this problem to ensure the safety of patients and trust in health interventions.

Background/Objectives: The World Health Organization’s Surgical Safety Checklist (WHO SSC) is known to reduce surgical complications and mortality, yet its implementation remains inconsistent across institutions. This study evaluated compliance with a modified SSC and assessed the impact of structured interventions to improve adherence at the University Hospital of Split, Croatia. Methods: A before-and-after study analyzed a sample of 1437 completed SSCs over four time points between April 2024 and May 2025: the baseline and after three successive interventions (an official letter from the hospital director emphasizing mandatory SSC use, individual meetings with department heads and head nurses reinforcing its importance, and a quality audit review of SSC completeness with leadership). Checklist completeness was assessed across five SSC sections: General Information, Sign In, Time Out, Sign Out, and Staff Info. Results: Overall checklist completeness increased from 78.3 ± 8.5% at baseline to 86.3 ± 2.5%, 92.0 ± 3.8%, and 94.7 ± 4.8% after the first, second, and third interventions, respectively (p < 0.001). All checklist sections improved significantly: General Info rose from 91.1 ± 6.0% to 98.6 ± 2.6% (p < 0.001); Sign In from 85.0 ± 11.4% to 97.0 ± 3.5% (p = 0.002); Time Out from 79.0 ± 14.6% to 96.4 ± 6.5% (p < 0.001); Sign Out from 70.2 ± 11.2% to 87.7 ± 11.0% (p = 0.003); and Staff Info from 70.7 ± 12.9% to 100.0 ± 0.0% (p < 0.001). Post hoc testing confirmed significant gains versus baseline for all three interventions (Dunnett’s test), with a further significant improvement between the first and third interventions (Tukey’s HSD, p = 0.011). Conclusions: Structured, low-cost interventions emphasizing administrative support, education, and communication significantly improved SSC adherence across a large tertiary hospital. This initiative represents a pioneering effort in Croatia and offers a scalable model for similar public healthcare systems aiming to enhance patient safety.

Background: Postoperative macular edema may limit visual recovery following cataract surgery. Although smoking is recognized as a risk factor for ocular inflammation, its impact on early postoperative macular morphology following cataract surgery has not been investigated. Methods: This prospective cohort study enrolled 88 elderly patients undergoing elective cataract surgery in a single university teaching hospital. The patients were divided into long-term smokers and lifelong non-smokers. Spectral-domain optical coherence tomography (OCT) was used to assess the central subfoveal thickness (CST), cube volume (CV), cube average thickness (CAT), retinal nerve fiber layer (RNFL), and cup-to-disk ratio (CDR) preoperatively and on the 1st, 7th, and 14th postoperative days (PODs). The phacoemulsification time and cumulative dissipated energy were recorded. Linear mixed-effects models were used to assess group-by-time interactions, and multivariable regression, adjusted for baseline covariates, was employed for analyses. Results: Eighty patients were included in the final analysis. Smokers had significantly thinner baseline CST than non-smokers. Both groups showed early postoperative CST increases, but only smokers exhibited sustained and significantly greater increases in CV and CAT on POD 14 (CV Δ +0.30 mm³ vs. +0.04 mm³; p = 0.026; CAT Δ +6.5 µm vs. +1.2 µm; p = 0.037). The RNFL and CDR changes did not differ significantly at earlier timepoints. However, smokers showed a notably greater RNFL thickening on POD 14 (Δ +4.2 µm; p = 0.001). Smoking status remained the strongest independent predictor of these changes (p < 0.001), while phacoemulsification parameters showed no significant interaction effects. Conclusions: Cigarette consumption independently predicts pronounced postoperative macular and RNFL thickening after uncomplicated elective cataract surgery. These transient structural changes could complicate early glaucoma assessment and should be considered when interpreting postoperative OCT findings in smokers.

