Uveal melanoma distant metastasis prediction system: A retrospective observational study based on machine learning.

Uveal melanoma (UM) patients face a significant risk of distant metastasis, closely tied to a poor prognosis. Despite this, there is a dearth of research utilizing big data to predict UM distant metastasis. This study leveraged machine learning methods on the Surveillance, Epidemiology, and End Results (SEER) database to forecast the risk probability of distant metastasis. Therefore, the information on UM patients from the SEER database (2000-2020) was split into a 7:3 ratio training set and an internal test set based on distant metastasis presence. Univariate and multivariate logistic regression analyses assessed distant metastasis risk factors. Six machine learning methods constructed a predictive model post-feature variable selection. The model evaluation identified the multilayer perceptron (MLP) as optimal. Shapley additive explanations (SHAP) interpreted the chosen model. A web-based calculator personalized risk probabilities for UM patients. The results show that nine feature variables contributed to the machine learning model. The MLP model demonstrated superior predictive accuracy (Precision = 0.788; ROC AUC = 0.876; PR AUC = 0.788). Grade recode, age, primary site, time from diagnosis to treatment initiation, and total number of malignant tumors were identified as distant metastasis risk factors. Diagnostic method, laterality, rural-urban continuum code, and radiation recode emerged as protective factors. The developed web calculator utilizes the MLP model for personalized risk assessments. In conclusion, the MLP machine learning model emerges as the optimal tool for predicting distant metastasis in UM patients. This model facilitates personalized risk assessments, empowering early and tailored treatment strategies.

Real-world large sample assessment of drug-related dry eye risk: Based on the FDA adverse event reporting system database.

PURPOSE AND DESIGN: This study aimed to evaluate the risk of drug-related dry eye using real-world data, underscoring the significance of tracing pharmacological etiology for distinct clinical types of dry eye. METHODS: Analyzing adverse event reports in the Food and Drug Administration Adverse Event Reporting System (FAERS) from January 2004 to September 2023, we employed disproportionality analysis and the Bayesian confidence propagation neural network algorithm. The analysis involved categorizing drugs causing dry eye, assessing risk levels, and conducting segmental assessments based on the time of onset of drug-related dry eye adverse reactions. RESULTS: In the FAERS database, adverse reactions related to dry eye were linked to 1160 drugs. Disproportionality analysis identified 33 drugs with significant risk, notably in ophthalmic (brimonidine, bimatoprost), oncology (tisotumab vedotin, erdafitinib), and other medications (isotretinoin, oxymetazoline). The top three drugs with the highest risk of drug-related dry eye are isotretinoin (Bayesian confidence propagation neural network (BCPNN) = 6.88), tisotumab vedotin (BCPNN = 6.88), and brimonidine (BCPNN = 6.77). Among different categories of drugs, respiratory medications have the shortest mean onset time for drug-related dry eye, averaging 50.99 days. The prevalence skewed towards females (69.9 %), particularly in menopausal and elderly individuals (45-70 years old, mean age 54.7 ± 18.2). Reports of drug-related dry eye adverse reactions showed an annual increase. CONCLUSION: Informed clinical decision-making is crucial for preventing drug-related dry eye. Assessing the risk of dry eyes associated with both local and systemic medications helps optimize treatment and provide necessary cautionary information.

Drug-Related Keratitis: A Real-World FDA Adverse Event Reporting System Database Study.

Purpose: This study aimed to assess the drug risk of drug-related keratitis and track the epidemiological characteristics of drug-related keratitis. Methods: This study analyzed data from the U.S. Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) database from January 2004 to December 2023. A disproportionality analysis was conducted to assess drug-related keratitis with positive signals, and drugs were classified and assessed with regard to their drug-induced timing and risk of drug-related keratitis. Results: A total of 1606 drugs were reported to pose a risk of drug-related keratitis in the FAERS database, and, after disproportionality analysis and screening, 17 drugs were found to significantly increase the risk of drug-related keratitis. Among them, seven were ophthalmic medications, including dorzolamide (reporting odds ratio [ROR] = 3695.82), travoprost (ROR = 2287.27), and brimonidine (ROR = 2118.52), and 10 were non-ophthalmic medications, including tralokinumab (ROR = 2609.12), trazodone (ROR = 2377.07), and belantamab mafodotin (ROR = 680.28). The top three drugs having the highest risk of drug-related keratitis were dorzolamide (Bayesian confidence propagation neural network [BCPNN] = 11.71), trazodone (BCPNN = 11.11), and tralokinumab (BCPNN = 11.08). The drug-induced times for non-ophthalmic medications were significantly shorter than those for ophthalmic medications (mean days, 141.02 vs. 321.96, respectively; P < 0.001). The incidence of drug-related keratitis reached its peak in 2023. Conclusions: Prevention of drug-related keratitis is more important than treatment. Identifying the specific risks and timing of drug-induced keratitis can support the development of preventive measures. Translational Relevance: Identifying the specific drugs related to medication-related keratitis is of significant importance for drug vigilance in the occurrence of drug-related keratitis.

Transgenic dry eye mouse models: powerful tools to study dry eye disease.

Dry eye disease (DED) is one of the most common chronic multifactorial ocular surface diseases with high prevalence and complex pathogenesis. DED results in several ocular discomforts, vision fluctuation, and even potential damage of the ocular surface, bringing heavy burdens both on individuals and the society. The pathology of DED consists of tear film hyperosmolarity and immune responses on the ocular surface. Mice are widely used for developing models that simulate human DED features for investigating its pathogenesis and treatment. DED can be classified into aqueous-deficiency dry eye (ADDE) and evaporative dry eye (EDE). ADDE can be further divided into Sjögren syndrome dry eye (SSDE) and non-Sjögren syndrome dry eye (NSSDE). SSDE mouse models include natural strains, typified by non-obese diabetic (NOD) mice, and genetically engineered ones, like Aire-/- and Id3 knockout mice. Intrinsic EDE mainly refers to meibomian gland dysfunction (MGD). Eda-/- Tabby, Sod1-/-, Elovl1-/- are the most common transgenic MGD mouse models. Transgenic mouse models provide useful tools for studying the pathogenesis of DED and evaluating its novel therapies. This review compares the major transgenic dry eye mouse models and discusses their applications in DED research.