Background/Objectives: Written discharge explanations are crucial for patient understanding and safety in emergency eye care, yet their tone and clarity vary. Large language models (LLMs, artificial intelligence systems trained to generate human-like text) can produce patient-friendly materials, but direct, blinded comparisons with clinician-written texts remain scarce. This study compared patient perceptions of a routine clinician-written discharge text and a GPT-5-generated explanation, where GPT-5 (OpenAI) is a state-of-the-art LLM, based on the same clinical facts in emergency ophthalmology. The primary objective was empathy; secondary outcomes included clarity, detail, usefulness, trust, satisfaction, and intention to follow advice. Methods: We conducted a prospective, single-blind, within-subject study in the Emergency Ophthalmology Unit of the University Hospital Centre Split, Croatia. Adults (n = 129) read two standardized texts (clinician-written vs. GPT-5-generated), presented in identical format and in randomized order under masking. Each participant rated both on Likert scales with 1–5 points. Paired comparisons used Wilcoxon signed-rank tests with effect sizes, and secondary outcomes were adjusted using the Benjamini–Hochberg false discovery rate. Results: Empathy ratings were lower for the GPT-5-generated text than for the clinician-written text (means 3.97 vs. 4.30; mean difference −0.33; 95% CI −0.44 to −0.22; p < 0.001). After correcting for multiple comparisons, no secondary outcome differed significantly between sources. Preferences were evenly split (47.8% preferred GPT-5 among those expressing a preference). Conclusions: In emergency ophthalmology, GPT-5-generated explanations approached clinician-written materials on most perceived attributes but were rated less empathic. A structured, human-in-the-loop workflow—in which LLM-generated drafts are reviewed and tailored by clinicians—appears prudent for safe, patient-centered deployment.

Background and Objectives: Vernix caseosa peritonitis (VCP) is rare. Nonspecific symptoms of acute abdomen during early puerperium make preoperative diagnosis of VCP challenging. We aimed to identify risk factors, early diagnosis and treatment options, and the association between the timing and severity of VCP diagnosis and maternal outcomes. Materials and Methods: We searched PubMed, PubMed Central, and Google Scholar. Articles were analyzed according to the PRISMA guidelines. The search items included: ‘vernix caseosa peritonitis, ‘vernix caseosa granuloma, ‘maternal meconium peritonitis’, ‘maternal meconium granuloma’, ‘vernix caseosa’, ‘peritonitis’, ‘pregnancy’, ‘puerperium’, ‘postpartum’, and ‘gravid’. Additional studies were extracted by reviewing the reference lists of retrieved studies. Demographic, clinical, obstetric, diagnostic, and treatment parameters, and outcomes were collected. Results: Out of 55 published VCP case reports, 46 were available. Most involved term pregnancies (84.8%) and were delivered by Cesarean section (CS) (87%), with no difference in parity distribution (χ²(2) = 1.1875, p = 0.5523) or fetal sex (m: f = 53.3%: 46.7%). Common symptoms included abdominal pain and fever over 38 °C, while dyspnea or tachypnea was unexpectedly frequent (23.9%/15.2%). The interval from delivery to surgery ranged from 4 to 13 days (average 8 days), with no difference between CS and vaginal deliveries. Preoperative VCP was diagnosed in only 4.3% of cases, and intraoperative diagnosis occurred in 60.9%. Intraoperative microbiology and histopathology (vernix components) were positive in 6.5% and 89.1%, respectively. The birth weight was normal (3656 ± 509 g), with no maternal or neonatal deaths. Conclusions: VCP primarily develops in term pregnancies delivered by CS, without other risk factors. Despite extremely low preoperative and unexpectedly low intraoperative diagnosis and treatment delay of several days, there is no maternal or fetal mortality. The time to symptom onset is similar between women who delivered vaginally and those who had a CS. All women with VCP after vaginal delivery had previous deliveries. Abdominal pain with a fever over 38 °C and dyspnea or tachypnea in the early puerperium suggests VCP. Surgical lavage is the primary treatment, while corticosteroids have been reported to be beneficial in several cases, and antibiotics seem to have a limited role.